Patent Application
20090240401
The invention relates to a trapping prevention means and a method for controlling and regulating a motor-driven adjusting device, in particular a seat adjusting means in a motor vehicle.
A trapping prevention means is necessary in motor-driven seat adjusting devices in motor vehicles, for example in window winders, sliding roofs, sliding doors, tailgates etc, for safety reasons, in order to stop and possibly reverse the motorized drive when necessary, that is to say if an object or body part is trapped. Trapping prevention of this kind is in particular also desired in motorized seat adjustment means. Characteristic variables of the motorized drive are normally evaluated in order to determine whether trapping has occurred. Such characteristic variables are, for example, the motor voltage, the motor current or the rotation speed. The motor moment is normally determined from these characteristic variables, and an excess force is determined from said motor moment in turn. The excess force is given by the difference between the total force exerted by the motor and a total adjusting force which is required, in particular, to overcome the friction and to accelerate the adjusting device. However, it is difficult to determine the adjusting force since, for example, the friction can vary during the course of the adjustment process on account of areas with severe running difficulties. In addition, aging effects or else temperature influences can have a considerable influence on the friction. Temporarily varying acceleration forces are also taken into account when determining the excess force. Thus, for example, according to EP 1 310 030 B1, a large number of individual forces are added up at a summation point in order to determine the resulting excess force and an excess force or a trapping force is determined by comparison with the force currently exerted by the motor.
EP 1 299 782 B1 discloses a trapping prevention means in which the current profile of the force exerted by the motor over the adjustment path is compared with the profile of the force of a previous actuation process. However, if there is a relatively long period of time between the two actuation operations, the ambient conditions, for example temperature, may have significantly changed. Employing the force profile of a previous actuation process may therefore prove problematical in order to be able to use the previously measured force profile as the current profile of frictional force.
EP 0 714 052 B1 discloses a trapping prevention means for a side window or a sliding roof, in which the adjustment process is divided into equal time windows which lie in the region of 100 msec. In this case, this time window width should be selected on the basis of the trapping instance to be detected which occurs most slowly. In order to determine the excess force, the measured values of the current time point are compared with those of a reference time point which is at a distance of one window width from the current measurement time point and at which trapping has not occurred.
Reliable identification of trapping prevention in the event of seat adjustment is more complex than in relation to window winders or sliding roofs, in the case of which the glass pane moves toward a fixed stop.
The invention is based on the object of providing a simple trapping prevention means and also a straightforward method for reliably detecting a trapping instance, particularly in the case of seat adjustment.
According to the invention, the object is achieved by means of a trapping prevention means as claimed in patent claim 1. Accordingly, provision is made for a plurality of movement classes to be defined and for a distinction to be made between said movement classes in order to monitor for a trapping instance, and for a decision criterion to be derived from detected characteristic variables of the motorized drive, on the basis of which decision criterion the current state of the adjusting device is associated with one of the movement classes. In this case, the movement classes include, in addition to a running difficulty of the adjusting device, trapping of an object and run-up against an end stop, in particular, also the movement class of a sudden reaction and/or the movement class of a load movement of a load on the adjusting device.
The distinction between these movement classes is based on the consideration that, in the event of seat adjustment, particular situations can occur which have to be taken into account in the evaluation. Firstly, when a person is trapped, a so-called panic reaction can be expected as a sudden reaction. It is therefore assumed that, in particular situations and depending on the person concerned, said person will brace himself against the adjusting movement of the seat with all his force if he subjectively senses a dangerous situation.
As an alternative or in addition to the movement class of sudden reaction, the load movement of a load on the adjusting device is provided as a further movement class. This movement class applies when the person sitting on the seat moves during the adjustment process. On account of a change in load of this type, the current total load of the motor can be both increased and reduced. All the movement processes which are essential to the decision are covered by classification into this total of five movement classes in particular, and so a trapping instance can be reliably identified with only a low error rate.
The motor torque or a variable which is correlated with the motor torque is usually used as the decision criterion. This correlated variable is, for example, the motor rotation speed detected as a characteristic variable or else the motor current. The profile of the motor moment in the event of a panic reaction or a load movement differs from a normal trapping instance, in which only the seat moves toward an object. Distinguishing between these movement classes, in particular also identifying a movement class in relation to the sudden reaction and/or the load movement, therefore ensures that that special trapping situations which differ from the typical and normal trapping instance are also detected and identified as such.
In this case, a running difficulty of the adjusting device is understood to be the total friction, which has to be overcome by the drive motor, of the adjusting device, with this total friction usually varying over the adjustment path during the adjustment process and sometimes also including running difficulty peaks. Trapping of an object, in particular a person, is here understood to be the case in which the seat is moved toward a person who is either sitting on a back seat and is therefore pushed into the rear seat, or who is sitting on the seat to be adjusted and is moved, for example, toward the steering wheel or the dashboard, but without having to exert an excessive counter-force. That is to say, in this case, the normal trapping situation in which the person does not exhibit any pronounced reaction is assumed. Finally, the movement class of run-up against an end stop involves the situation in which the seat adjusting means moves into its front or rear end position in the event of a translatory adjustment or into the upright or inclined end position when the inclination of a backrest is adjusted. These end positions are usually defined by a mechanical end stop.
According to a preferred refinement, a spring model for the adjusting device is used as the basis for classification purposes and at least one spring constant is derived from the detected characteristic or input variables as the decision criterion. The use of a so-called spring model is based on the consideration that, on account of the compliance of the cushioning in a seat in the event of a trapping instance, said cushioning yields in the manner of a spring and therefore exerts a spring force against the adjusting movement. This is proportional to the path covered, with the proportionality factor being the spring constant. This spring constant is used as a decision criterion, that is to say the value or a variable derived from said value of the spring constants is used to make a decision as to which of the movement classes the current state of the adjusting device is to be associated with.
In this case, the spring constant is a variable derived from the total load of the motor. Therefore, a characteristic change in the total load of the motor, preferably a characteristic change in the motor moment, is used, in particular, as a decision criterion. The total load of the motor is therefore understood to be, in particular, the total torque exerted by the motor or the resulting total force exerted by the motor. Since other characteristic variables of the motor, for example the motor current or the motor rotation speed, are linked to the motor moment, it is furthermore also possible to use the motor current or, for example, the motor rotation speed as the decision criterion, in addition to the motor moment. The spring constant is preferably determined from the change in the motor moment or one of these characteristic variables.
In this case, the mathematical derivative of the total load is preferably used as the decision criterion. The derivative is generally understood to mean the change in the value of the total load in an interval, for example a time or distance interval. In this case, these intervals may be both infinitesimally small in the mathematical sense and also have predefined, fixed interval widths, so that the values for the total load have to be detected or determined only at defined sampling points. Since the total load is correlated to the force exerted by the motor, the spring constant or at least a variable which correlates with this spring constant can be directly given by the derivative of the total load.
According to a preferred development, the same value range for the decision criterion, but with different profiles of the decision criterion, is associated with the movement class of the load movement and the movement class of run-up against an end stop. This refinement is based on the knowledge that a load movement and run-up against an end stop in the spring model are represented by a spring constant of a comparable level, but the spring constant is highly time-dependent in the case of a load movement. In contrast, the mechanical stop can be described substantially by a constant spring constant. In other words, this refinement is based on the consideration that load influences can lead to a sharp increase in the total load of the motor in the short term but this is considerably reduced again after a short period of time, whereas the total load of the motor increasingly rises in the event of movement towards an end stop.
Different value ranges for the derivative are expediently associated with the individual movement classes. The lowermost value range is associated with the movement class a) the running difficulty, the following value range is associated with the movement class b) trapping of an object, the subsequent value range is associated with movement class c) run-up against an end stop, and the highest value range is finally associated with the movement class d) the sudden reaction. Identification of the respective movement classes and therefore identification of a trapping instance, specifically identification of the movement class b) trapping of an object and d) sudden reaction, is therefore ensured on the basis of these value ranges for the derivative, with delimitation with respect to further movement classes too.
According to an expedient development, the values or value ranges for the decision criterion, in particular the value ranges for the derivative and further threshold values or variables and values derived from the derivative, which are required for the classification operation, are determined with the aid of a measurement process on a physical model. In this case, the measurement results obtained are stored as values which can be used in the classification operation. This is done, for example, by the parameter values being stored in a table or a characteristic map and an unambiguous association of the individual values to the different movement classes being taken from this characteristic map. As an alternative, an association function can be provided on the basis of these values in the manner of a fuzzy logic. Instead of measurement on a physical model, a theoretical model or empirical values can be used as an alternative or in addition.
The profile of the spring constants or the derivative, that is to say the change in said spring constants, is preferably used for the association to the individual movement classes, in particular whether the movement class b), trapping of an object, is present. In this case, a trapping instance is identified when the value of the spring constants/derivative remains constant or possibly increases in a certain way. This is based on the consideration that, in the event of a normal trapping instance, that is to say without a panic or sudden reaction, the trapped person is expected to exert a certain counter-force. In the spring model which forms the basis, this is expressed by the spring constant (spring stiffness), which characterizes the compliance of the cushion, being superposed by a counter-force exerted by the person, so that the resulting spring constant increases. The check as to whether the value of the derivative increases therefore takes into account the expected behavior of a person in the event of a trapping instance.
In addition, identification of a trapping instance is preferably based on a predefined lower load threshold value, that is to say a predefined motor moment or a total force which is derived from this, being exceeded. The relevant decision criterion is determined only after this is exceeded. This is based on the consideration that an indication of a trapping instance is present only when there is a significant change in the total load, and that it is necessary to evaluate the profile of the total load with regard to the decision criterion and with regard to the presence of a trapping situation only in this case.
With regard to evaluation which is as simple as possible, at least and preferably exactly three load threshold values are defined in this case, with one value of the decision criterion in each case being determined and evaluated between two load threshold values in each case. Since the decision criterion is primarily considered to be the derivative of the profile of the total load, that is to say the change in the total load, meaningful evaluation without a great deal of outlay on computation is possible by means of this measure even on the basis of few measurement and detection points. In order to determine the derivative, the respective value pair at the three load threshold values in particular is, in this case, stored and suitably interpolated, for example linearly to the next value pair. The value pairs are formed from the respective load threshold value and an associated variable value, for example distance or time. This interpolation is then used to determine the value of the derivative for the respective interval of the variables, for example a specific time or distance interval, without problems.
In other to further make the decision as to whether trapping has occurred, provision is preferably additionally made for an upper load threshold value to be defined, this threshold having to be exceeded in order to conclude that trapping has occurred.
According to a preferred refinement, a nominal load which represents the total friction of the adjustment system is determined for the purpose of determining and defining the lower load threshold value which has to be exceeded in order to even begin the computational check as to whether trapping has occurred. In this case, the load threshold value is defined as a characteristic deviation of the currently detected total load from the nominal load. In order to determine the nominal load, the following process is followed in this case in particular: during a start phase in each case at the beginning of an actuation operation of the adjusting device, the total load detected for this time point is determined and stored as a nominal load. In this case, the load is, in particular, the motor moment, the force exerted by the motor or else a variable which is correlated with this, for example the detected and, in particular, averaged motor rotation speed or the detected motor current.
According to the invention, the object is also achieved by a method having the features of patent claim 18. The advantages and preferred refinements given with regard to the trapping prevention means can therefore correspondingly also be transferred to the method.
Some of the individual features and combinations of features in the patent claims, possibly with the addition of further features or combinations of features from the description, are also independent of the features of independent patent claims 1 and 2. We reserve the right to submit partial applications which do not or do not fully contain the features of claim 1 or 2.
Exemplary embodiments of the invention are explained in greater detail below with reference to the figures, in which:
The method for reliable detection of a trapping instance explained below with reference to the figures applies in particular to use in a motor-driven seat adjusting means in the motor vehicle sector. A device of this type has an adjusting mechanism which comprises a seat support which can usually be longitudinally adjusted in guide rails which are slightly inclined with respect to the horizontal. A backrest whose inclination can be adjusted is also attached to the seat support. In this case, the rotation point of the backrest is arranged such that it is somewhat spaced apart from the guide rails. Furthermore, the adjusting device comprises a respective drive motor both for translatory adjustment in the longitudinal direction of the seat support and for inclination adjustment of the backrest. These motors are usually a DC motor or a rotation speed-controlled DC motor.
When seats are automatically adjusted, there is a risk of a person being trapped in the seat to be adjusted or else between the seat to be adjusted and a back seat. A trapping instance of this kind leads to a high motor torque and therefore correlates to a higher force expended by the motor. This total torque generated by the motor is also generally called the total load in the present case. Identification of a trapping instance is problematical particularly in the case of seat adjustment of this type since the force to be additionally applied by the motor does not necessarily exhibit an abrupt increase in the event of trapping on account of the soft seat cushion.
The method described below is suitable, in particular, for a seat adjusting means, but can, in principle, be applied to other adjusting devices, for example window winders, sliding doors, trunk lids, sliding roofs, etc. too.
The computational and mathematical treatment of an adjusting device of this kind with the aid of a control device is explained in greater detail below with reference to
In the event of trapping, a spring model is assumed in order to physically and mathematically describe in a simple model the real processes when a person is trapped between the seat and a further seat or the dashboard. In the physical model shown in
Taking this physical model as a basis, the following equation 1 is given for the motor voltage u:
u=R·i+Ldi/dt+u
ind Equation 1
This can be differentiated to give the equation 1′ for the variable di/dt:
di/dt=1/L(u−R·i−K1n) Equation 1′
with the following relationship, according to which the induced voltage uind is proportional to the rotation speed n and the proportionality factor is K1, having been taken into account here:
uind=K1n Equation 2
Furthermore, the motor moment MMot is proportional to the motor current i multiplied by a proportionality constant K2:
MMot=K2i Equation 3
For the right-hand side of the physical model according to
M
Mot
−M
L
=Jdn/dt Equation 4
The moment of inertia J is actually made up of several components, in particular the moment of inertia of the motor and that of the mechanical parts of the seat. Since very large transmission ratios are generally provided for motorized seat adjusting means, the proportion of the total moment of inertia of the mechanical parts can be ignored and it is sufficient to take into account the moment of inertia of the motor for the calculation. The following equation, according to which the trapping moment ME is proportional to the spring force FF, with the proportionality factor K3 being a weighting parameter which takes into account the geometry of the adjusting mechanism, can be derived from the spring model for the trapping moment ME. In this case, the weighting parameter takes into account, for example, the lever length, the lever transmission ratio or the position of the adjusting mechanism. Information about the areas of risk, that is to say, for example, the distances between the seats which, in particular, are also dependent on the body size, are additionally incorporated in the weighting parameter. The spring force FF is in turn proportional to the rotation angle φ−φK covered, with the proportionality factor being the spring constant c. In this case, φK is the rotation angle at the time point at the beginning of the trapping instance, that is to say when contact is made for the first time between the seat to be adjusted and the trapped person.
M
E
=K
3
F
F
=K
3
c(φ−φK) Equation 5
A mathematical model or a corresponding calculation algorithm, which can be represented by the control loop illustrated in
By taking into account the supplementary spring model, a second mathematical model can be derived, with the aid of which the actual situation can be checked for the presence of a trapping instance. This second model can be represented by a control loop according to
The rotation angle φ is given by integration of the rotation speed n. The trapping moment ME is built up on account of the spring constant c. The load moment ML determined last by means of the first mathematical model according to
The variables L, R and K1 and K2 are motor-specific characteristic variables which are known when using a specific type of motor or at least can be determined by experiments. The moment of inertia J and the constant K3 are variables which characterize the adjusting mechanism or the interaction of the motor with the adjusting mechanism, which variables can be and also are likewise determined, in particular, by experiments on reference models. In this case, the constant K3 is determined separately for each type of adjusting device. In this case, the values of the parameter K3 are measured and stored, particularly with the aid of measurements on an actual model of the adjusting device. It should be noted here that, in particular, the weighting parameter K3 which represents the mechanism of the seat adjusting means is dependent on other variables, for example angle of inclination of the backrest or current longitudinal position of the seat. Therefore, a table of values or a characteristic map for the parameter K3 is created overall and stored in a memory of the control device. The respectively valid parameter values are then taken from this table of values or characteristic map in each case depending on the current position of the seat, and adopted in the calculation for the first or second model. In this case, the values of these parameters can also be processed using fuzzy logic.
In the method, a distinction is made between a start phase I and a monitoring phase II. The start phase I is divided into two sub-phases IA and IB, with the sub-phase IA representing a start-up phase of the motor 2 during which the motor 2 is adjusted to a specific, substantially constant motor moment MMot. The motor moment MMot remains at this level if there are no frictional changes, running difficulties or trapping situations. The second sub-phase IB serves to determine a nominal moment MG. This corresponds to the motor moment MMot which is output by the motor 2 during this sub-phase IB and is also called the total moment or total load. The nominal moment MG is determined, in particular, by calculating the average value of the values for the motor moment MMot over the second sub-phase. As an alternative to this, the average value is calculated over the entire start phase I and the start-up effects are ignored.
The start phase I becomes the monitoring phase II at a time point t0. In this case, the time point t0 is formed such that the adjusting device has covered a predefined adjustment path up until this time point. The value for the nominal moment MG determined during the start phase I is first stored as a comparison value for the monitoring phase II. During the monitoring phase II, a significant or characteristic deviation is defined as a difference from the nominal moment MG and a limit value which is called lower load value M1 is stored. The profile of the motor moment MMot is now monitored in order to determine whether this lower load limit value M1 is exceeded. In this case, the averaged profile of the rotation speed n is used as a criterion for the profile of the motor moment MMot.
In this case, both the value for the nominal moment MG and, with it, the lower load value M1 are preferably adapted during the adjustment process. Different frictional values and local running difficulties usually occur, specifically over the adjustment path, so that the motor moment MMot varies and, for example, also increases continuously over a relatively long adjustment path. If the nominal moment MG were not adapted, there would be a risk of the load value M1 being exceeded, this being a triggering criterion for checking whether trapping has occurred. In this case, the nominal moment MG is adapted, for example, by moving average value calculation over a predefined time window or else by means of continued average value calculation, starting from time point to.
If the load value M1 is exceeded, this is judged to be an indication of a possible trapping instance. At this time point, a changeover is made from the first mathematical model to the second mathematical model and the spring model is now taken into consideration for the calculation. When the changeover is made to the second model, at least one variable which is still determined with the first model is adopted here as an input variable for the second model. This variable is, for example, the value for the last actual nominal moment MG, since this represents the sum of all the moments acting on the drive, excluding the trapping moment ME.
The changeover to the second mathematical model is therefore made at time point t1, at which the load value M1 is exceeded. Therefore, the monitoring phase II is also divided into two sub-phases IIA and IIB, with the first mathematical model being used for monitoring purposes during the first sub-phase IIA and the second mathematical model being used during the sub-phase IIB.
The second mathematical model is now used to check whether trapping has actually occurred. This is explained in greater detail below with reference to
When checking a seat adjusting means for a trapping instance, the profile of the motor moment MMot is examined to determine which of the following movement classes are present:
The characteristic profiles for these movement classes of the motor moment MMot are illustrated in
As can be seen from the individual curve sections in
In the spring model which forms the basis, the increase in force or motor moment MMot corresponds to the gradient or derivative, and therefore to the spring constants c, for evaluation of these different situations. Therefore, the spring constant c, which can be obtained by means of the derivative, is used as the decision criterion as the critical criterion for classifying the currently measured profile of the motor moment MMot. In addition, further decision criteria, which have to be satisfied, are provided for unambiguous association. The term “derivative” is to be understood very broadly here. It is essential for characteristic variables for the profile of the respective motor moment MMot to be determined, from which characteristic variables conclusions can be drawn as to which movement classes a) to e) are present.
In the exemplary embodiment, an average load value M2 and a maximum load value M3 are defined in addition to the lower load value M1 in order to identify the different movement classes. If the respective load value M1 to M3 is reached, the associated adjustment path x1 to x3 (or else the associated time point t) is stored and value pairs (M1, x1), (M2, x2) and (M3, x3) are respectively formed. As an alternative to this, it is also possible to predefine fixed travel points during the sub-phase IIB and to determine the respectively current motor moment MMot at these travel points.
A value for the gradient c1, c2 is then determined in each case from the value pairs, in particular by simple linear interpolation or another mathematical interpolation. This is indicated in
Some movement classes a) to e) differ additionally or sometimes only by virtue of the profile of the increase. By determining three value pairs, two intervals are used for evaluation purposes, so that it is possible to identify whether the increase in force is increasing, remaining the same or possibly even decreasing.
In addition to the decision criterion of the derivative (gradient c1, c2), a further decision criterion used is the maximum load value M3 being exceeded. Therefore, a trapping instance is identified only when the derivative moves in a predetermined value range and at the same time the maximum load value M3 is exceeded. With regard to the derivative, the decision value used is not only the absolute value but also the profile of the absolute value.
As can be seen from comparison of
The derivative is of particular importance for associating the currently measured profile with the individual movement classes a) to e). For association in terms of which value of the derivative or which profile of the derivative is to be associated with which of the movement classes a) to e), it is expedient—similarly to in the case of the weighting factor K3—to store the individual values or profiles of the derivative in a table or in a characteristic map from which association with the individual movement classes can be performed directly or with the aid of a fuzzy logic, taking into account further boundary parameters. In this case, the table or the characteristic map is preferably likewise determined in the manner of a calibration process on the basis of a specific physical model, or empirical values are employed.
| Number | Date | Country | Kind |
|---|---|---|---|
| 20 2006 002 525.1 | Feb 2006 | DE | national |
| Filing Document | Filing Date | Country | Kind | 371c Date |
|---|---|---|---|---|
| PCT/EP2007/001319 | 2/15/2007 | WO | 00 | 10/30/2008 |
