Applicant hereby claims foreign priority benefits under U.S.C. §119 from Danish Patent Application No. PA 2006 00845 filed on Jun. 22, 2006, the contents of which are incorporated by reference herein.
The present invention relates to a fluid controller, e.g. for use in a steering system for a vehicle. In particular, the invention relates to a fluid controller operable to control a flow of a fluid from a source of pressurized fluid to a pressure operated device, the controller comprising:
Machines such as vehicles or ships, and in particular off-highway machinery such as wheel loaders, excavators, dozers, tractors, harvesters and similar heavy duty machines often operate with hydraulic steering systems. In fluid controllers for steering a vehicle or a ship, the valve member, e.g. in the form of a spool, often forms a neutral configuration and two operating configurations on opposite sides of neutral, corresponding to a left-turn and a right turn of the vehicle. Similar fluid controllers are, however, also applied in machines of a completely different kind, e.g. for controlling lifting in a crane, movement of robots, or in general for controlling a machine.
Typically, the steering system receives a steering input from an operator of the machine who thereby indicates a desired activity of the machine, e.g. a desired angular movement of a steering wheel of a vehicle. The steering input could be provided by an operator via a handle such as a steering wheel, a joystick, a push button etc. In a fluid controller, the steering input is converted into a hydraulic flow to a pressure operated device, e.g. a hydraulic actuator or cylinder which moves the steering wheels of a vehicle. A fluid controller of the above mentioned kind is disclosed e.g. in U.S. Pat. No. 5,992,458 and in U.S. Pat. No. 5,638,864 wherein the controller is disclosed in combination with an unequal area steering cylinder. Typically, the fluid controllers comprise a housing with a valve member which is movable in the housing and which cooperates with the housing to open and close various passages between the source and ports of the pressure operated device and possibly between other ports of the pressure operated device and a receptacle for the fluid.
Typically, the steering input which is provided on the handle is converted into a reference which is indicative of a desired position of the valve member relative to the housing. Based on the reference, the valve member is moved in the housing, e.g. by use of an electrically or hydraulically operated actuator. Some of the existing fluid controllers comprise a sensor which provides an output indicative of an actual position of the valve member relative to the housing.
To detect faults in the system, an integrated observer may determine a theoretical model output which is indicative of a theoretically correct position of the valve member relative to the housing based on a specific reference. The observer may thereby provide a residual value which indicates a possible difference between the output as provided by the sensor and the model output determined from the reference and a model of the system.
By comparing the residual value with a threshold value, an alert can be provided when a specific reference does not lead to a desired position of the valve member relative to the housing. This situation may occur e.g. if movement of the valve member in the housing is restricted.
U.S. Pat. No. 6,766,230, U.S. Pat. No. 6,470,300 and U.S. Pat. No. 6,687,585 all disclose various fault detection systems and methods for vehicles, engines etc.
In the existing fluid controllers, a discrepancy exist between on the one hand to set the threshold value relatively close to zero thereby to detect a majority of the situations in which the valve member does not move in an intended way, and on the other hand, not to be alert unless there is an error which is significant.
It is an object of a preferred embodiment of the invention to improve fluid controllers of the above-mentioned kind, and in particular to provide a system which at one time is sensitive to detect important errors and which is not too sensitive to be interrupted by inessential errors.
Accordingly, the invention provides a fluid controller of the kind mentioned in the introduction wherein the fluid controller is adapted to provide a gradient of the reference and to scale the threshold value based on the gradient. Since faults are detected merely by use of a sensor which detects the position of a valve member, i.e. corresponding to the spool sensor existing already in many commercially available steering systems, a more complicated closed loop control of a steering system by use of wheel position sensors etc. can be avoided. The invention thereby provides for a simple and reliable steering system.
The gradient expresses the speed at which the reference is changed, e.g. in increments per time unit. If a reference* is received a time step of a length X after a previously received reference, the gradient g could be expressed generally as:
In practice x could be a fixed value, i.e. the reference is delivered by the processor at a fixed frequency.
If the gradient is high the residual value will tend to be high, and a fault alarm is disposed to be activated unjustifiably. According to the invention, the threshold value may be increased as a function of the gradient, and the sensitivity of the fault alarm is therefore adjusted based on the gradient. This facilitates a system with less unjustified alarms and with an acceptable sensitivity. The threshold value could e.g. be scaled proportionally to the gradient. As an example, the threshold value could be expressed as an integer multiplied with the gradient.
The comparing of the threshold value with the residual value may be used to trigger a fault alarm, e.g. for stopping the machine which is controlled by the fluid controller.
The processor could be a computer system which forms part of the fluid controller, e.g. formed in one component with the housing and valve member or at least be attachable to the housing and valve member. The processor could be software implemented in a computer system, e.g. comprising a standard integrated circuit for computers such as PCs or in an ASIC.
The reference could be expressed by a number, and the gradient could be expressed as the percentage by which the number has changed since the last time a reference is received or a value by which the number changes per time unit. Correspondingly, the threshold value may be expressed by a number, and this number is changed as a function of the gradient.
Correspondingly, the residual value could be expressed by a number which is compared with the threshold value for generating an alarm signal under predetermined conditions.
The observer could be implemented as software in a standard computer system such as a PC or the like, or the observer could form part of an application specific integrated circuit (ASIC) and be an integrated part of the processor of the fluid controller.
The valve member could be either slidably or rotatably received in the housing, and cooperation between the housing and the valve member may open and close various passages depending on the position of the valve member relative to the housing. As an example, a passage between the source of pressurised fluid and one out of several steering ports, e.g. for right and left turning of a movable machine could be opened simultaneously with opening of passages between other of the ports and a receptacle for collection of fluid which has been used for steering purposes in a previous steering sequence.
As it will be discussed in further details later, the observer may operate in accordance with the Luenberger observer-based fault detection method.
In one embodiment, the reference is provided at a fixed frequency.
When the residual value has been compared with the scaled threshold value, the fluid controller may activate an alarm signal, e.g. if the residual value exceeds the threshold value. The alarm signal could be any kind of signal, e.g. a stop signal which prevents further operation of the machine in question, an acoustical or optical alarm etc.
The invention is particularly useful in combination with a Steer by Wire (SbW) steering system, and in particular in combination with an SbW system with two parallel systems. One of the parallel systems may operate the steering system under normal conditions. An error in this main steering system can be detected in accordance with the present invention, and the steering system could be adapted, based on a detected fault, automatically to shift to the other one of the two parallel systems. In this case, the driver could be notified to have the main system repaired and optionally be instructed to drive at a low speed.
In a second aspect, the invention provides an error detection method for a fluid controller of the kind comprising:
In a third aspect, the invention provides a steering system for a vehicle, said system comprising a fluid controller operable to control a flow of a fluid from a source of pressurized fluid to a pressure operated device, the controller comprising:
Any of the more specific details mentioned relative to the first aspect of the invention may apply to the second and third aspects of the invention.
In the following, a preferred embodiment of the invention will be described in further details by use of the following denotations:
PVE: An electrical actuation unit for hydraulic proportional valves
PVE H-bridge: see
LVDT: Linear variable displacement transducer
ASIC: Application specific integrated circuit.
PVG32: A proportional valve.
RHP: Right half-plane
SbW architecture: Steer-by-wire architecture
1oo1: one-out-of-one
1oo1D: one-out-of-one with diagnostics
2oo2: two-out-of-two
3oo3: three-out-of-three
The embodiment is described with reference to the drawing in which:
Further information is available in JIE Chen, Ron J. Patton (1999). Robust Model-Based Fault Diagnosis for Dynamic Systems. Kluwer Academic Publishers.
In the following example, a PVE, i.e. an electronically actuated valve, is applied in an H-bridge as disclosed in
An LVDT sensor measures the actual spool position whereafter an ASIC compares this to the commanded reference. A P-controller is used for controlling the duty cycle in a 40 Hz pwm signal that directly controls the switch patterns for the four on-off valves. In
Although the two plots show fairly good correspondence, the simulation model, when based solely on first principles, seems to miss the initial inverse behaviour as the measured response exhibits, see
To account for this the simulation model is extended with a right half plane (RHP) zero. With this modification the simulation model response corresponds closely to the measured data, see
From experiments it is found that the PVE dynamics may be approximated by the following transfer function:
In
As seen from
In this section it is chosen to explore the applicability of a Luenberger observer-based fault detection method [?]. However before introducing the observer, the PVE transfer function (1.1) is first considered in a state-space formulation:
{dot over (x)}(t)=Ax(t)+Bu(t)+R1f(t)
y(t)=Cx(t)+Du(t)+R2f(t) (1.2)
where R1 and R2 are the fault entry matrices. Next the observer structure is introduced:
ż(t)=Fz(t)+Ky(t)+TJu(t)
w(t)=Gz(t)+Ry(t)+Su(t) (1.3)
where z(t) is the observer state vector. If the system (1.2) is observable the observer exists and its output w will asymptotically approach a linear transformation of the system state x, that is limt→∞[w(t)−Lx(t)]=0, where L is a matrix performing the linear transformation, if the following conditions are met [?]:
F has stable eigenvalues
TA−FT=KC
J=TB−KD
RC+GT=L
S+RD=0 (1.4)
where the matrix T is a transformation matrix.
If the matrix L is chosen as L=C then an output estimation is achieved:
ŷ(t)=w(t)+Du(t) (1.5)
From (1.2) and (1.5) the residual vector r(t) is formed:
Now by combining (1.3) and (1.6) the following residual observer is obtained:
ż(t)=Fz(t)−Ky(t)+Ju(t)
r(t)=L1z(t)+L2y(t)+L3u(t) (1.7)
If we form the error vector as:
e(t)=z(t)−Tx(t) (1.8)
then the error dynamics can be expressed as:
This combined with the requirements in (1.4) reduces to:
ė(t)=Fe(t)+KR2f(t)−Tr1f(t) (1.10)
The error dynamics thus solely depends on the presence of faults.
The residual vector may be expressed as:
If the matrices are chosen such that:
L1T+L2C=0
L3+L2D=0 (1.12) (1.13)
then (1.11) reduces to:
r(t)=L1e(t)+L2R2f (1.14)
The residual vector is seen to solely depend on the presence of system faults.
In the following this method of fault detection is implemented on the validated first principle PVE model. It is chosen to use a full order observer where the observer poles are six times as fast as the poles in the observed system, i.e. the system in (1.2). The Q matrix is chosen as Q=130000.
When simulating the steering valve spool position without any valve failures, the response becomes as sown in
Next the fault scenario where the spool returns to neutral is considered. This kind of fault will not be detected by the existing fault detection method already implemented in the PVE, since the consequence of the fault does not cause the spool position to exceed the reference. In
From
It may be argued that even though the observer does not immediately detect when the fault occurs, it does rather quickly detect when the valve fails. Looking at the valve spool position response in
The magnitude in spool position error is thus critical for quick failure detection and because of this it may seem advantageous to use a small threshold value. However as seen from
To overcome the necessity of large fixed thresholds in order to provide robustness to false alarms it is now suggested to use the gradient of the reference signal to adaptively scale the threshold value. In situations where large gradients occur the threshold is momentarily increased. In this way the false alarm peaks will not result in real false alarms. In
As seen from
The observer based fault detection method used is particularly useful for pilot operated hydraulic valves, as the considered PVE, since practically no external disturbances exist. In case external disturbances did exist, they would impact the state velocity vector in (1.2) in the same way faults would do. It will in this situation be required to decouple the external disturbances from the residual generator. One way of doing this may be to use an unknown input observer [?]. However, for the PVE case considered here this will imply that at least one additional state must be measured in order to both decouple the disturbance and detect failures. If the unknown input observer approach is adopted it will be required to model how the external disturbances map into the state velocity space. This may prove to be a difficult task. However, for pilot operated valves do practically only flow-induced spool forces act as external disturbances. But since the generated actuation force, from the pilot pressure, is usually significantly larger at nominal flow rates than the flow induced forces, it may be justifiable to neglect external disturbances in the model description.
In the design procedure it has now been shown that primary steering valve failures are detectable. The required safety strategy regarding online fault detection is therefore fulfilled. The final step is now to design a dynamic performance control strategy and validate it together with the performance of the fault detection method. This will, however, not be done in this case study as it is similar to what is done in conventional design methods.
While the present invention has been illustrated and described with respect to a particular embodiment thereof, it should be appreciated by those of ordinary skill in the art that various modifications to this invention may be made without departing from the spirit and scope of the present invention.
Number | Date | Country | Kind |
---|---|---|---|
2006 00845 | Jun 2006 | DK | national |
Number | Name | Date | Kind |
---|---|---|---|
5115640 | Stephenson et al. | May 1992 | A |
5313389 | Yasui | May 1994 | A |
5386365 | Nagaoka | Jan 1995 | A |
5638864 | Haarstad et al. | Jun 1997 | A |
5992458 | Novacek et al. | Nov 1999 | A |
6014598 | Duyar et al. | Jan 2000 | A |
6470300 | Benzinger et al. | Oct 2002 | B1 |
6687585 | Rizzoni et al. | Feb 2004 | B1 |
6766230 | Rizzoni et al. | Jul 2004 | B1 |
6795761 | Lee et al. | Sep 2004 | B1 |
6816804 | Lee | Nov 2004 | B1 |
20040143379 | Borroni-Bird et al. | Jul 2004 | A1 |
20050159866 | Takeuchi et al. | Jul 2005 | A1 |
Number | Date | Country | |
---|---|---|---|
20070295924 A1 | Dec 2007 | US |