FR2865175A1 - VEHICLE STABILITY CONTROL SYSTEM USING SEVERAL PREDICTIVE ALGORITHMS AND A SELECTION PROCESS - Google Patents

Abstract

System for controlling the stability of a vehicle, the system comprising means for imparting a longitudinal force to the tire and means for calculating a slip GOpt by determining the values of the adhesion coefficient i, for each slip Gi corresponding to levels successive "i", using in parallel "n" calculation algorithms each determining a target value of said slippage thus making it possible to obtain as many target values GCn as algorithms used, the system selecting as value of the optimal slippage GOpt the best of the target values GCn by subjecting the "n" target values GCn to comparisons aimed at eliminating the least probable target values, the comparisons being made on the basis of a function f(λ) of a descriptor of the physical functioning of the rotation of the tire on the ground allowing characteristic values to be calculated. For example, an interesting characteristic value is the variation with respect to time of G.

Description

FIELD OF THE INVENTION

  The present invention relates to stability control systems of a vehicle. In a particular application, the invention relates to systems for preventing the locking of wheels during heavy braking, popularized under the designation ABS. More generally, the invention relates to all systems intended to maintain the vehicle on a stable trajectory by automatically acting on actuators such as those determining a torque to the wheels, or those determining the steering of one or more wheels, or even those concerning the suspension, which is known to have an effect on the control of the trajectory (for example active antiroll). In the particular application mentioned above, the actuators are the brakes of a wheel or the member imposing a driving torque to a wheel.

STATE OF THE ART

  As a reminder, the longitudinal adhesion coefficient g of the tire is the quotient of the longitudinal force divided by the vertical force applied, ie the load applied to the tire (in the simplified case of a pure braking force and the skilled person will easily generalize); the slip G of the tire is G = 0% when there is no slip between the speed of the tire and the speed of the vehicle, that is to say if the tire rolls freely, and G = 100% if the tire is blocked in rotation. Typically, depending on the environment (nature of the soil (asphalt, concrete), dry or wet (water height), temperature and level of wear of the tire), the value of the slip G can vary enormously ( g is about 0.15 on ice and about 1.2 on dry soil).

  It is known that the braking of a vehicle will be all the more effective that one will manage to operate the tread with a slip G corresponding to the maximum value of the coefficient of adhesion (sometimes also called coefficient of friction). The maximum value of the coefficient of adhesion is called. But the average driver is not able to dose braking to meet this condition.

  P10-1595-EN This is the reason why a vehicle stability control system has been developed which automatically modulates the braking force so as to aim at a predetermined slip target, which is supposed to correspond to the maximum of the coefficient of friction. adhesion.

  In particular, the patent application EP 1371534 published on December 17, 2003 proposes a sliding control method using a quantity called Invariant that the inventors' researches have made it possible to discover, this quantity being thus called because it is substantially constant whatever the tire under consideration and whatever the grip of the ground on which the tire rolls.

  If this method makes it possible to determine a sliding target for which one is actually much closer to the real maximum grip coefficient of the tire in the actual driving circumstances, there are however cases where it is possible to determine a better target for improve braking efficiency (or acceleration).

BRIEF DESCRIPTION OF THE INVENTION

  The invention proposes a system for controlling the stability of a vehicle using a plurality of predictive algorithms and a selection process.

  In a general formulation, the invention proposes a system for controlling the stability of a vehicle in which a characteristic parameter Q of the operation of a tire of the vehicle intended to roll on a ground varies according to a parameter P according to a certain law, an optimum value of said parameter P being imposed by a controller directly or indirectly so as to act on at least one of the elements selected from the group comprising the torque applied to the tire, the steering angle of the pneumatic, the camber angle of the tire and the vertical force applied to the tire, wherein the controller includes means for: P10-1595-EN É Determine estimates or measurements (P ;, Q;) for successive levels i values; E use in parallel n computation algorithms each determining a target value of said parameter and thus obtaining as many target values Pc "as algorithms used, E selecting as optimal value of the parameter the best of the target values Pc" by subjecting the n target values Pc "to comparisons aimed at eliminating the least likely target values.

  The desired result is to maintain the value of a parameter Q at a value chosen as being ideal in the situation of the vehicle of the moment. In the present specification, there is described in detail an application to the control of the sliding of a tire, typically during a braking maneuver or during a maneuver acting on the yaw of a vehicle (function known as ABS in the first case or function known as ESP in the second case). Finally, we mention an application to control the trajectory using actuators other than those acting on the torque wheels.

  In a first application, the invention therefore proposes a system for controlling the stability of a vehicle in which the parameter P is the slip G of the tire and the characteristic parameter Q is the adhesion coefficient of the tire, the system comprising means for printing to the tire a longitudinal force, means for modulating the longitudinal force and means for calculating the slip parameter during each activation of the means for printing the tire a longitudinal force as follows: É determine the values of the coefficient adhesion for each slip G; Corresponding to successive levels i of the longitudinal force, E to use in parallel n calculation algorithms each determining a target value of said slip and thus to obtain as many target values Gc "as algorithms used, E select as optimal slip value G pt the best of the target values Gc "by subjecting the n target values Gc" to comparisons aimed at eliminating the least likely target values.

  The present invention therefore proposes, as a function of observations made at the very beginning of each torque variation maneuver (typically at the very beginning of each braking maneuver, but this is also valid at the beginning of each acceleration or for other maneuvers such as a steering), from the observations (measurements or estimates), calculate the target slip values according to several algorithms and select among these targets the optimum slip value finally retained by the slip control. The desired result is therefore to maintain the value of the coefficient of adhesion of the tire to a value chosen as being ideal in the situation of the vehicle of the moment, for example to the maximum possible value.

  The application of choice of the present invention lies in controlling the sliding of a wheel under braking. All the following description concerns in this case a modulation device of the longitudinal force acting on the brake control. Let us indicate once and for all that in this case, we initialize (i = 0) the operations indicated above, and in more detail below, at the beginning of each braking maneuver. But if it is decided to apply the present invention to control the sliding of an accelerating wheel, the longitudinal force modulation device acts on the engine torque to the wheels and is initialized (i = 0) the indicated operations at each request for variation of the engine torque greater than a predetermined torque threshold.

  Note also that, in the context of the present invention, it is of little importance that the tread which is treated the adhesion characteristic is that of a tire or a non-pneumatic elastic bandage or a caterpillar. The terms tread, tire or tire, bandage, elastic bandage, caterpillar or even wheel should be interpreted as equivalent. Note also that the determination of the values of the coefficient of adhesion for each slip G; can be done by direct measurement or estimation from other measurements or estimation of other quantities such as the effort in the ground plane and the vertical load.

  The selection of the optimal slip is based on a comparative evaluation between all the algorithms. For example, for each algorithm from 1 to n, after acquisition of sufficiently Gi values, a comparative balance is made by calculating the absolute value of the relative difference between the target Gcn of the algorithm n and the target Gcsel # "- 'of the previous selection n - 1. Using a characteristic value (X) of the physical operation of the rotation of the tire on the ground, a tolerance threshold on the difference calculated previously is chosen and on the basis of this characteristic value, the selection is as follows: if the deviation is less than the tolerance threshold, select Gcn as optimal slip value Gop, if the deviation is greater than the tolerance threshold, select Gcse '# n- 'as the value of the optimal slip Gopt For the first choice, a preferred algorithm will have been experimentally determined, for example the algorithm statistically giving the target closest to the ideal target. tions gives a first selection for the target, ie a first reference value, hence the target designation Gcseu '' '.

  We can take as characteristic value X the variation with respect to the time of G because it describes quite well the type of soil on which one is; it is known that on a weakly adhering ground the wheel goes very quickly to blocking so the variation with respect to time will be great and vice versa.

  The tolerance threshold is calculated by Boolean logic or fuzzy logic or by any mathematical function.

  The remainder of the description illustrates the invention used to select, according to the circumstances, the most relevant of three algorithms: a first algorithm incorporating the invariant principle illustrated in the aforementioned patent application for all that concerns the determination of the target slip using this principle, which will be referred to herein as the Invariant algorithm convenience, also referred to as Invt; A second algorithm, new in itself, which will be referred to herein as the Medium algorithm, also referred to as Avg; P 10-1595-EN a third algorithm, new in itself, which will be referred to herein as Wet algorithm convenience, also referred to as Wet.

BRIEF DESCRIPTION OF THE FIGURES

  The following figures show: FIG. 1 is a block diagram of the invention; Figure 2 illustrates a particular treatment of early acquisitions of measurements or estimates; Figure 3 illustrates in greater detail said particular treatment; Figure 4 illustrates another particular treatment of measurement or estimation acquisitions; Figure 5 illustrates the Invariant algorithm; Figure 6 illustrates the Medium algorithm; Figure 7 illustrates the Wet algorithm; Figure 8 is a block diagram illustrating an implementation of the invention.

  DESCRIPTION OF THE BEST MODE FOR CARRYING OUT THE INVENTION Reminders on the Invariant Algorithm There are only brief reminders concerning the Invariant algorithm. In particular, see Example 1 described in the aforementioned patent application, relating to the automatic maintenance of the operation of a tire on a vehicle at a certain coefficient of adhesion, for example the maximum coefficient of adhesion The Invariant algorithm allows in particular to slaved the slip G so as to maintain an optimum value GOPt predetermined. Said predetermined slip G pt is more particularly and in a nonlimiting manner, chosen so that the predetermined value of the adhesion coefficient corresponds substantially to the max value. In this case, it is agreed to call this particular optimum value Gmax. It is observed that the position of the maximum for this tire varies according to the ground. There is an invariant Invt common to these different soils.

  By applying the Invariant determination equation to the slip coefficient analysis, the Invariant is calculated in particular as follows: (Gmax) Invt = G /, with p having a positive value and less than 1. G (max PEG) With regard to p, its value is preferably between 0.25 and 0. 75, for example it is 0.5, the reader being referred to the description of the request for aforementioned patent as to the consequences of choosing the value of p. The following is obtained: G (G = Gmax) 0.58 (G = G max / 2) An Invariant Invt was introduced by processing the data at a first slip value G, corresponding to the maximum of the maximum adhesion coefficient, and at a second value corresponding to 50% of the first value, for example 50% of the previously mentioned slip. We can thus deduce a first value of Invt.

  It is then necessary to proceed to the determination of the slope ai of the curve; = f (G,) illustrated in FIG. 5. This is done by direct calculation a ,, - i / Gi or a regression is used. adapted, for example a linear regression. For example, two coefficients P10-1595-FR are calculated.

G

  particulars AL 'and BL'n by the following linear regression, applied to n measurement or estimation points: AL; n = nEgeA EGeAe BL; n = EaEeG2 EGeEEeG n.EG2- (EG) 2' n.EG2- ( EG) 2 which gives the expression of the slope a = AL'nE G + BL'n Then, the computation of G0 "i is done as follows: 1 pEInvt Another well-adapted regression is an exponential regression.

  Thus, a first algorithm called algorithm Invariant, performs the following operations: É Determine the slope a; of the line passing through the origin and by (G ;, É Calculate AP coefficients by direct calculation or by appropriate regression from a sufficient number of pairs (a ;, G;) in order to model a curve of variation 15 a; = f (G ;, AP), É Calculate a first target slip Gc '' using a predetermined invertor "Invt".

  We have seen that Invariant Invt is worth substantially 0.58. But, in practice, Invariant Invt can be used as a debugging parameter of the system.

  Description of the Middle Algorithm

  The Mean algorithm consists in detecting the change in the curvature of the curve (G). It is understood that such a change is indicative of the proximity of the maximum adhesion. For this, we treat the measurements or the estimates of slip values and coefficient of adhesion associated with two regressions, one of which aims to model a curve of variation which passes necessarily by the origin and whose other aims at modeling. a P10-1595-EN GoPt - - BLin 1 Invt A Lin curve of variation which does not necessarily pass by the origin, ie whose first point is left free.

  Thus, a second algorithm performs the following operations: É Determine coefficients A [a "g / Pi by direct computation or by appropriate regression, from a sufficient number of pairs (;, G;), so as to model a first curve of variation = f (G ;, A [avg / pl) necessarily including by convention the origin, and the pair or pairs (;, G;), in which; is not equal to zero, É Determine an indicator of the average slope al of said first variation curve, É Determining coefficients B [a "g / P] by direct calculation or by appropriate regression, from a sufficient number of pairs (;, G;), so model a second variation curve = f (G ,, B [a "gf pi) including the pair (s) (;, G;), in which; is nonzero, É Determine an indicator of the mean slope a2 of the second variation curve, É As long as the difference between ai and a2 is less than a predefined slope threshold If the difference between al and a2 exceeds the predetermined slope threshold, determine a target slip Gia "g using at least the same value as at least one pair of values (G, pt,). last pair of values (G ;,;). Preferably, a second condition causing the determination of the second target slip is added using at least the last pair of values (G ;,) .i .; the second condition being as soon as G, exceeds a predetermined threshold, by example 15%.

  It is particularly appropriate to use linear regressions. In this case, we look at the difference between two linear regressions, as shown in Figure 6. In this case, the first variation curve, represented by a dashed line in Figure 6, is a first line = Aa "g G ;, including by convention the origin (to a near correction Go, correction which will be explained later) and the or the pairs (;, G;), obtained by a first linear regression calculating a first coefficient Aa "g, P10 -1595-EN É The second curve of variation, represented by the center line in Figure 6 (less vertical curve), is a second line; = A; ,, É G; + am including the pair or pairs (;, G;), which is left free from the origin, and is obtained by a second linear regression calculating coefficients A,; ,, and B1; n ( which does not exclude a Go correction which will be explained later).

  Of course, in the case of linear regressions, the indicators al and a2 are directly the respective slopes Aa "g and A ,, ,, of each of the straight lines.

  It has been determined experimentally that, in the case of linear modeling (linear regressions), the predetermined slope threshold for the difference between a1 and a2 is preferably of the order of 30%.

  In the end, the target slip Gca "g retained can be simply equal to the last value G. As an alternative and in a more precise way, the target slip is determined.

MAX

  Gcavg =. AVC with f3 a focus parameter. When we speak in the present memory of a debugging parameter, it means that, even if there exists for this parameter a value or a range of values representative of a physical reality, it is possible in practice arbitrarily to use this parameter as a fine-tuning button of the practical operation of the stability control system of a vehicle, ie a parameter used in the development of the system. Just add that the parameter (3 has an analogy with the Invariant, which translates for f3 by a value of about 1.04 to bring closer to the value 0.58 for the Invariant.

  As we will see in the following, without this being limiting, it is proposed that the algorithm Average is chosen as the reference value for the determination of the target slip because it has been observed experimentally that it is the one that is the most robust for all types of tires and all types of floors. It is therefore this algorithm that serves to validate target choices based on the other algorithms. An example of a choice process between the algorithms is described at the end of this memo.

P10-1595-EN

  -11-Description of the Wet Algorithm

  As the name implies, this algorithm is specifically designed for wet target calculation (in fact all low friction soils). Indeed, on this type of soil, the wheel goes very quickly to the blocking and one risks to realize too late that the maximum of the curve (G) is exceeded.

  The principle of this algorithm is to study the evolution of the derivative of the slip as a function of time. If we see a change too fast, ie a change of the slip too fast, we can assume that the system is no longer stable because we have passed or we are in the process of exceed the slip corresponding to the maximum adhesion. This algorithm therefore consists of: É As the values of G; are calculated, calculate the variation with respect to the time of G, É As long as the variation is greater than a low threshold, calculate coefficients A [wet , p by direct calculation or by a suitable regression so as to model the variation with respect to the time of G by a curve of variation which is a function of (G, A [wet, p]), É As soon as said variation is greater at a high threshold, determine a third GCWet target slip using at least the last values of A [wet, p]. Preferably, said algorithm is used only with slip values greater than 4%. Advantageously, said low threshold is of the order of 1% per second and said high threshold is of the order of 3% per second.

  A simple application of this algorithm uses a linear regression, so that we calculate the Awet and Bwet coefficients as follows: E dG _AWEr.G + BWET dt and the third target slip GCwet is thus determined as follows : P10-1595-EN - 12 - G cavet = dG _ tgt _ BWET

awet

  Although an experimental value of 2% per second for the dG_tgt parameter has given good results as shown in FIG. 7, in practice this parameter can be arbitrarily used as a fine-tuning knob for the practical operation of the control system of the stability of a vehicle, as evoked about the parameters f3 and Invariant.

  Suggestions for Improvements Regarding Initial Data Acquisition Before proceeding, some corrections to the low slip values and coefficient of adhesion obtained at the very beginning of a braking maneuver are proposed. At the beginning of the braking, it was found that the curve g (G) can present a strange behavior. The purpose of this first part of the algorithm is to correct this behavior. In Figures 2 and 3, it can be seen that on a wet ground, the foot of the curve g (G) does not appear linear and that for a zero g, the slip is not. This is probably due to errors in the value of the measured slip. This is obviously not representative of the physical phenomena in the contact of the tread on the ground. This is troublesome for the target slip computation algorithms that are based on the study of the slope of the curve g (G). Of course, this is very dependent on the concrete technological means implemented on the vehicle to acquire this information. Therefore, the indications given in this context are simple suggestions that it is useful to apply if we face this problem, but which are not limiting. More generally, it is useful to authoritatively correct the shape of the foot of the curve of variation of the coefficient of adhesion depending on the slip if it has a strong pace implausible.

  The first part of the data processing will therefore consist in calculating the value of the slip from which the data can be used to reliably calculate a target slip or optimal slip. Let's call this slip Go. Figure 2 P10-1595-EN 13 - shows that this slip Go is worth about 3%. A more plausible course of the curve (G) is obtained by connecting Go to the substantially linear part of the increasing portion of the curve.

  Therefore, preferably, the stability control system of a vehicle is such that one proceeds, before all operations using the curve of variation of; as a function of G ;, to a correction of the departure of said curve by eliminating the first real pairs (;, G;) as long as the variation as a function of G, is not substantially constant and looking for the slip associated with a coefficient of zero adhesion (this being of course not limiting) such that the pair (0, Go) and the non-eliminated pairs (;, G;) are substantially aligned, and using a curve starting from (0, Go ) and joining the non-eliminated pairs (;, G), so that for any value of G, larger than G, G; is replaced by G; Go.

  For this, for example, an algorithm is used which comprises the following steps: E systematically eliminate all slip values associated with an adhesion coefficient of less than 0.01; É Calculate continuously regressions of t and G as a function of time, preferably exponential regressions given the shape of the foot of the curve in the example illustrated by means of Figure 2 and Figure 3: p = eAP (t TStart) + Bp G = e AG É (t TStart) + BG The acquired values can be considered as representative of reality when the estimated or measured coefficient of adhesion is greater than 0.1 or when the slip exceeds 4%.

  Figure 3 illustrates how to determine Go from the curves giving the acquired values respectively for the coefficient of adhesion as a function of time and for the slip as a function of time. We look for the value of the time for which the regression on the curve is worth a certain value, for example 0.05 (see the horizontal segment between a null abscissa and an ordinate of 0.05 and the dotted curve). The value of Go will be the value of the regression on the slip curve at this time (see the vertical segment between the previously obtained point P10-1595-EN-14) and a point on the continuous line curve, giving the value of the Go slip).

  Therefore, before all the operations using the curve of variation of i as a function of Gi, a correction of the start of said curve is carried out by eliminating the first real pairs (lai, Gi) as long as the variation of the function of Gi n is not substantially constant and looking for the slip Go associated with a coefficient of zero adhesion such that the pair (0, Go) and the non-eliminated pairs (;, Gi) are substantially aligned, and using a curve starting from (0 , Go) and joining the non eliminated pairs (i, Gi). Then, in all the algorithms used, for any value of Gi greater than GB, Gi is replaced by Gi Go.

  Up to now, it has been assumed that values of p are calculated or estimated. However, in some cases, the method of obtaining the coefficient of adhesion (from the braking force itself estimated on the basis of brake pressure given the particular characteristics of each vehicle and from the wheel speed) does not give a satisfactory result (the curve (G) calculated is too flat or continuously upward). We know that this is not realistic. To correct this problem, a numerical correction of the calculated can be implemented. This correction is based on the speed of evolution of the slip as a function of time. Indeed, if the speed of the wheel (and therefore the sliding) is racing fast, it is that we are in the unstable zone of the curve (G). Therefore the curve (G) should decrease, which is used in a stability control system of a vehicle in which, when the variation with respect to the sliding time becomes greater than a predetermined threshold of variation, we proceed, before all the operations exploiting the curve of variation of pi as a function of Gi, to a correction of the end of the curve by replacing the values of; corresponding to slips bringing the variation with respect to the sliding time beyond said predetermined threshold of variation, by values corrected as follows: u, corr = pi. I Maxl G; 1 where Acorr is a focus coefficient and can be specific to each algorithm. For example, a good practical value was found to be 0.2 for the Medium algorithm. P10-1595-EN Figure 4 illustrates this correction. Thanks to this correction, the curve (G) returns to a form certainly more in conformity with the physical reality, which allows the algorithms to produce reliable targets.

  Note that, if the value of the wax is itself modified by this correction, all the algorithms used are based on the shape of the curve and not on its values. The reader is also referred to the aforementioned patent application in which it was pointed out that the Invariant algorithm makes it possible to calculate a sliding target without even calculating the associated coefficient of adhesion, the latter being useless for the proper functioning of the invention. Sliding control of a vehicle wheel.

  Selection between the target slides given by the different algorithms Different targets were calculated thanks to the presented algorithms (Invariant, Medium, Wet). The general principle of selection is illustrated in Figure 1. It is seen that one exploits measures or estimates, say in general acquisitions G slip; and the slip coefficient i associated with each i of the slip values. A target value G01 of the slip is calculated in parallel by means of n algorithms Alg i. Finally, one chooses among these target values by making comparisons using a function f (,) to find one or more characteristic values of the physical operation of the rotation of the tire on the ground.

  We will now describe a practical, nonlimiting case of implementation of the choice of the final value, ie the optimal slip used for the regulation of the sliding of a wheel.

  Figure 8 illustrates the operations. We retain the Middle algorithm here as a reference that serves to judge others. The principle of the algorithm of choice between the different targets is based on the fact that one of the targets has a reference value in which one has good confidence whatever the type of soil. The choice will be made from the difference with this value of P10-1595-FR reference and with the help of the derivative of the slip as a function of time (dG) to characterize the types of soils.

  The selection of the optimal slip, among the target values given by the Wet and Mean algorithms, is effected for example as follows: É Calculate the relative difference in absolute value GE between Gcavg and GcWet É Preselect a value Gcsen of the way following: as long as said variation with respect to the time of G is lower than a low limit, to retain as a value Gcsel # i the value of Gcavg, when said variation with respect to the time of G is greater than the low limit, a first critical threshold AGmaX whose value depends on said variation with respect to the time of G and:> if GE is smaller than the first critical threshold AG ", retain as value Gcsel # i the value of Gcavg - * if GE is greater than the first critical threshold AGmax, hold as value Gcsel # i the value of Gc1Net Take as the final selection of the value of the optimal slip G Pt the value Gcsel # i The critical threshold is preferably chosen by a process s of fuzzy logic, to be variable and better adapted to real circumstances.

  As a purely illustrative example, good results have been obtained by varying the critical threshold between 0% and 2.5% for a value of the variation of the slip as a function of time varying between 3.5 (low limit) and 4% by second (upper limit), and setting the critical threshold to 2.5% for values above this upper limit.

  Then, if an invariant algorithm is used, the last step is replaced by a selection between the above preselection and the target value given by the invariant algorithm, operating as follows: P10-1595-EN É if said variation relative to the time of G is greater than a predetermined choice threshold, then the value of the optimal slip G pt is equal to Gcsel # '; If said variation with respect to the time of G is less than a predetermined choice threshold and if the difference GEZ in absolute value between and Gcse '#' is less than an optimized threshold OG pt, then the value of the optimal slip G pt is equal to Ga '; If said variation with respect to the time of G is greater than a predetermined choice threshold and if the difference GEZ in absolute value between Gcm \ and Gcse '#' is greater than an optimized threshold AG pt, then the value of the optimal slip G pt is equal to Gcse '#'.

  The thresholds of choice and optimized threshold are, in turn, debugging parameters of the system. By way of illustration only, good results have been obtained by setting the choice threshold at 3% per second and setting the optimized threshold at 5%.

  We have just seen that the selection process happens in two stages. chained. The first step involves the Medium algorithm and the Wet algorithm, and the second step involves the Invariant algorithm. The following is another process for the first and second steps.

  The selection of the optimum slip, among the target values given by the invariant and average algorithms, is effected for example as follows: Preselect a value Gcse '#' as follows: if said variation with respect to the time of G is less than a predetermined choice threshold, then the value Gcsel # 'is equal to Gcav; if said variation with respect to the time of G is greater than a predetermined choice threshold and if the difference GE2 in absolute value between Gci vt and Gcav is less than an optimized threshold OG pt, then the value Gcse1 # 'is equal to Ga '; if said variation with respect to the time of G is greater than a predetermined choice threshold and if the difference GE2 in absolute value between Gcmvr and Gcse '#' is greater than an optimized threshold AG pt, then the value Gcselm 'is equal to Gcavg . 30 Take Gcse '#' for final selection of optimal slip value G pt.

  P10-1595-EN Then, using the Wet algorithm, replace the last step with a selection between the above preselection and the target value given by the Wet algorithm, by doing the following: É Compute l relative difference in absolute value GE between G csel # 1 and GcWet E Select the value of G pt in the following way: as long as said variation with respect to the time of G is lower than a low limit, retain as value Gopt the value of G csel # 1, when said variation with respect to the time of G is greater than the low limit, a first critical threshold AG "is determined whose value depends on said variation with respect to the time of G and: - * if GE is less than the first critical threshold AG ", to retain as value Gopt the value of G csei # 1 -> if GE is greater than the first critical threshold AG", to retain as value Gopt the value of Gcwet Let us note that what was said during the first presentation of a selection process at p Critical thresholds, choice threshold and optimized threshold remain valid for this second presentation of a selection process.

  Application to other phenomena In the aforementioned patent application, the possibility of other applications of the Invariant algorithm has been demonstrated, for example to the analysis of the drifting thrust developed by a tire or elastic tire in a zone of operating close to the saturation of the drift thrust. This is because of the similarity of the laws of variation of these physical phenomena. In the same way, the present invention has broader applications than the only comparison between the predictions of different algorithms modeling the coefficient of adhesion as a function of the slip. To close the subject, let us simply mention (without even this addition being limiting as will have been understood) that the invention also applies to a stability control system of a vehicle for predicting the P-10. 1595-EN - 19 value of the drift angle 8 of a tire where the lateral force (also called drifting thrust) is maximum.

  In this case, the parameter P is the drift angle θ of the tire and the characteristic parameter Q is the drift thrust Fdet of the tire. This is to predict when the tire will reach its maximum and therefore will not be able to respond to its primary function of allowing the vehicle to rotate, in order to maintain the operation of the tire to a predetermined value of the drift thrust Fdet, or to warn the driver. To maintain the operation of the tire at a predetermined target value, it is possible, possibly automatically, to carry out preventive interventions for reducing the speed of the vehicle to avoid critical driving situations (if the vehicle does not rotate as desired by the driver, this may result in an accident). To proceed with these actions, it is also useful to select between several target values given by different algorithms.

  In this case, the invention relates to a system comprising means for controlling a parameter according to the orders printed by the driver of the vehicle on his control means and according to the commands issued by a trajectory controller of the modulation means of the parameter and means for calculating the drift angle parameter and at each activation of the means for printing the parameter as follows: É at each activation of the variation control system of, for at least two different levels i of drift angle , to note different values of Fy ,, and the associated angle of drift 8, obtained by estimation or direct measurement, E to use in parallel n calculation algorithms each determining a target value of said slip and thus to obtain as many target values in that of used algorithms, select as optimal value of the drift angle the best of the 8cn target values by submitting the n 8cn target values to comparisons aimed at eliminating the least likely target values.

P10-1595-EN - 20 -

Claims (31)

  1. Vehicle stability control system, in which a characteristic parameter Q of the operation of a vehicle tire intended to roll on a ground varies according to a parameter P according to a certain law, an optimum value of said parameter P being imposed by a controller directly or indirectly so as to act on at least one of the elements selected from the group comprising the torque applied to the tire, the steering angle of the tire, the camber angle of the tire and the vertical force applied to the tire, wherein the controller comprises means for: É determining estimates or measurements (P ;, Q;) for successive levels i of values; E use in parallel n computation algorithms each determining a target value of said parameter and thus obtaining as many target values Pc "as algorithms used, E selecting as optimal value of the parameter the best of the target values Pc" by subjecting the n target values Pc "to comparisons aimed at eliminating the least likely target values.   The vehicle stability control system according to claim 1, wherein the parameter P is the slip G of the tire and the characteristic parameter Q is the tire adhesion coefficient, the system comprising means for printing at the tire. tire a longitudinal force, means for modulating the longitudinal force and means for calculating the sliding parameter GOPt during each activation of the means for printing the tire a longitudinal force as follows: É determine the values of the coefficient of adhesion for each slip G; corresponding to successive levels i of the longitudinal force, E use in parallel n calculation algorithms each determining a target value of said slip and thus obtain as many target values Gc "as algorithms used, P10-1595-FR -21 - E select as optimum slip value G pt the best of the target values Gcn by subjecting the n target values Gcn to comparisons aimed at eliminating the least likely target values.   A vehicle stability control system according to claim 2, wherein the selection of the optimum slip is made by performing the following operations for each algorithm from 1 to n: E Calculate the absolute value of the relative deviation between the target Gan of the algorithm n and the previous selection Gcsei # nt É Take a characteristic value (X) of the physical operation of the rotation of the tire on the ground and use this characteristic value to calculate a tolerance threshold on the previously calculated deviation, É Proceed to the selection as follows: If the deviation is less than the tolerance threshold, select Gcn as the optimum slip value Go) t, If the deviation is greater than the tolerance threshold, select Gc Sei # nI as value of optimum slip Goet.   The vehicle stability control system according to claim 3 wherein the characteristic value X is the variation with respect to the time of G. 5. The system of claim 4 wherein the tolerance threshold is calculated by a process selected from the group containing fuzzy logic and Boolean logic and a mathematical function.   The vehicle stability control system according to one of claims 2 to 5, wherein a first algorithm uses an Invariant and performs the following operations: É Determine the slope a; from the line through the origin and by (G ;, P10-1595-EN É Calculate Ap coefficients by direct calculation or by appropriate regression from a sufficient number of pairs (a ;, G;) to model a variation curve a; = f (G ;, An), É Calculate a first target slip using a predetermined invert "Invt".   A vehicle stability control system according to claim 6, wherein the invariant is determined as follows: (G max) Invt = G /, with p having a positive value and less than I. G (max pe) A vehicle stability control system as claimed in claim 7, wherein two particular A, B and A coefficients are computed by a regression selected from the group consisting of linear regression and exponential regression.   The vehicle stability control system according to one of claims 2 to 5, wherein a second algorithm performs the following operations: Determine coefficients A [avg! P by direct calculation or by appropriate regression, from a sufficient number of pairs (;, G;), so as to model a first variation curve; = f (G ;, A [avg / pl) necessarily including by convention the origin, and the pair or pairs (;, G;), in which is different from zero, É Determine an indicator of the average slope ai of said first curve of variation, É Determine coefficients B [avg / p] by direct calculation or by an appropriate regression, starting from a sufficient number of pairs (;, G;), so as to model a second variation curve ; = f (G ;, B [avg / p]) including the one or more pairs (;, G;), in which; is different from zero, É Determine an indicator of the average slope a2 of the second curve of variation, P10-1595-EN - 23 - É As long as the difference between al and a2 is less than a predetermined slope threshold, resume the operations for each new value pair acquisition (G ,, É As soon as the difference between al and a2 exceeds the predetermined slope threshold, determining a target slip Gca "g using at least the last pair of values (G ;,; ).   A vehicle stability control system according to claim 9 wherein, as soon as the slip G exceeds a predetermined threshold, the target slip Gca "g is determined using at least the last pair of values (G; ,;).   The vehicle stability control system according to claim 9, wherein the first variation curve is a first straight line g, = Aa "gG, including by convention the origin, and the pair or pairs (;, G;), obtained by a first linear regression calculating a first coefficient Aa "g, É The second variation curve is a second straight line; = Al ,,,. G; + B11 including the pair or pairs (;, G;), obtained by a second linear regression calculating coefficients Ahn and Bi A vehicle stability control system according to claim 9 or 11, wherein the target slip GCa "g is determined to be equal to the last value G 1.   Vehicle stability control system according to claim 9 or 11, in which MAX   which determines the target slip GC g = / 3 AEAVC with f3 a focus parameter.   The vehicle stability control system of claim 13 wherein 13 is about 1.04.   P10-1595-EN The vehicle stability control system according to claim 13, wherein f3 is a parameter used in debugging the system.   16. Vehicle stability control system according to one of claims 9 to 15 wherein the predetermined slope threshold for the difference between at and a2 is of the order of 30%.   Vehicle stability control system according to one of claims 2 to 5, in which a third algorithm performs the following operations: As the values of G; As long as the variation is greater than a low threshold, calculate coefficients A [Wet, p by direct calculation or by a suitable regression so as to model the variation with respect to the time of G by a curve of variation which is a function of (Gi, A [Wet, p]), É As soon as said variation is greater than a high threshold, determine a third target slip Gc 'and using at least the last values of A [wet , p iE Vehicle stability control system according to claim 17, in which a linear regression is used and the coefficients AWet and Bwet are calculated in the following way: dG _AWET.G + BwET dt and the third target slip GcWet is determined in the following way: ÉG caver dG _tgt BWET awet   The vehicle stability control system of claim 18 wherein dG_tgt is used as a focus parameter.   P10-1595-EN - 25 - 20. Vehicle stability control system according to one of claims 2 to 19, wherein the longitudinal force modulation device acts on the brake control and is initialized (i = 0) the indicated operations at each start of the braking maneuver.   21. Vehicle stability control system according to one of claims 2 to 19, wherein the longitudinal force modulation device acts on the engine torque wheels and is initialized (i = 0) operations indicated at each request for variation of the engine torque greater than a predetermined torque threshold.   22. A vehicle stability control system according to one of claims 2 to 21, wherein is carried out, before all operations using the curve of variation of; as a function of G;, to a correction of the departure of said curve by eliminating the first real pairs G;, G;) as long as the variation of; as a function of G, is not substantially constant and looking for the slip Go associated with a coefficient of zero adhesion such that the pair (0, Go) and the non-eliminated pairs (;, G;) are substantially aligned, and using a curve starting from (0, Go) and joining the non-eliminated pairs (;, G;), so that for any value of G; bigger than Go, G; is replaced by G, Go.   Vehicle stability control system according to one of Claims 2 to 22, in which, when the variation with respect to the time of the slip becomes greater than a predetermined threshold of variation, the procedure is carried out before all the operating operations the curve of variation as a function of G ;, to a correction of the end of said curve by replacing the values of corresponding to shifts bringing the variation with respect to the sliding time beyond said predetermined threshold of variation, by values corrected as follows: corn Maxi dG Pi. [dt 1 or Acorr is a predetermined parameter.   24. The vehicle stability control system of claim 23, wherein Acorr is about 0.2.   P10-1595-EN - 26 - The vehicle stability control system of claim 23 wherein Acorr is used as a focus parameter.   A vehicle stability control system using both the algorithms of one of claims 9 to 16 and one of claims 17 to 19, wherein the selection of the optimal slip is made from the following procedure: É Calculate the relative difference in absolute value GE between Gca "g and Gcwet É Preselect a value Gcsel # i in the following way: as long as the said variation with respect to the time of G is lower than a lower limit, retain as value Gcsei # i the value of Gca "g, when said variation with respect to the time of G is greater than the low limit, a first critical threshold is determined whose value depends on said variation with respect to the time of G and: if GE is less than the first critical threshold AG ", take as a value Gcsel # i the value of Gca" g if GE is greater than the first critical threshold AG ", retain as a value Gcsel # i the value of Gcwet, É take as the final selection of the value of optimal slip G o} t the value Gcsel # i.   A vehicle stability control system using both the algorithms of one of claims 6 to 8 and one of claims 9 to 16, wherein the selection of the optimal slip is from the following procedure: É Preselect a value Gcsei # t in the following way: if said variation with respect to the time of G is lower than a predetermined choice threshold, then the value Gcsel # i is equal to Gca "g; The time ratio of G is greater than a predetermined choice threshold and if the difference GE2 in absolute value between GcrnI and Gca "g is less than an optimized threshold AG pt, then the value Gcset # i is equal to Gcm; P10-1595-EN if said variation with respect to the time of G is greater than a predetermined choice threshold and if the difference GE2 in absolute value between Gcm and Gcsel # 1 is greater than an optimized threshold AG Pt, then the value Gesel # 1 is equal to Gca g.   É Take Gcse1 for final selection of optimal slip value G Pt The vehicle stability control system of claim 27 further using the algorithm of one of claims 17 to 19, wherein the final selection is replaced by the following steps: E Calculate the relative difference in absolute value GE between G csel # 1 and Gc'et É Select the value of G Pt as follows: = as long as the said variation with respect to the time of G is lower than a low limit, to retain the value G Pt as the value of G csei # 1, when said variation with respect to the time of G is greater than the low limit, a first critical threshold AGmax is determined whose value depends on said variation with respect to the time of G and: -p if GE is less than the first critical threshold AG ", to retain as GoPt the value of G csei # 1 if GE is greater than the first critical threshold AGmax, to retain as value G Pt the value of GcWet The vehicle stability control system of claim 28 wherein the critical threshold is calculated by a process selected from the group containing the fuzzy logic and the Boolean logic and a mathematical function.   The vehicle stability control system of claim 27 wherein the optimized threshold is a focus parameter.   A vehicle stability control system according to claim 1, wherein the parameter P is the drift angle S of the tire and the characteristic parameter Q is the tire drift Fdet, the system comprising means for controlling a parameter according to the commands printed by the driver of the vehicle on his control means and according to the commands issued by a trajectory controller, means for modulating the parameter, and means for calculating the parameter angle of drift 8 Pt at each activation of the means for printing the parameter as follows: É at each activation of the variation control system of, for at least two different levels i of drift angle, record different values of Fyi, and the associated drift angle ô; obtained by estimation or direct measurement, E use in parallel n calculation algorithms each determining a target value of said slip and thus obtain as many target values Sc 'as algorithms used, E select as optimal value of the drift angle the best Sc target values by subjecting the n target values Sc to comparisons aimed at eliminating the least likely target values. P10-1595-EN