Speed control apparatus and method for rapid transit vehicles

BACKGROUND OF THE INVENTION
Rapid transit vehicles employing cam type propulsion control units present several problems for the automatic speed control of those transit vehicles. The discrete states or levels of tractive effort inherent in cam propulsion systems and the inability to command certain state to state transistions are somewhat incompatible with microprocessor based continuous type speed control systems such as described in U.S. Pat. No. 4,282,466 of T. C. Matty and entitled Transit Vehicle Motor Effort Control Apparatus and Method in relation to a chopper propulsion motor control system. Furthermore the cam state to state transition delay times and the increased equipment wear resulting from decreases in both positive and negative tractive effort present difficulties for the speed control apparatus and method.
It is desired for the Washington Metropolitan Area Transit Authority (WMATA) to provide an improved speed control apparatus and method responding to two primary control criteria, namely the vehicle speed has to be maintained within a predetermined speed band, such as a plus zero to minus four mile per hour speed band, relative to the commanded vehicle speed, and the number of adverse state changes has to be kept below six changes per minute. An adverse state change is defined as a decrease in either positive or negative tractive effort. Decreases in tractive effort are obtained in cam propulsion apparatus by opening resistor shorting switches. Since this action severely stresses these switches, the adverse change restriction is intended to reduce wear on the propulsion equipment and therefore extend equipment life. The previous automatic train speed control equipment controlling the cam propulsion units for the WMATA transit vehicles requests more than six adverse state changes per minute.
The previous WMATA train control apparatus includes twelve train lines over which the propulsion equipment and the ATC equipment communicate. There are ten specific propulsion train line patterns and five specific brake train line patterns in addition to a coast train line pattern which can be selected by the ATC equipment. These sixteen train line patterns, corresponding to sixteen states of the propulsion system operation, provide sixteen different levels of positive and negative tractive effort. WMATA, however, only uses fourteen of these states and only fourteen will be considered hereafter, since two propulsion states are not used. The adverse change limitation on train state transitions applies to changes in the train line pattern outputted by the ATC equipment.
The problem of speed maintaining within a four mile per hour band is not particularly difficult. A prior art proportional or proportional plus integral controller will successfully maintain speed with some modification to account for closed loop stability and certain unavailable state transitions while attempting to decrease tractive effort while in propulsion. This approach however results in more adverse state changes than desired.
A speed control approach simulating an ideal human operator is appropriate. An ideal human operator observes the train speed relative to the speed control band, and approximates the train acceleration in relation to forthcoming track conditions. He understands the undesirable effects of adverse state changes and therefore acts to minimize them. If he is speed maintaining low in the speed band and accelerating slightly he would take no action at that time. If on the other hand he were decelerating while low in the speed maintaining band he would transfer to a state providing more tractive effort. In addition he would not overreact, but would wait out the train delays to see the effect of the state transitions. Furthermore if a decrease in tractive or retarding effort were required he might transfer to a state sufficiently removed from the present state to assure that sufficient reduction occurs in an effort to reduce the number of adverse state transitions. A return to the coast state in this example would reverse the acceleration or deceleration under most conditions and result in the fewest adverse transitions. In this manner by monitoring speed, acceleration and the speed band while being aware of train delays, adverse transitions, and track conditions, an ideal human properly controls train performance. In addition there is a learning process that a human operator undergoes to improve his decision making ability with experience.
SUMMARY OF THE INVENTION
A speed control apparatus and method are provided for a transit vehicle, wherein a vehicle time to limit, defined as the time for speed to change from a present speed to a speed band limit, is provided as the difference between a predetermined speed band limit and the vehicle speed divided by the vehicle acceleration. This time to limit is compared with a determined vehicle response delay time for controlling when a transition from the present tractive effort state to a new tractive effort state takes place. The implementation of this speed control apparatus and method is through matrix mathematics.

BRIEF DESCRIPTION OF THE DRAWINGS
FIG. 1 shows a prior art track signal block arrangement for controlling the operation of a transit vehicle;
FIG. 2 shows a prior art vehicle carried speed control apparatus for determining the movement speed of a transit vehicle along a roadway track;
FIG. 3 shows a block diagram of the prior art cam propulsion control apparatus that has been operational for several years with the WMATA transit system in Washington DC;
FIG. 4 shows a block diagram of the improved cam propulsion control system of the present invention;
FIG. 5 shows the motor curves for a well known traction motor operating at the system line voltage of 650 volts;
FIG. 6 shows the motor curves for a well known traction motor operating at the lower range line voltage of 455 volts;
FIG. 7 shows the motor curves for a well known traction motor operating at the higher range line voltage of 845 volts;
FIGS. 8A and 8B show the speed regulation routine of the present invention;
FIG. 9 shows the matrix transfer subroutine of the present invention;
FIG. 10 shows the speed maintaining subroutine of the present invention;
FIGS. 11A and 11B show the speed band subroutine of the present invention;
FIG. 12 shows the time to limit subroutine of the present invention;
FIG. 13 shows the change states subroutine of the present invention;
FIG. 14 shows the train line output subroutine of the present invention;
FIGS. 15A and 15B show the program stop subroutine of the present invention;
FIG. 16 shows the look ahead subroutine that is operative when in program stop and the commanded speed limit is below the program stop profile speed;
FIG. 17 shows the seek subroutine operative to make a program stop table search;
FIG. 18 shows the flare-out subroutine of the present invention;
FIG. 19 shows the direction of move macro routine of the present invention;
FIG. 20 shows the propulsion, brake and coast transition states of the present invention;
FIG. 21 shows a typical speed band in relation to the command speed limit;
FIG. 22 shows the program stop speed limit crossing the command speed limit in relation to a program stop at a passenger station; and
FIGS. 23A, 23B and 23C show a functional block diagram to illustrate the operation of the present speed control apparatus and method for rapid transit vehicles;
FIG. 24 shows the grade determination routine; and
FIG. 25 shows the matrix search routine that determines the number of entries in the grade correction factor table.

DESCRIPTION OF THE PREFERRED EMBODIMENT
In FIG. 1 there is shown a prior art transit vehicle control signal block arrangement whereby respective desired command speed signals are transmitted to each of successive track signal blocks 100, 102, 104, and so forth. A transit vehicle 106 is shown within the signal block 100 as it moves along the track rails 108 and 110. The vehicle 106 will electrically short circuit and prevent the command speed signal from transmitter 112 from reaching the associated signal receiver 114 for the purpose of detecting that the vehicle 106 occupies the signal block 100. The transmitter 113 is similarly operative with previous signal block 101, the transmitter 115 and receiver 116 are similarly operative with signal block 102, and the receiver 117 is similarly operative with signal block 104.
In FIG. 2 there is shown a prior art speed control apparatus 201 carried by the transit vehicle 106 of FIG. 1. The speed control apparatus 201 determines the movement speed of the transit vehicle 106 along the track rails 108 and 110. Each signal block has an input command speed signal, and the track signal antennas 116 carried by the vehicle 106 as shown in FIG. 1 provides this input command speed signal to a receiver and decoder 200, which provides the input command speed at output 202. The speed regulation apparatus 204 compares the input command speed from output 202 with the vehicle actual speed from tachometer 206 for providing a speed error to the P signal generator 208, which then provides a tractive effort request P signal to the propulsion and brake system 210 for controlling the propulsion and the brake operation of the vehicle. A safety overspeed control 212 responds to the input command speed at output 202 and compares it in a failsafe manner with the vehicle actual speed from the tachometer 206 for providing an enable signal 214, which enables the P signal generator 208 to respond to the speed error from the speed regulation apparatus 204. A program stop control 216 responds to program stop control signals, provided from suitable wayside marker devices or a transposed cable positioned along the roadway track adjacent to a signal station, and sensed by a vehicle carried antenna 218 for determining the operation of the P signal generator 208 to control the program stop speed of the transit vehicle and thereby the stopping position of the vehicle in relation to the passenger station.
In FIG. 3 there is shown the prior art speed control apparatus presently in operation at the Washington Metropolitan Area Transit Authority (WMATA) transit system in Washington DC, and which includes a speed maintaining and program station stop speed control apparatus 300 operative with the transit cars provided on that system. The A and B cars include respective operator's console displays 301 and 302 operative with a console interface 305, and which indicate the train number, the train destination, train length, the regulated speed, the limiting speed, the actual speed and other data. The display in the lead car is driven by the ATO or automatic train operation module 304. The tachometer 306 provides an actual vehicle speed input signal 308 through the tachometer signal processing apparatus 307. The automatic train protection or ATP equipment 310 provides an input command speed signal 311 from the track signal block occupied by the train. The train grade from a wayside grade marker and a programmed stop speed limit from a wayside program stop marker when the train is approaching a passenger station are supplied to the marker decoding apparatus 312, which operates to decode and provide the grade signal 313 and the program stop speed limit signal 314. Respective operator's consoles 301 and 315, and 302 and 316 are provided in the A and B transit cars, with the operator being located in the lead car of the train. The ATS or automatic train supervision subsystem 318 provides a performance speed limit 319 that can be reduced below the ATC command speed to effect a performance modification because of bad weather or track conditions that require restricted operation of the train, provides a train I.D. signal 320, a one-half power acceleration limit 321 and a train length signal 322. The train line pattern apparatus 324 provides binary voltages on respective train lines for providing a desired one of the available states of power or brake operation, with there being available eight different power states, five different brake states and a coast state for this purpose. There are two WMATA propulsion states, switch 1 and switch 2, shown as states 1/3 and 2/3 of FIG. 20, that are also available but not used by WMATA. The train line module 326 includes relay devices that provide adequate current handling capability to energize the train lines 328 leading to the vehicle cam controller propulsion apparatus 330 in each car of the train. The control apparatus shown in FIG. 3 is provided in each A car, and is operative in the lead car of the train for controlling the propulsion system 330 in each car of the train through the train line 328 which passes through the train and couples with the propulsion system 330 in each car of that train.
In FIG. 4 there is shown the improved cam propulsion speed control system of the present invention, including a programmed microprocessor based speed maintaining and program stop logic apparatus 401. A weighted pendulum balance accelerometer 440 provides an acceleration indication signal. The elements shown in FIG. 4 that are common with FIG. 3 have been given the same reference numbers. The provision of the tachometer signal processing logic 441, the acceleration signal processing logic 446, the train line pattern logic 448, the marker decoding logic 450 and the serial link logic 452 is believed to be within the routine understanding of persons having ordinary skill in this art, and the functional operations of these logic devices are not unique to this application.
In general the time period required for the train vehicle to change from a present speed to a speed band limit is called the time to limit. The time period required for the train vehicle to respond to and provide a new requested tractive effort state is the delay time and is a function of the cam controller operation, the control system response, the propulsion motor operation in relation to the known motor curves, the mass of the train and so forth. A change of state will be made when the time to limit is determined to be equal or less than the delay time of the train.
The vehicle carried ATP or automatic train protection system responds to the track signal provided command speed for controlling the actual speed of the train to be within a desired speed band, such as a plus zero to minus four miles per hour band relative to that command speed. In addition there is an ATS system that provides an additional speed limit for each section of the roadway track. The speed maintaining control system responds to the lower of the ATP command speed limit and the ATS speed limit. When a program stop signal is provided in relation to a passenger station, the speed maintaining control system responds to the lowest of these three signals as the velocity reference signal.
A grade indicating marker is provided in the WMATA system in relation to each of predetermined portions of the roadway track and furnishes a grade indication signal that is considered to be too inaccurate to be used by the control apparatus and method of the present invention, which determines the grade by differentiating the speed indication signal from the tachometer in an averaging manner to remove noise and comparing this differentiated signal with the accelerometer signal. The previous WMATA control system obtained the grade from the provided marker.
The WMATA train control system operates with a cam controller and includes 12 voltage train lines connected to determine the operation of the propulsion motor control system. Respective binary voltage signals are placed on the train lines to establish a desired control pattern to determine the operation of each vehicle in the train in accordance with a wayside provided command signal. The propulsion equipment provides one of a desired power or a desired brake effort in response to the control pattern of the train line signals. The cam controller on the lead car of each train determines the discrete level of tractive effort for the whole train. It is required that the speed of the train be maintained within a predetermined speed band in relation to the command speed from the roadway track, as shown in FIG. 21. In addition there is a limited number of adverse control changes established, with an adverse change being defined as a decrease in either positive or negative traction effort. To control the train, responses are made to the acceleration of the train from a pendulum accelerometer and the actual speed of the train from a tachometer.
A time to limit determination, when the vehicle is accelerating, is made by establishing the difference between a predetermined speed band limit and the present actual speed of the train, with that difference being divided by the present acceleration of the train, in accordance with the relationship: ##EQU1## This time to limit determination, when the vehicle is decelerating, is made in accordance with the relationship: ##EQU2## An assumption is made that the present acceleration will not change substantially in the short period of time, about 100 milliseconds, before the microprocessor next performs this calculation. In addition the delay time required for the train to change from the present tractive effort state to the new tractive effort state is established as a function of where the present operation is located on the known motor curves, the mass of the train, etc. Then a comparison is made of this time to limit and this delay time by subtracting the delay time from the time to limit and for as long as the time to limit is greater than the delay time no state transition change is made in the operation of the cam controller, in accordance with the relationship: ##EQU3##
The head end vehicle of the train includes an automatic train protection system or ATP equipment that provides the command speed limit. In addition there is provided an automatic train supervision system or ATS which provides a speed limit that is considered safe for track conditions. The train speed determining cam controller responds to the lower of these two speed limits.
The weighted pendulum balance accelerometer apparatus measures the train acceleration, excluding the effect of track grade. A way to establish the total train acceleration is to differentiate the tachometer measured speed. The result would include the grade, but this calculation is considered too noisy for the desired speed control purpose. Furthermore, the wayside grade markers are too inaccurate to be used in obtaining grade acceleration. Therefore the grade acceleration is computed by differentiating the tachometer output and comparing the result to the accelerometer output and averaging the difference over an 800 millisecond time period. This can be done because the grade does not change very fast. This computed grade acceleration is then added to the instantaneous accelerometer acceleration to obtain the total acceleration. Differentiation of the tachometer output in an averaging manner to directly obtain total acceleration is prohibited due to the rapid changes possible in total acceleration.
A determination is made of all delay times associated with the state-to-state transitions of the train and these are placed in a fourteen by fourteen delay time matrix, (with provision being made for sixteen by sixteen in the microprocessor software). The time to limit calculation is updated ten times a second to determine the present time to limit (TTL). The delay time for a possible state transition is obtained from the delay time matrix and compared with the calculated TTL to see if the quantity TTL minus delay is greater than or equal to zero. When it becomes zero, a state transition is made. To fill a 14.times.14 decision matrix the 14.times.14 TTL transition reward matrix is multiplied by a 1.times.14 probability vector. The control system is designed to benefit by actual operation learning experience. Initially the probability or experience vector will include zeros and ones in its matrix. These could be changed to reflect how likely a particular state is to be chosen in accordance with real time learning experience such as a human operator would receive. The multiplication of matrices is per se a mathematically well known operation, i.e. the multiplication of the reward matrix by the probability vector to provide the next state decision matrix.
In FIG. 22, for program stop control of the train, and after the first program stop control marker is detected, the cam controller responds to the lowest of the ATS speed limit signal, the ATP command speed limit signal 2200 and the program stop speed limit signal 2204. The desired deceleration for program stop is minus 2 miles per hour per second. This can be implemented by effectively adding two to the determined train acceleration. The time to limit calculation for program stop control, if the train acceleration is above minus two miles per hour per second, then becomes TTL equals upper program stop speed limit minus speed divided by acceleration plus two, in accordance with the relationship: ##EQU4## The time to limit calculation for program stop control, when the train acceleration is less than minus two miles per hour per second, then becomes TTL equals lower program stop speed control limit minus speed divided by acceleration plus two in accordance with the relationship: ##EQU5## All units of speed are in miles per hour and all units of acceleration are in miles per hour per second.
There are eight power states, one through eight, with the larger number providing more propulsion effort. There are five brake states, nine to thirteen with the higher number providing more brake effort. There is one coast state providing essentially no effort. The grade must be considered, since when in power and going up a positive grade a return to coast or zero propulsion will usually be adequate to control the speed, but when in power and going down a negative grade, a return to coast may not be adequate to slow the train enough to turn the speed around, and a brake state might be required for this purpose.
In FIG. 5 there are shown the motor curves for a well known traction motor type 1462A. This traction motor is available at the present time from the Westinghouse Electric Corporation. These motor curves are for this traction motor operating at a system line voltage of 650 volts, and relate to tractive effort as a function of motor current for various speeds of operation. In relation to determining the time delays in the vehicle response to a change in command speed, it is necessary to establish how much time is required to obtain a new tractive effort to provide a transition change to the new command speed. The four lower curves marked FS1, FS2, FS3 and FS4 are for the series operation of the four propulsion motors on each car. The four upper curves marked FS1, FS2, FS3 and FS4 are for the parallel operation of these motors.
In FIG. 6 there are shown the motor curves for the same traction motor operating at a lower range line voltage of 455 volts.
In FIG. 7 there are shown the motor curves for the same traction motor operating at an upper range line voltage of 845 volts.
It has been determined that these motor curves can be divided into three bands of speed in relation to the time delay characteristics of the motor operation. Within the lowest speed band of zero to 23 miles per hour, the time delays are reasonably constant. Within the middle speed band of 23 to 50 miles per hour, the time delays are reasonably constant, and within the highest speed band of 50 to 75 miles per hour, the time delays are reasonably constant.
To obtain a tractive effort value for a certain vehicle speed and motor state and line voltage, the first step is to find the vehicle speed on the speed axis then proceed perpendicularly from the speed axis until the desired motor state curve intersects this speed. If the motor state curve does not intersect this speed one of two things is done. If the speed is higher in value than what the motor curve reaches, a nominal value in tractive effort should be assumed. If the curve starts above the desired speed the following formula should be used, the tractive effort equals the mass of the car times the maximum acceleration permitted by the propulsion equipment for the motor state divided by 21.95 miles per hour per second. From the point where the vehicle speed crosses the motor state curve, then proceed parallel to the speed axis until the tractive effort curve for the motor state is reached. If the motor curve is the FS1 curve, go to the tractive effort FS1 curve. If the motor curve is the FS2 curve, use the tractive effort FS2 curve, etc. Once this point is found, go parallel to the line current axis until the tractive effort axis is reached. This is the tractive effort value for that speed and motor state. By reversing this process a given tractive effort and motor state will obtain a vehicle speed. For the WMATA project, multiply the obtained tractive effort by four to give the total tractive effort per car since there are four motors on each car and the tractive effort per car is the sum of these motors.
In FIG. 8 there is shown the speed regulation or speed maintaining routine 800. When this routine is called, at step 801 a check is made to see if the train is operating in the ATO mode and not in overspeed. If not, at step 802 the trainline output data is cleared and the routine exits at return 803. If so, at step 804 there interrupts are disabled, and at step 806 the current trainline status is stored as ATO trainline status to ensure no confusion when switching from manual to ATO. At step 808 the desired data are input and placed in memory. The interrupts are again enabled at step 810. At step 812 a check is made to see if the speed limit is zero. If yes, the routine goes to the overspeed control step 1002 of FIG. 10 for requesting full service brakes. If not, at step 814 the grade determination routine of FIG. 24 is called. At step 816 a determination is made to see if the train actual speed SPD is less than or equal to 23 miles per hour, which relates to the lowest speed band shown in FIGS. 5, 6 and 7. Within each of the speed band shown in FIGS. 5, 6 and 7 the delay times are considered to be substantially constant for the purpose of the propulsion control operation for a transit vehicle including a particular type of propulsion control. If the actual speed is not within the lowest band at step 816, at step 818 a check is made to see if the actual speed is within the middle speed band. If the actual speed is not less than or equal to 50 miles per hour, at step 820 a block of delay data starting at tab 50 is loaded from ROM into a work area using the matrix transfer routine 900 shown in FIG. 9, where at step 902 the desired matrix is moved into the work area. If the actual speed is less than or equal to 23 miles per hour, at step 824 a block of delay data starting at tab zero is load from ROM into a work area. If the actual speed is greater than 23 miles per hour and less than or equal to 50 miles per hour, at step 826 a block of delay data starting at tab 23 is loaded from ROM into the work area. This delay data is taken from the appropriate motor curves shown in FIGS. 5, 6 and 7. Now a correction for the actual track grade is begun. The time delays loaded at one of the steps 820, 824 or 826 were based on a substantially level track and are appropriate for that track condition. However the WMATA track system can vary in grade from plus 4% to minus 4%, so suitable grade correction factors are required. At the appropriate one of steps 828, 830 and 832 the grade correction factors for the actual speed range is obtained in conjunction with the entry search routine shown in FIG. 25 at steps 834, 836 or 838.
The ATS subsystem 318 shown in FIG. 4 provides a full power or half power flag signal 321 that determines if a 50% limit is placed on the train acceleration. At step 846 in FIG. 8B a check is made to see if this half power flag is set, and if it is then at step 848 a check is made to see if the train line is requesting the three highest levels of power operation that will provide the three miles per hour per second maximum power. If the power state six, seven or eight is requested, at step 850 the state TH is set equal to zero, which is coast. At step 852 the change states subroutine shown in FIG. 13 is called to provide the necessary state changes and the bookkeeping functions. Then the train line output subroutine shown in FIG. 14 is called by the latter subroutine to output the next desired state pattern on line 325 shown in FIG. 4 leading to the cam controller propulsion system 330. The program operates such that the only way to get from power state six, seven or eight into one of the lower power states one, two, three, four or five is by returning to coast, and then to let the speed maintaining operation go to a desired one of the power states one to five. If the half power flag is not set at step 846, at step 853 a check is made to see if the program stop flag is set. If so, at step 854 a check is made to see if the operation is already in program stop. If yes, at step 856 the program stop speed band parameters are obtained, and at step 858 the P stop subroutine shown in FIG. 15 is called. At step 854, if the operation is not already in program stop, at step 860 a check is made to see if the marker update flag is set, and if not the routine goes to step 856. If yes, at step 862 the program stop flags used in a program stop control of a vehicle approaching a passenger station are reset and cleared. At step 864 the speed maintaining subroutine shown in FIG. 10 is called.
The speed maintaining subroutine shown in FIG. 10 at step 1000 checks to see if the actual train speed is greater than the command speed limit minus a control factor at the top of the speed band. At present this control factor UPB is set at plus zero, but it is provided to make available a conservative margin such as 1/2 mile per hour if desired in the speed control of the train operation. If yes, this indicates the vehicle is going faster than the command speed limit, so at step 1002 the train line pattern TH state is output to request state 12, which is full service brakes to correct the overspeed. The program then goes to step 1003 to clear state delay time and at step 1004 calls for the P change subroutine shown in FIG. 13 because of the requested change of state. If the actual speed is not over the command speed limit at step 1000, at step 1006 a check is made to see if the actual speed is less than or equal to the command speed limit minus a control factor at the bottom of the speed band, which at the present is minus four and a half miles per hour. This checks the actual speed in relation to the bottom of the predetermined speed band. If not, at step 1008 the time to limit determining speed band subroutine shown in FIG. 11 is called. If yes, the train is under speed so at step 1010 a check is made to see if the half power flag is set, and if it is not set then at step 1012 a state of eight is output, which requests full propulsion as shown in FIG. 20. The train is underspeed and full propulsion is requested. At step 1010 if the answer is yes, the maximum half power state is requested at step 1014. This is selectable, and can be set at state five or whatever state is desired for this purpose. In the speed maintaining subroutine shown in FIG. 10, at step 1006 if the actual speed is within the predetermined desired speed band such as plus zero to minus four and a half miles per hour below the command speed as shown in FIG. 21, the speed band subroutine shown in FIG. 11 is called at step 1008.
In the speed band subroutine shown in FIGS. 11A and 11B at step 1100 a check is made to see if the program stop flag is zero. If yes, at step 1102 a check is made to see if the train is accelerating, and if yes at step 1104 the probability rector is determined by obtaining from ROM a permission Vector. This vector can be obtained from RAM data corresponding to state probability history. If this is found to be necessary for a particular application, the filling of the reward matrix, steps 1114 to 1118 would be modified slightly to multiply each matrix element (TTL element) by the value in PVEC. At step 1106 a portion of the time to limit equation is determined and sets STTL (speed TTL) equal to the command speed limit minus the upper conservation factor in the upper part of the speed band, which is presently set at plus zero but can be selected as desired, and minus the train speed to establish how far the actual speed is below the upper speed limit. Going back to step 1102, if the acceleration is less than zero then at step 1108 the probability vector PVEC is set equal to a vector from ROM as previously described.
At the present time this permission or probability vector has all zeroes with the exception of a single one that selectively determines the next transition state of the train operation, but the capability is provided for future utilization with positive real numbers in this vector such that the decision matrix element value would be determined by a multiplication of this probability vector times the TTL reward matrix element value established in FIG. 12. The speed band subroutine could fill the probability vector with values in accordance with the probability of being in a particular state based on how often the operation has been there over a predetermined past period of time. At step 1112 the counter CNT is set to zero, and each time through the rest of this loop the counter CNT is incremented by one and when the counter equals the number of possible states the program operation is completed. If the probability vector were filled with real numbers this operation multiplies the reward matrix row element values by the probability vector element values. Presently, at step 1114 the P vector is shifted right to fill the P vector only for those states of the train operation permitted. At step 1116 the time to limit subroutine shown in FIG. 12 is called.
At step 1118 in FIG. 11 the present train line state TRNLN is put into memory location TH (train line hold), the counter is set to zero and CTTL is initialized. The ACT routine sequentially looks at the decision matrix row for each of the possible state transitions and selects the highest value and its associated transition. For each pass through the routine every time a value is found that is higher than the last value then the number of that new state is put into TH and the transition will be to this desired state. The decision matrix is the product of the probability vector times the reward matrix. At step 1120 if the half power flag is set, at step 1122 if the new train operation state is six, seven or eight then at step 1124 the new state is changed to the maximum half power state. If the half power flag is set at step 1120 and if the new state is not six, seven or eight at step 1122, or if the half power flag is not set, then at step 1126 a check is made to see if the desired state TH is the same as the present train line state. If yes, at step 1128 the same old data is output, and if not at step 1130 a check is made to see if a long enough waiting time has occurred since the last transition change to have the full performance benefit of that change. If not enough waiting time has taken place, the same outputs are refreshed at step 1128. If enough waiting time has been provided, at step 1132 N delay is set equal to the delay from the matrix in relation to the requested state transition, with the train line TRNLN being the present state and TH being the next desired state. At step 1134 the change states subroutine PCHG shown in FIG. 13 is called to make the desired change of state. The maximum delay is three seconds from full propulsion to full brake.
In FIG. 12 the delay associated with the particular state to state transition is established in relation to the corresponding element in the delay matrix DT. At step 1200 the limit is established as equal to the delay matrix element DT corresponding to the present train line state TRNLN and the new state CNT, where CNT equals TH from steps 1118 to 1119. At step 1202 a check is made to see if the program stop flag is set equal to zero. If yes, at step 1204 a check is made to see if the acceleration is positive or negative. If positive, at step 1206 TTLT is set to equal the vehicle delay time plus grade correction DD to give the overall delay times the acceleration which results in the maximum .DELTA. speed allowed. If negative at step 1204, at step 1208 TTLT is set equal to the vehicle delay time plus grade correction EE times the absolute value of the acceleration which results in the maximum allowed .DELTA. speed.
This .DELTA. speed divided by acceleration gives the time required to reach the speed limit from the present speed, but instead of dividing .DELTA. speed by acceleration and comparing it with a time delay, the time delay is multiplied by acceleration and compared to .DELTA. speed to facilitate the operation of the microprocessor.
If the program stop flag at step 1202 does not equal zero, then at step 1210 if the program stop flag equals one at step 1212 TTLT is set equal to the vehicle delay limit LIMIT plus the grade correction factor FF times the acceleration plus two. If the program stop flag is not one at step 1210, which indicates deceleration of more than two miles per hour per second in program stop, at step 1214 the limit plus grade correction FF is multiplied by the absolute value of the acceleration minus two. In program stop the control objective is dynamic equilibrium of the speed control at a deceleration of minus two miles per hour per second. At step 1216, TTLV is set equal to the time to limit TTLT minus 100 times STTL. STTL is the actual .DELTA. speed computed by taking the difference between the speed band limit and the actual speed. 100 is a scaling constant. The .DELTA. speed between the speed band limit and the actual speed is STTL as determined in FIG. 11 at one of the steps 1106 and 1110. TTLT was just computed as the allowable .DELTA. speed limit. A comparison is made of these and is put in the appropriate decision matrix row at step 1218. When the speed band subroutine is completed the appropriate decision matrix row will be filled with values representing the desirability of state transitions.
The change states subroutine in FIG. 13 at step 1300 checks to see if the new requested state is the same as the present state, and if yes at step 1302 the output train line is refreshed. If no, at step 1304 the delay time is read from the delay matrix. The direction of the last state-to-state transition (positive or negative) is determined by the direction of move macro routine shown in FIG. 19 and which is called in step 1306. At step 1306 the direction of move from the previous state to the present state determines the up down flag UPDN for that previous state transition. At step 1308 this flag is stored in location DIR1. At step 1310 the direction of the present move from the present state to the desired new state is found to determine the up down flag UPDN for the new state transition. At step 1312 a check is made to see if the new state transition up down flag equals the previous state transition flag, and if yes at step 1314 the new delay NDEL is set equal to the delay remaining for the last transition plus the delay for the new transition. If not, at step 1316 the new delay equals the absolute value of the new delay minus the remaining delay for the last move. At step 1318 a check is made to see if this new total delay time is greater than a predetermined maximum delay, and if yes at step 1320 this maximum limit is used in place of the new delay. At step 1322 the train line output subroutine shown in FIG. 14 is called. The purpose for this delay time addition or subtraction is to find the actual state-to-state transition time for use in the wait-out period after the new state transition is made.
The train line output subroutine in FIG. 14 at step 1400 disables the interrupts and at step 1402 resets the TTIME timer with the new delay time NDEL. At step 1404 the interrupts are enabled, and at step 1406 TR2 is set with the now old train line state TRNLN and the present train line pattern TRNLN is set with the new desired train line state. At step 1408 the train lines are output through the designated output ports leading to the train line module 326 and the cam controller propulsion system 330.
In FIG. 15 there is shown the program stop subroutine operative with the program stop marker signals received from the roadway track. The WMATA transit system has a first program stop marker positioned about 2700 feet from the center line of the passenger station platform. The second program stop marker is positioned about 1200 feet. The third marker is positioned about 484 feet, and the fourth is positioned about 160 feet from the center line of that station platform. The objective is to position the head end of the train at the proper position relative to the platform center line.
In the program stop subroutine shown in FIG. 15 at step 1500 the program stop table search subroutine shown in FIG. 17 is called. At step 1502 a comparison is made to see if the lesser of the ATP command speed limit and the ATS speed limit is greater than or equal to the program stop speed limit. If no, at step 1504 a flag is set for speed maintaining. If yes, at step 1505 the speed limit to be used is set to be the programmed stop speed limit, and the check flag is set equal to zero. At step 1506 a flag is cleared since the operation is not in speed maintaining. At step 1508 a check is made to see if the actual speed is less than or equal to the speed limit minus the upper correction, which is plus zero miles per hour. If no, there is an overspeed. At step 1507 a check is made to see if the train is operating in flare out and if yes, the flare-out subroutine shown in FIG. 18 is called. The flare-out subroutine provides a jerk-limited station stop to increase passenger comfort. If not, step 1509 checks to see if the train is in speed maintaining mode, and if not, step 1510 sets the state for full service brakes, the check flag is set to zero at step 1511 and the change states subroutine shown in FIG. 13 is called at step 1513. If there is no overspeed at step 1508, at step 1512 a check is made to see if the actual speed is less than the flare-out speed limit. If yes, at step 1514 a check is made to see if the program stop speed limit is less than or equal to the flare-out speed limit. If yes, at step 1516 the flare-out subroutine shown in FIG. 18 is called. If not, at step 1518 a check is made to see if the speed is less than or equal to the bottom of the speed maintaining band minus an additional programmable offset. If yes, at step 1520 the speed maintaining subroutine shown in FIG. 10 is called. If not, at step 1522 a check is made to see if the speed maintaining flag is set. If yes, at step 1524 the speed limit below program stop profile subroutine shown in FIG. 16 is called. This subroutine is used when the program stop speed limit is still above the command speed limit but closing in on the command speed limit and a transition to a program stop brake state must be made now in anticipation of the minus two mile per hour per second program stop speed limit deceleration. If the speed maintaining flag is not set, the train is operating in program stop and at step 1526 a check is made to see if the acceleration is greater than or equal to minus two miles per hour per second. If yes, at step 1528 the program stop flag is set for accelerating, and if not at step 1530 the program stop flag is set for decelerating. At step 1532 a check is made to see if the present train line state is emergency brake. If yes, at step 1534 the output is refreshed, and if not at step 1536 the speed band subroutine shown in FIG. 11 is called. At step 1538 a check is made to see if the train line state is zero, corresponding to coast. If yes, at step 1540 the train line output is refreshed, and if no at step 1542 the speed band subroutine is called.
As shown in FIG. 22, the program stop speed limit 2204 comes down at minus two miles per hour per second. If the command speed limit 2200 remains at some command speed, such as 40 miles per hour, when the train passes the 1200 foot marker where the program stop speed limit is a higher speed, such as 55 miles per hour, after a period of time, such as about 12.5 seconds, the program stop speed limit will intersect the command speed limit as shown in FIG. 22. To avoid overshooting the program stop speed limit 2204, since the speed maintaining calculation would provide an acceleration in relation to the program stop speed limit 2204 at the speed maintaining acceleration plus two, this subroutine shown in FIG. 16 is provided to correct such an overshoot of the program stop speed limit 2204 that might otherwise cause an undesired correction that would adversely stress the propulsion equipment on each vehicle.
In FIG. 16 the speed limit below program stop profile subroutine is shown. At step 1600 a check is made to see if the check flag is set to one. If yes, the subroutine has already been executed and at step 1601 the train line outputs are only refreshed. Since this flag is set to one later in this same program, the first time through this program it is not set to one. At step 1602 a check is made to see if the acceleration is less than or equal to minus two miles per hour per second, and if so the program stop flag is reset at step 1604 and the speed maintaining subroutine shown in FIG. 10 is called at step 1606. If the answer is no at step 1602, at step 1608 the program stop speed difference is set equal to the program stop speed limit minus the actual speed. At step 1610 some scaling is applied. At step 1612 the delay limit is obtained from the delay matrix for the present state in relation to a brake state, BSTATE, which is presently set at nine or minimum brake. At step 1614 the time to limit .DELTA. speed difference is computed by subtracting the delay limit plus a grade correction factor FF times the acceleration plus two from the scaled program stop speed difference. At step 1616 if the time to limit .DELTA. speed difference is greater than or equal to zero, no action is required at the present time so the program stop flag is set to zero at step 1618 and the speed maintaining routine is called at step 1620. If at step 1616 the answer is yes, then the program stop speed limit is too close to the command speed limit and corrective action must be taken to avoid overshoot. At step 1622 the BSTATE is requested which is state nine or minimum brake, and the check flag is set to one. At step 1624 the change states subroutine shown in FIG. 13 is called.
In FIG. 17 there is shown the seek subroutine operative to make a search of a predetermined program stop table. This table has two entries, one being the distance to the centerline of the passenger station platform and the other being the maximum MPH at that distance. The subroutine takes the present distance to go and searches the table for the next greater distance, with the corresponding MPH from the table being the maximum allowed MPH for the train at the present distance. At step 1700 the interrupts are disabled. At step 1702 the distance to go counter is read, and at step 1704 the interrupts are enabled. At step 1706 a comparison is made of the most significant byte of the three bytes in the distance to go counter in relation to the last pointer, and if the same, the old pointer is used at step 1708. If not, at step 1710 in an effort to determine which table should be used, a comparison is made with zero, and if yes, a new table entry pointer is loaded at step 1712 for that pointer. If not equal to zero at step 1710, then at step 1714 a check is made to see if this byte is equal to one, and if so, at step 1716 a different table entry pointer is loaded. If not equal to zero and not equal to one, at step 1718 a check is made to see if this byte is equal to two and if so, at step 1720 another table entry pointer is loaded. If none of the checks at steps 1706, 1710, 1714 and 1718 is yes, then at step 1722 a default is established and the table entry pointer is set for the highest value which is about 80 MPH. At step 1724 the most significant byte of the distance to go counter is stored as the past distance to go for use the next time through this loop. At step 1726 the least significant byte is compared with the pointer. If equal, at step 1728 the value in the pointer table is put into the past pointer table to use next time through the loop, and the program stop speed limit takes the corresponding entry speed limit value. At step 1730, if the distance to go counter is less than the entry value pointed to, a four byte back up is made at step 1732 since there are four bytes in the table for every entry, and this is repeated until the distance to go counter is less than the pointer value at step 1730. At step 1734 the pointer value is incremented by four locations, at step 1736 a search is made to find the next higher distance by comparing the distance to go with the pointer and if the distance to go is greater this indicates a value is found. At step 1728 the pointer is stored and the program stop speed limit is established. If less, at step 1738 the next higher distance increment is selected for the pointer and this loop continues until a distance to go value is found that is greater than the pointer distance.
In the flare-out subroutine shown in FIG. 18, at step 1799 the in flare-out flag is set and at step 1800 the next state is commanded to be coast and the time in flare-out is incremented 1/10 second since the routine is run every 100 milliseconds and each run is 1/10 second. At step 1802 a check is made to see if the counter is greater than a predetermined time, such as 3.6 seconds. If not, at step 1804 the change states subroutine shown in FIG. 13 is called to command the coast. If yes, at step 1803 a check is made to see if the profile speed is less than 1.5 MPH and if yes at step 1805 the next state is set at full service brake which is 12 and the timer value is set at maximum, then at step 1804 the change states subroutine is called and return. If no at step 1803 the routine goes to step 1804 and return. This subroutine provides jerk limiting when coming into a station by going into coast for a while and then full calling for service brakes.
In FIG. 19 the direction of move macro routine, which is used to determine the adding or subtracting of time delays, is shown. At step 1900 a check is made to see if the next or "to" state of the train is brake. If yes, at step 1902 a check is made to see if the present or "from" state is in brake. If the present "from" state is brake and the next "to" state is not in brake, at step 1904 the flag UPDN is set to one, since the move is down from propulsion or coast to brake. At step 1906, a check is made to see if the present "from" state is greater than the next "to" state, and if not at step 1908 a down movement is indicated.
If the "from" state is greater than the next "to" state at step 1906, then at step 1910 an up move is indicated. If not going to a brake state at step 1900, this means the next state is either coast or propulsion and at step 1912 a check is made to see if the "from" state is brake, and if yes, at step 1914 an up move is indicated. If the "from" state is not brake at step 1912, then at step 1916 a check is made to see if the "to" state is greater than the "from" state. If not, at step 1918 a down movement is indicated, and if yes, at step 1920 an up move is indicated by the UPDN flag.
The present control system uses the acceleration of the train in relation to the difference between the command speed band limits and the actual speed to establish a time to limit. When the train is accelerating in power, the upper command speed limit in relation to the predetermined speed band is used for this purpose. When the train is decelerating in brake, the lower band speed limit is used for this purpose. Then the vehicle response delay time is compared with the time to limit to determine when a state transition must occur. In this way a longer waiting period is provided between state transitions such that the train speed changes operate within the full width of the determined speed band and the required restriction on adverse state changes is satisfied. Reducing the width of the speed band, for example speed maintaining between plus zero and minus three miles per hour would result in more state-to-state changes.
In FIG. 23 there is shown a functional block diagram to illustrate the operation of the speed maintaining and program stop logic apparatus 401 shown in FIG. 4. This apparatus includes an Intel 8080 microprocessor programmed with the software control program shown in FIGS. 8 to 19. At block 2300 a check is made to see if the ATP equipment 310 shown in FIG. 4 is indicating an overspeed operation of the train on input 303 to the speed maintaining and program stop logic apparatus 401, which is done at step 800 in FIG. 8. When the ATP equipment 310 indicates an overspeed condition, block 2302 outputs full service brakes which is done at step 801 in FIG. 8. Block 2304 samples the input data, such as the ATP command speed limit on input 311, the accelerometer indicated acceleration on input 442, the ATS performance modification speed limit on input 319, and the grade marker information on input 313 (not presently used) and puts this in memory, which is done at step 803 in FIG. 8. At block 2306 the delay time matrix is filled using the motor curves and jerk limit requirements that were prevously put in the memory look up table in relation to the present state and the next desired state, such that the train speed is used to find the desired data from the look up table which data is then loaded into a memory buffer, and this operation is done at steps 806 to 818 in FIG. 8. Each location in that look up table corresponds to a state transition. At block 2308 the grade correction factors are obtained from a storage table that was previously established from empirical data, and which is done at steps 818 to 844 in FIG. 8. At block 2310 a check is made to see if a half power acceleration limit is provided on the input 321, which is done at steps 846 to 852 in FIG. 8. At block 2312 a check is made to see if the train is operating in a speed maintaining mode or in a program stop mode as determined by the program stop marker providing an indication on input 314, and which is done at step 853 in FIG. 8.
Assume the train is operating in speed maintaining mode. At block 2314 a determination is made to see if the actual speed on input 308 from the tachometer 306 is within the predetermined plus zero mile per hour to minus four and a half mile per hour speed band below the command speed on input 311, which is done at steps 1000 and 1006 in FIG. 10. It is readily apparent that this speed band can be adjusted in mile per hour width as may be desired to improve the speed control of the train. If the actual speed is above the speed band, at block 2316 full service brakes are output which is done at step 1002 in FIG. 10. If the actual speed is below the speed band, at block 2318 maximum permitted propulsion power is output which is done at step 1010 in FIG. 10. If the speed is within the speed band, at block 2320 the permission or probability vector is determined, which is a 1.times.14 vector that can have numbers according to the historical probability of each state but at the present time has all zeroes and a single one; the speed band subroutine is called at step 1008 in FIG. 10 and this determination is done at steps 1100 to 1114 of FIG. 11 to fill this probability vector. At the present time if the train is in propulsion state one and accelerating, the permission vector establishes the next transition state to be two and if decelerating the next transition state to be zero (coast). At block 2322 the time to limit calculation is performed, which is called at step 1116 of FIG. 11 and is done at steps 1200 to 1214 of FIG. 12. At block 2324 a comparison is made between the time to limit values with the state transition delays, which is done at step 1216 of FIG. 12. At block 2326 the 14.times.14 reward matrix is filled with the results of the time to limit calculations and comparison with the state transition delays, which is done at steps 1200 to 1218 of FIG. 12. The control concept here is to multiply the probability vector times the reward matrix to provide a decision matrix. However the probability vector at present is a permission vector with only zeroes and a single one in it, so the operation here is simplified to shift through the probability vector and for each one (presently a single one) the corresponding reward values are put into one corresponding decision matrix row for the 14 possible states to which a transition can be made from the present state, and which is done at steps 1118 to 1119 in FIG. 11 and includes blocks 2328 and 2330 in FIG. 23. At block 2332 a half power check is made, which is done at step 1120 to 1124 in FIG. 11. The final new state is then compared with the present state in step 1126. If the new state chosen is the same as the present state, at block 2334 the old train line pattern is refreshed which is done at step 1128 in FIG. 11. If a new state is chosen, in step 1130 a check is made to see if the delay time associated with the last change has expired. If it has not, the present train line pattern is refreshed. If it has expired, at block 2336 a transition delay time is obtained from ROM which is done at step 1132. Step 1134 calls the change states subroutine in FIG. 13. Bookkeeping in relation to the delay times is required because the train might not have reached the full tractive effort after the last state change so only a partial change has so far taken place, and it is necessary to consider this partial change in relation to a currently chosen new state transition. At block 2338 the new train line output subroutine shown in FIG. 14 is called by step 1322 of the change states subroutine shown in FIG. 13.
At block 2312 if a program stop marker was sensed, at block 2340 the program stop speed limit based on distance to go is found in a lookup table which is called at step 1500 in the program stop subroutine shown in FIG. 15 and is done by the seek program stop table search subroutine shown in FIG. 17. At block 2342 a comparison is made between the ATP command speed limit on input 311 and the program stop speed limit from block 2340, which operation is done at step 1502 in FIG. 15. If the command speed limit is less than or equal to the program stop speed limit, at block 2344 the ATP command speed limit is loaded for the control of train speed which is done at step 1504. At block 2346 an overspeed check is made which is done at step 1508. If an overspeed is determined, at block 2348 full service brakes is output which is done at steps 1510 to 1513. At block 2342 if the program stop speed limit is more restrictive, at block 2343 the program stop speed limit is loaded for the control of train speed, which is done at step 1506. At block 2345, an overspeed check is made which is done at step 1508. If the train is not in an overspeed condition, at block 2350 a check is made to see if the flare-out subroutine should be implemented which are steps 1507 and 1516. If flare-out is to be initiated, at block 2351 the flare out subroutine shown in FIG. 18 is executed, and this is done at step 1516. If not in flare-out, at block 2352 a check is made to see if the actual speed is below the plus zero to minus eight miles per hour program stop speed band, which is done at step 1518. If the speed is too far below the program stop speed limit, a return to speed maintaining is made at line 2356 and this is done at step 1520. At block 2358 a program stop speed limit look ahead check is made where the ATP command speed limit does not change but the program stop speed limit is decreasing at the rate of minus 2 miles per hour per second, such that the program stop speed limit with the time to limit calculation could become the more restrictive, which is done by the speed limit below program stop profile subroutine shown in FIG. 16 and is called at step 1524. At block 2360 the time to limit for a transition to BSTATE (minimum brake) is computed, which is done at step 1612. At block 2362 the time to limit is compared with the delay time, step 1614, and if the time to limit is greater at line 2364 a return is made to speed maintaining, which is done at step 1616 and 1620. If the time-to-limit is not greater than the delay time, at block 2366 minimum brake is commanded, which is done at step 1622 and 1624. If the train is operating in program stop, the look ahead check is no longer applicable, and block 2368 determines if the train is accelerating or decelerating relative to minus two miles per hour per second and this is done at step 1526 in FIG. 15. At block 2370 the accelerating or decelerating program stop flag is set and this is done at step 1528 for accelerating and step 1530 for decelerating. The operation then essentially goes back to speed maintaining at line 2372 by calling the band subroutine, FIG. 11, in steps 1536 and 1542.
In FIG. 24 there is shown the flow chart of the roadway track grade determination routine, which at block 2400 obtains the vehicle acceleration due to grade from RAM storage. The tachometer routine previously differentiated the averaged tachometer speed signal and includes the accelerometer output to calculate the acceleration due to the track grade, which is then added to the accelerometer value. This program determines the grade range of this acceleration. At block 2402, if the grade is greater than positive 2.5%, which corresponds to a certain value of the acceleration, at step 2404 a first bit in a flag word located in RAM is set. If the grade is between positive 2.5% and 0.75% at block 2406, corresponding to a different value of acceleration, another second bit in that five bit flag word is set at block 2408. At block 2410 if the grade is between positive 0.75% and negative 0.75%, at block 2412 a third bit in that flag word is set. At block 2414 if the track grade is between minus 0.75% and 2.5%, then at block 2416 a fourth bit is set in the flag word. At block 2418, if none of the other bits have been set by this routine, the fifth bit in the flag word is set. At block 2420, the grade data is stored in memory.
In FIG. 25 there is shown the flow chart of the matrix routine that determines the total number of entries in the grade correction factor table, which is arranged in five groups of three entries. The first is for accelerating during speed maintaining, the second is for decelerating during speed maintaining and the third is for program stop. At block 2500 the number of entries for each table group is loaded. At block 2502 the grade correction factors are cleared. At block 2504 the grade information is obtained, which is a word with five bits, and one of those bits is set by the grade determination program shown in FIG. 24. The first and least significant bit is for the steepest uphill grade and this ranges over to the fifth bit which is for the steepest downhill grade. At block 2506 the routine points to one of the vehicle speed ranges delete 00 between 0 and 22, delete 23 between 23 and 50, and delete 50 when over 50, as shown at appendix pages A25 and A26 to identify which of the five groups of grade correction factors corresponds to the bits of the grade determination. This is established by the carry set condition. At block 2508 the carry is rotated, and if the carry is not set at block 2510, the routine increments to point to the next of these areas. The carry is determined to be set when the area group is found that does correspond. At block 2512 the first entry of the group is obtained and stored in the DD factor. At block 2514 if the number from the table is negative, at block 2516 FF is put into the upper byte to signify a negative correction factor. At block the routine gets the second factor from the table, and at block 2520 checks if the number is negative. If so, at block 2522 FF is put into the upper byte. If not, at block 2524 the grade correction factor for the program stop is obtained, and a check is made at block 2526 to see if it is a negative number. If so, at block 2528 the upper byte is made FF, and if not an exit is taken from the routine.
In the time to limit subroutine shown in FIG. 12, the time required to reach the speed band limit is determined as a difference between the present speed and the speed band limit, which is the upper limit for acceleration and the lower limit for deceleration, divided by the train acceleration. This time to limit is then compared with the train response delay time. Since a microprocessor can multiply easier than it can divide, the actual calculation is to compare .DELTA. speed difference with the delay time multiplied by the acceleration. This comparison determines making or not making the state transition by setting the value in the reward matrix. One speed control objective is to wait as long as is reasonable before making a state transition to minimize adverse transitions, and another speed control objective is to stay within the speed band.
As compared to human speed control operation of a transit vehicle, a programmed microprocessor control system cannot see the track conditions as a human can, and the speed and acceleration sensing devices are not entirely accurate in the information they provide to the microprocessor system. Also vehicle delays, although measurable, vary with vehicle mass and the propulsion system supply line voltages and are therefore not known exactly at any given time.
The time to limit or TTL of the vehicle is the time remaining prior to the vehicle exceeding the upper or lower speed limit, and is a function of vehicle actual speed relative to the respective speed band limit and the vehicle acceleration. The time to limit must be compared to vehicle response delay time encountered prior to achieving a new state with sufficient tractive or retarding effort to maintain speed within the speed band. To determine the time to limit when accelerating, the time to limit equals the upper speed band limit minus the actual speed divided by the acceleration, and when decelerating the time to limit equals the lower speed band limit minus the speed divided by acceleration. When the time to limit equals or falls below the vehicle response time, the state transition must occur. Therefore if the time to limit minus the delay is greater than zero, no state transition takes place. If the time to limit minus delay is less than or equal to zero, the state transition occurs.
In order to provide sufficient corrective action under most conditions and to minimize the number of adverse state transitions, the state transitions which result in decreases in positive or negative tractive effort will be returned to coast. Increases in positive or negative tractive effort will step through the successive train line patterns, since each such change is not penalized by the system specification. The WMATA motor curves indicate that the vehicle response delay times change relative to speed. These delays are set to be conservative even if the worst case vehicle mass and line voltage conditions are considered. Once a transition occurs another speed maintaining transition is not permitted until the first transition delay time elapses.
Depending on the grade of the roadway track, when descending for example a negative 4% grade a return from propulsion to coast may not guarantee deceleration of the train. Similarly ascending a positive 4% grade for example a return from brake to coast may not guarantee acceleration of the train. Therefore the state transition delay times must be modified to allow for subsequent transitions to provide sufficient tractive effort relative to the grade condition. Grade information is available from the track markers which indicate track grade in one of five levels. This information was found to be insufficient to provide delay correction factors which result in satisfactory train performance for various grades, train masse and line voltages. Therefore, the grade is computed using tachometer differentiation and accelerometer acceleration as previously described. The previous tractive effort state history processed continually can further optimize performance relative to variables such as weather, line voltage and vehicle mass.
For any given state, 14 state transitions exist including the stationary or no change transition. Some of these transitions are desirable to maintain train speed and some are undesirable and others are unobtainable due to propulsion equipment configuration. A control approach is required to manipulate the potential 196 state transitions and the decision mechanism in a compact and efficient manner. The logical nature of the state transitions and microprocessor based control is operative with a state transition decision matrix. The elements of this state transition decision matrix representing the desirability of each state transition can be obtained from a 14.times.14 reward matrix and a state transition probability vector. The 196 elements of the reward matrix are each a function of the present train line pattern or state, the new state, vehicle speed and acceleration, speed maintaining band, vehicle delays and grade. The state probability vector is defined such that its elements represent the probability of being in state i after N transitions. The state probability vector elements will vary with such variables as weather conditions, line voltage and vehicle mass. Given the reward matrix and the state transition probability vector, the elements of the state transition decision matrix can be determined and the best state transition decision can be made. Each state transition decision matrix element is assigned a value based on the aforementioned parameters. These values indicate the overall desirability of choosing that particular state transition. The state transition associated with the matrix element with the highest desirability is implemented. Filling of the matrix and choosing the element of greatest value is a repetitive and computational process well suited to microprocessor based control systems. This approach to transition decision making also results in improved logic understanding and facilitates trouble shooting and modification when necessary.
For programmed station stopping, the speed control logic employing a state transition decision matrix with time to limit based elements is also applicable. The WMATA marker system provides program station stopping distance information at 2700, 1200, 484 and 160 feet from the center line of the station platform. From this distance information, a speed profile can be generated which provides the specified two mile per hour per second deceleration rate. After passing a marker the cumulative distance travel information can be maintained with reasonable accuracy by integrating the tachometer. Since once under program stop subsystem control the speed limit decreases at two miles per hour per second and the train deceleration must also approximate this rate. To achieve this deceleration the time to limit TTL calculation must be modified such that if acceleration is above minus two miles per hour per second, then TTL equals the upper program stop speed limit minus the speed divided by acceleration plus two, and if acceleration is below minus two miles per hour per second then the time to limit TTL equals the lower program stop speed limit minus speed divided by acceleration plus two, where all units of speed are miles per hour and all units of acceleration are miles per hour per second.
The transition decision matrix elements are modified not only by the time to limit calculation procedure change but also in the decision mechanism logic used for selection of certain states. For example only brake states are typically permitted when the program stop speed limit is controlling and the speed is within plus zero and minus four miles per hour of this limit. Once a program stop speed marker is detected a check of the time to limit is continually made using both the commanded speed limit and the program stop profile speed limit. The more restrictive limit is chosen and its logic executed. A flare-out technique is also implemented just prior to stopping. This routine consists of a return from brake to coast for one to two seconds and then an application of full service brakes. In this manner a jerk limited stop is achieved providing increased passenger comfort.
DESCRIPTION OF INSTRUCTION PROGRAM LISTING
In Appendix A there is included an instruction program listing that has been prepared to provide the desired operation of a transit vehicle in accordance with the disclosed speed control apparatus and method. The instruction program listing establishes the functional operation of the present invention, and is written in the assembly language of an Intel 8080 microprocessor computer system. Many of these computer systems have already been supplied to customers, including customer instruction books and descriptive documentation. A person skilled in this art could routinely apply the attached program in relation to specific transit vehicles to be controlled for a particular vehicle track system.
The following page is Appendix -A1- ##SPC1## ##SPC2## ##SPC3## ##SPC4## ##SPC5## ##SPC6##

Number	Name	Date
3218537	Stamm	Nov 1965
3305712	Hoffman	Feb 1967
3324373	Stamm	Jun 1967
4217643	Anderson et al.	Aug 1980
4270716	Anderson	Jun 1981
4282466	Matty	Aug 1981
4410154	Matty	Oct 1983
4459668	Inoue et al.	Jul 1984

Speed control apparatus and method for rapid transit vehicles

Information

Patent Number

Date Filed

Date Issued

Inventors

Original Assignees

Examiners

Agents

CPC

US Classifications

Field of Search

US

International Classifications

Abstract

Description

Claims

US Referenced Citations (8)

Non-Patent Literature Citations (3)

Entry
Krings B. J., Alternative Systems for Rapid-Transit Propulsion and Electrical Braking, Westinghouse Engineer Mar. 1973, vol. 33, No. 2, pp. 34-41.
Hillman et al., The Washington Metro Automatic Train Supervision System, ASME Publication 73-ICT-79, Sep. 1973, Intersociety Committee on Transportation Conference.
Greenway et al., Automatic Train Control and Communications for Washington Metro, Communications Society IEEE, Nov. 1974, pp. 14-21 and 30.