METHOD FOR MANUFACTURING A SYSTEM FOR ESTIMATING THE CLAMPING FORCE EXERCISED BY AN ELECTRIC BRAKE

Abstract

A method for manufacturing a system for estimating the clamping force exerted by an electric brake intended to be mounted on a given motor vehicle, said estimation system comprising a storage area including executable machine instructions, memory, and a processor for generating an estimated clamping force, said brake being mounted on a test vehicle or a test bench conforming to the given motor vehicle, said method comprising the steps: a) selecting characteristics of the brake and its environment from a group of characteristics of the brake and its environment, said characteristics being measured or evaluated by means equipping the brake or the vehicle, said brake also comprising a clamping force sensor, said step a) implementing a first algorithm,b) development of a function to calculate the estimated clamping force from the selected characteristics and a second artificial intelligence algorithm. [Fig. 5]

Description

Title of the invention: METHOD FOR MANUFACTURED A SYSTEM FOR ESTIMATING THE CLAMPING FORCE EXERCISED BY AN ELECTRIC BRAKE. TECHNICAL FIELD AND PRIOR ART

[0001] The present invention relates to a method for manufacturing a system for estimating a clamping force exerted by an electric brake and a braking system comprising such an estimation system.

[0002] A motor vehicle is equipped with a brake at each wheel. This can be a disc brake or a drum brake.

[0003] The brake can be a hydraulic brake or an electromechanical brake designated EMB (“Electromechanical Brake” in Anglo-Saxon terminology).

[0004] Parking brakes are increasingly electrically activated, and it is interesting to be able to produce fully electrically actuation service brakes. For example, a screw-nut system operated by an electric motor causes the brake pads to be applied against the disc in the case of a disc brake, and the brake linings against the drum in the case of a drum brake.

[0005] We wish to be able to know the clamping force exerted by each brake in order to be able to check if the expected braking level is reached and / or to control the service brake in the case of an electromechanical brake.

[0006] Furthermore, we wish to reduce or limit the number of sensors, in particular we do not wish to implement a force sensor to know the clamping force.

[0007] Currently, the rotor position in the brake is used to deduce the clamping force using predetermined curves that relate the rotor position to the clamping force. However, it appears that these curves do not always adequately reflect the actual behavior of the brake. Indeed, knowing the rotor position alone is not always sufficient to estimate the clamping force with sufficient accuracy in certain configurations, particularly for an electromechanical service brake, because a large number of parameters are involved.

[0008] Document EP 3 853 087 describes a braking system in which the braking commands were generated by a device equipped with a fuzzy logic processor and validated by a decision algorithm. Description of the invention

[0009] It is an object of the present invention to provide a method for manufacturing a system for accurately estimating the clamping force exerted by an electric brake requiring a reduced number of sensors.

[0010] The stated goal above is achieved by a method of manufacturing a system for estimating the clamping force of a brake implementing at least a first algorithm for selecting relevant characteristics for estimating the clamping force and training at least a second artificial intelligence algorithm intended to allow the estimation of the clamping force from the selected characteristics.

[0011] The development of the braking system includes a selection step among a large number of characteristics of the brake and its environment, this step including at least the classification of the characteristics by means of at least one algorithm, this classification taking into account in particular the relative importance of the different characteristics, and the choice among said classified characteristics of certain characteristics, and a learning step of an artificial intelligence algorithm in order to establish a relationship between the clamping force and the chosen characteristics.

[0012] In a particularly advantageous embodiment, the classification step and the selection step are carried out by an algorithm, possibly artificial intelligence but not necessarily.

[0013] In one example, a polynomial function is extracted from the trained artificial intelligence algorithm and the brake estimation system mounted on the production vehicle includes a standard processor applying the established function to estimate the clamping force.

[0014] In another example, the estimation system directly integrates the trained algorithm and includes a processor; in particular a graphics processor, offering greater computing power to exploit the trained algorithm.

[0015] In another example, the braking system incorporates the trained artificial intelligence algorithm and includes a graphics processor. The processor then estimates the clamping force using the trained intelligence algorithm. Furthermore, the trained artificial intelligence algorithm continues to learn during the operation of the vehicle's brakes; that is, it continues to adapt, which advantageously allows it to take into account the aging of the braking system / vehicle.

[0016] The term "artificial intelligence algorithm" means any fuzzy logic or artificial intelligence, advantageously self-learning, preferably any neural network, preferably any deep neural network. Neural networks are software and / or hardware neural networks...

[0017] Thanks to the invention, a virtual clamping force sensor is made which is capable of accurately estimating the clamping force of an electric brake with a reduced number of real sensors. The development of the virtual sensor uses a large input number of features of which only some are selected to build the trained algorithm, which allows the features to be selected in a particularly relevant way and adapted to each of the targeted configurations.

[0018] Furthermore, the fact of carrying out a preliminary step of selecting the characteristics of a given specific brake, which takes place during the calibration phase of the brake on a platform of the vehicle model intended to receive the brake, ensures a strong adaptation of the clamping force estimation function to this brake integrated into this platform and therefore an accurate estimation of the clamping force exerted by said brake, while limiting the number of input parameters and therefore the time required to carry out the learning, and in fact the adaptation time of the brake to the given platform is reduced.

[0019] In other words, the development of the braking system implements at least a first algorithm in the parameter selection phase from among a large number of parameters and at least a second artificial intelligence algorithm enabling the system to construct a function to estimate the clamping force from the selected characteristics, which makes it possible to increase the robustness and accuracy of the system.

[0020] The present invention relates to a method for manufacturing a system for estimating the clamping force exerted by an electric brake intended to be mounted on a given motor vehicle, said estimation system comprising a storage area including executable machine instructions, a memory and a processor for generating an estimated clamping force, said brake being mounted on a test vehicle or a test bench conforming to the given motor vehicle, said method comprising the steps: a. selection of brake and environment characteristics from a group of brake and environment characteristics, said characteristics being measured or evaluated by means equipping the brake or said vehicle, said brake also comprising a clamping force sensor, said step a) implementing at least a first algorithm, b. development of a calculation function for the clamping force estimated from the selected characteristics and at least a second algorithm which is of an artificial intelligence type.

[0021] In an advantageous example, the selection step a) comprises a sub-step of classifying the features and a sub-step of choosing among said features so as to select the selected features, at least said classification step being carried out by an artificial intelligence algorithm.

[0022] For example, in step b), the second artificial intelligence algorithm is trained from the selected features and the measured clamping forces, so as to obtain a trained artificial intelligence algorithm considered satisfactory.

[0023] In one embodiment, during step b) a polynomial function is extracted from the second trained artificial intelligence algorithm considered satisfactory intended for use by the processor of the estimation system to calculate the estimated clamping force.

[0024] In another embodiment, in step b) the calculation function is the second trained artificial intelligence algorithm considered satisfactory.

[0025] For example, at least the second artificial intelligence algorithm is implemented by a neural network.

[0026] In an advantageous example, step b), and possibly also step a), is (are) repeated at several wear levels up to a wear level considered sufficient.

[0027] The present invention also relates to an electric brake for a motor vehicle comprising means for applying a clamping force to an element attached to a wheel of the motor vehicle, and a system for estimating the clamping force of an electric brake obtained by the process according to the invention and means for collecting selected characteristics, in particular without using a sensor for direct measurement of said clamping force, said collection means comprising sensors and / or means for evaluating all or part of the selected characteristics.

[0028] In one embodiment, said polynomial function being stored in the storage area.

[0029] In another embodiment, the trained artificial intelligence algorithm is stored in the storage area, the processor being a graphics processor.

[0030] The electric brake may then include means for the continuous learning of the trained artificial intelligence algorithm, said means ensuring at least the detection of a drift in the trained artificial intelligence algorithm. The continuous learning means include, for example, a backpropagation loop configured to continue training the trained artificial intelligence algorithm during the use of the brake on the motor vehicle.

[0031] Another object of the present invention is a method for learning the electric brake according to the invention, the electric brake being mounted on the given motor vehicle, said vehicle comprising means for measuring or estimating the deceleration of said vehicle, said method comprising: - comparison of the estimated clamping force to the measured or estimated deceleration, - monitoring the variation of deceleration over time as a function of clamping force, - verification of the appearance of a drift in the estimated clamping force.

[0032] Advantageously, when the occurrence of a drift is detected, said method includes a learning step for the trained artificial intelligence algorithm taking into account the drift.

[0033] Preferably, the learning process takes place during the entire use of the brake. BRIEF DESCRIPTION OF THE FIGURES

[0034] The following description will be better understood with the aid of the attached drawings, in which: [Fig. 1] is a schematic representation of a vehicle comprising a braking system capable of implementing the invention, [Fig.2] is a schematic representation of an example of a system according to the invention for estimating the clamping force applied by an electric brake intended to be mounted on a production vehicle, [Fig.3] is a schematic representation of the step of selecting the selected characteristics of the manufacturing process according to an example of the invention, this step taking place with the brake mounted, for example, on a test vehicle. [Fig.4A] is an example of a classification graph of the characteristics of the brake and its environment, [Fig.4B] is another example of a classification graph of the characteristics of the brake and its environment, [Fig. 5] is a schematic representation of an example of a learning process for the second artificial intelligence algorithm of the manufacturing process according to an example of the invention, this step taking place with the brake mounted, for example, on a test vehicle. [Fig.6] is a schematic representation of an example of generating control signals for an electric brake mounted on a production vehicle implementing the present invention, [Fig.7] is a schematic representation of an example of continuous learning of the clamping force estimation system of a braking system according to the invention mounted on the series vehicle. DETAILED DESCRIPTION OF PRODUCTION METHODS

[0035] In [Fig.1], we can see a vehicle V, represented schematically, comprising a braking system S including electric brakes FRI, FR2, FR3, FR4 each equipping one wheel.

[0036] In this example, service braking is provided by electric brakes FRI, FR2, FR3, FR4. These advantageously incorporate an electric parking brake.

[0037] Each electric brake comprises an actuator equipped with an electric motor and means for converting the rotational motion of the electric motor into a translational motion applying the brake pads against the brake disc or the brake shoes against the drum. The electric brake can be equipped with any type of electric motor, for example, a DC motor, such as a brushless DC motor.

[0038] The braking system may or may not include an ABS and / or ESP control unit.

[0039] The braking system includes an electronic control unit (ECU) (Electronic control unit in Anglo-Saxon terminology) for controlling brakes FRI, FR2, FR3, FR4. Brake control is achieved, for example, by pressing a brake pedal. Alternatively, brake activation is controlled automatically, particularly in the case of an autonomous vehicle.

[0040] The invention will be described in the context of an application to one of the electric brakes, but it will be understood that the invention applies to each electric brake.

[0041] Figure 2 schematically represents an example of a system SE for estimating the clamping force applied by a brake implemented according to the present invention comprising a processor P, a memory area M, and a storage area Z in which executable machine instructions I and a function for calculating the clamping force are loaded, the calculation function being obtained from a trained artificial intelligence algorithm AIA.

[0042] The term "artificial intelligence algorithm" means machine learning methods, such as, for example, Bayesian ARD regression, Bayesian ridge regression, Ordinary least squares Linear Regression, Linear Model trained with L1 prior as regularizer (also designated Lass), XGBoost, ElasticNet, KNeighborsRegressor, PLSRegression, ExtraTreesRegressor, AdaBoostRegressor, Lightgbm, or artificial neural networks, for example, a convolutional neural network.

[0043] An artificial neural network comprises several interconnected units between which signals are transmitted and which are arranged in layers. Each neuron receives an input signal which is modified by weights and biases, and generates an output signal which is transmitted to the other neurons via connections or synapses.

[0044] The manufacturing steps of a clamping force estimation system according to an example embodiment will now be described.

[0045] Typically, the system for estimating the clamping force of a brake depends on the brake and the model of vehicle on which the brake is mounted.

[0046] To develop the estimation system, the brake is mounted on a test vehicle or platform, corresponding to the production vehicle on which the brake is intended to be mounted, and the estimation system is built taking this brake environment into account. The brake and the vehicle are instrumented using real sensors in order to quantify and / or qualify the various parameters. The quantification of a parameter is, for example, a measurement, a calculation, or an evaluation, for example, by a mathematical model, by mapping, or even by an artificial intelligence algorithm.

[0047] Alternatively or in addition, the brake is mounted on a test bench which corresponds to the production vehicle model on which the brake is intended to be mounted.

[0048] The embodiment method includes the calibration phase of the brake on the platform. The term calibration is used here to designate the adaptation of the clamping force estimation system, which may form a subset of a larger calibration process. In this sense, this calibration phase comprises: a) The selection of characteristics from a group GP of a large number of characteristics Cl, C2, ..., Cn of the brake and its environment, which will depend, among other things, on the platform; b) The development of a calculation function for an estimated value of the clamping force from the selected characteristics, this function being designed from the selected characteristics and an artificial intelligence algorithm.

[0049] Very advantageously, the calibration phase extends over a sufficient period of time to take into account substantial, or even total, wear of the brake linings, which offers the advantage of calibrating the brake over a sufficiently long period and taking into account different states of the brake.

[0050] According to the invention, the design of the system for estimating the clamping force exerted by a given brake on a given platform is tailor-made.

[0051] The term “brake environment characteristics” means characteristics of the vehicle and / or characteristics of the terrain on which the vehicle is moving, the weather... Furthermore, the terms “characteristic” and “parameter” are used interchangeably.

[0052] The characteristics Cl, C2... Cn are, for example and without limitation: vehicle acceleration, vehicle deceleration, brake motor rotor position or piston position, brake pad temperature, brake disc temperature, brake pad wear that can be measured or evaluated, the clamping force applied to the brake pedal, vehicle speed, speed during brake activation and deactivation, braking duration, known vehicle mass, time interval between two braking operations, characteristics of the last braking operation(s), number of braking operations, road condition, road gradient, brake pad thickness, vehicle type, number brake pad type, friction coefficient, brake pad surface area, wear coefficient, weather conditions such as humidity, ambient temperature...

[0053] The test vehicle travels, for example, on a test track. During the calibration phase, measurements of the clamping force are taken using a sensor on the brake mounted on the test vehicle; these values ​​are related to the characteristics Cl, C2, ..., Cn, some of whose values ​​are measured using real sensors, others evaluated using a mathematical model, or mapped in a table, or even evaluated using an artificial intelligence algorithm forming virtual sensors.

[0054] Measurements or evaluations of characteristics take place for example every 50 ms, as well as the measurement of clamping force.

[0055] Thus, an actual clamping force value is associated with a set of characteristic values. We then have a set of actual clamping force values ​​associated with the values ​​of a number of characteristics that have a greater or lesser influence on the clamping force.

[0056] The selection step is shown schematically in [Fig.3].

[0057] The selection step comprises:

[0058] - a first sub-step of classifying the characteristics Cl, C2,... Cn, - a second sub-step of choosing characteristics from among the classified characteristics.

[0059] The first classification substep is performed using an algorithm chosen from several algorithms, including artificial intelligence algorithms adapted to classify features according to their importance, such as, for example, XGBRegressor, ExtraTreeRegressor, SHAP, DecisionTreeRegress, GradientBoostingRegressor, and RandomForestRegressor, and classical algorithms. During the first substep, the algorithm classifies the features Cl, C2, ..., Cn according to several criteria, which are, for example, their effect on the clamping value, the interdependence of the features, whether they contain enough signals or certain signals to estimate the clamping force, the degree of consistency or reliability of this feature / parameter for estimating the clamping force, and the strength of the parameter for estimating the clamping force.

[0060] Some algorithms consider parameters as independent parameters, while others consider parameters in relation to other parameters. In the latter case, the rankings change if one or more parameters are removed, whereas in the former case, the rankings remain the same if one or more parameters are removed.

[0061] The choice of the ranking algorithm is either made by a human being, or by a classical or artificial intelligence algorithm.

[0062] In [Fig.3], the features are classified into subgroups.

[0063] This classification is, for example, a division into subgroups, a classification by order of importance...

[0064] This classification is advantageously carried out by a deep neural network or Deep Learning in Anglo-Saxon terminology.

[0065] Figure 4A shows an example of a ranking obtained by the SHAP values ​​algorithm, with the group characteristics on the y-axis and an index between 0 and 3500 on the x-axis reflecting the influence of each characteristic on the clamping force. Figure 4B shows the ranking obtained by the GradientBoostingRegressor_importance algorithm, with the group characteristics on the y-axis and an index between 0.000 and 0.175 on the x-axis reflecting the importance assigned to each characteristic by the algorithm.

[0066] During the second selection substep, a choice is made among the ranked features. In one example, the choice is made by a human; in another example, the choice is made by a classical or artificial intelligence algorithm. In one embodiment, the same algorithm is used to perform the ranking and select the features from the group of features. For example, the selected features are those with the highest indices.

[0067] At the end of the selection step, the selected characteristics are obtained.

[0068] For example, the selected characteristics are Cl, C3, C5, C9, Cil, Cn. As an example, the temperature of the disc is a parameter generally retained because it has a significant impact on the variation of the brake response.

[0069] Not initially limiting the characteristics used for estimating the clamping force also allows for a selection specific to each brake and its environment, and to the platform on which the brake is mounted. Certain parameters may have a significant influence on a brake in a particular mounting configuration, and a lesser influence in another mounting configuration.

[0070] The clamping force estimation function is established on the basis of the selected characteristics.

[0071] An artificial intelligence algorithm is built from the selected features and is trained until an algorithm considered satisfactory is produced.

[0072] In one example, the untrained algorithm is an untrained DLlp neural network designed considering the selected features. For this, in particular The number of layers, activation functions and connections between neurons, weights and biases are established. Some characteristics of the untrained neural network, such as the number of layers and neurons, are established, for example, through experience and trial and error, while others are fixed at default values.

[0073] In [Fig.5], the neural network learning phase uses measured clamping force values ​​and measured, calculated or estimated Cl, C3, C5, C9, Cil, Cn characteristics on the test vehicle.

[0074] The features Cl, C3, C5, C9, Cl1, Cn serve as inputs in Anglo-Saxon terminology to the untrained neural network.

[0075] The input values ​​are modified by the weights, biases, and activation functions of the different neural layers. The network generates an estimated clamping force as output.

[0076] A backpropagation loop (BRI) is then applied, in which the estimated clamping force is compared to the actually measured clamping force; this value is called Ground Truth in Anglo-Saxon terminology. When the estimated clamping force is too different from the actually measured clamping force, particularly beyond a predetermined threshold, for example, the weights applied to the neurons are updated, as well as the biases, to reduce this difference.

[0077] The input signals are again fed into the modified network, and the new estimated clamping force is compared again with the measured value. When the estimated clamping force and the actual clamping force are within the expected error threshold, for example, between 100 N and 1000 N, the neural network is considered satisfactory. The connections and / or layers of the neural network can also be modified during the training phase. A large number of neural network modification steps can take place before obtaining the trained DL1 network.

[0078] The term "satisfactory neural network" and more generally "satisfactory artificial intelligence algorithm" means a trained artificial intelligence algorithm providing estimated clamping force values ​​within the limits of the expected error threshold.

[0079] In this example, the neural network is of the feedforward type. Alternatively, the neural network is of the backward type.

[0080] A function for calculating the clamping force, generally of the polynomial type, is extracted from the trained intelligence algorithm. This function is loaded into the storage area, typically in the form of an implementation in a program of executable instructions. This example has the advantage of requiring only a processor with standard computing power.

[0081] In another embodiment, it is the trained algorithm, for example a trained neural network or a clone thereof, which is loaded into the computer, preferably a processor offering increased power is then implemented, for example a graphics processor.

[0082] The clamping force estimation system thus developed is then installed during the production of the mass-produced vehicle. The clamping force estimation system is substantially identical to that obtained at the end of the manufacturing process learning phase.

[0083] In addition, the braking system includes means, such as sensors or evaluation means, enabling the selected characteristics for estimating the clamping force to be obtained.

[0084] For example, the vehicle has, at each brake, a wheel rotation sensor, a sensor for the supply voltage of the brake's electric motor, and a sensor for the current consumed by the electric motor. The data then provided to the neural network relates to the following characteristics: wheel speed, electrical voltage applied to the motor, and current consumed by the motor.

[0085] The characteristics that are estimated are for example the temperature of the brake disc, the wear of the brake pads, the outside temperature.

[0086] It will be understood that these lists are given only as examples, and that they may vary depending on the brake, the vehicle...

[0087] In one example, the same calculation function or trained algorithm is implemented to estimate the clamping force exerted by each brake, i.e., the front right brake, the front left brake, the rear right brake, and the rear left brake. In another example, a function or trained algorithm is established for each brake or for pairs of brakes within a vehicle type, thus providing the most accurate estimate of the clamping force.

[0088] In another particularly advantageous embodiment, the commercial vehicle includes a clamping force estimation system integrating the trained artificial intelligence algorithm DEL2 as described above, the system also including means ensuring continuous training or learning of the artificial intelligence algorithm during vehicle operation.

[0089] Figure 7 shows a diagram of a clamping force estimation system incorporating continuous learning means.

[0090] For this purpose, the continuous learning means 4 include means for comparing the estimated clamping force Fest and the obtained deceleration Dec, and for monitoring the evolution of their correlation over time. For example, if the The deceleration obtained Dec for the clamping force Fest and the deceleration obtained Deci when the brake is activated for the same estimated clamping force Fest are not within the expected error threshold (IDec - Decil). <a), une boucle de rétropropagation BR2 est mise en œuvre pour modifier l’algorithme et compenser la dérive.

[0091] The vehicle's deceleration is measured or estimated using data other than a measurement of clamping force, and the production vehicle may not include a sensor for this clamping force.

[0092] Such a continuous learning estimation system offers the advantage of taking into account changes in the brake and / or vehicle over time, resulting, for example, from their aging. Indeed, during the life of the brake and the vehicle, the model established by learning at the beginning of the brake's life may become obsolete. Thanks to the estimation system incorporating continuous learning, the trained algorithm continuously adapts to these changes to limit their effect. Such a system makes it possible to avoid using an estimate that would no longer be representative of reality; this situation is referred to as "concept drift" in Anglo-Saxon terminology.

[0093] In addition, such continuous learning makes it possible to limit the imprecision of the estimation in conditions not tested during the learning of the system.

[0094] In one embodiment, the estimation system comprises only comparison means to detect a variation in the relationship between clamping force and deceleration and to issue an alert when a deviation exceeds a given threshold. A signal is then sent to the driver, who is advised to have their vehicle checked at a garage. During this check, the estimation system is updated and validated by a technician. A test may be performed to verify that the updated system is functioning correctly.

[0095] In one embodiment, the update of the driven model can take place, for example, when the vehicle is in a parked position. This update does not necessarily require a visit to the garage and the intervention of a professional.

[0096] An example of brake control is shown schematically in [Fig.6].

[0097] The clamping force is estimated from selected characteristics Cl, C3, Cn; for example, characteristic Cl is measured and characteristics C3 and Cn are estimated. The estimated clamping force Fest is combined with the deceleration DC or the desired clamping force F. A PID (Proportional, Integral, Derivative) controller is then applied to determine the control signals SX for the brake motor, such as the current to be applied to modify the clamping force to obtain the desired deceleration. The clamping force is then estimated again, for example.

[0098] The system according to the invention makes it possible to estimate precisely the clamping force applied by a brake, in addition it makes it possible to simplify the control unit and to free up space for the operation of the machine learning means.

[0099] The invention applies to both disc brakes and drum brakes. REFERENCES

[0100] V: vehicle S: braking system ECU: Electronic control unit SX: control signals SE: estimation methods FRI, FR2, FR3, FR4: brakes F: desired clamping force P: processor M: memory area Z: storage area I: Executable machine instructions DL1: neural network DLlp: primitive neural network Cl, C2..., Cn: characteristics DEL2: trained artificial intelligence algorithm BRI, BR2: backpropagation loop FG: expected clamping force Fest: estimated clamping force GP: group of characteristics DC: deceleration PID: Proportional, Integral, Derivative control 4: means of continuous learning 6: Means of comparison

Claims

Demands

1. A method for manufacturing a system for estimating the clamping force exerted by an electric brake intended to be mounted on a given motor vehicle, said estimation system comprising a storage area (Z) including executable machine instructions, memory, and a processor (P) for generating an estimated clamping force, said brake being mounted on a test vehicle or a test bench conforming to the given motor vehicle, said method comprising the steps: a) selecting characteristics of the brake and its environment from a group of characteristics of the brake and its environment, said characteristics being measured or evaluated by means equipping the brake or said vehicle, said brake also comprising a clamping force sensor, said step a) implementing at least a first algorithm,b) development of a function for calculating the estimated clamping force from the selected characteristics and at least a second algorithm which is of an artificial intelligence type.

2. A manufacturing method according to claim 1, wherein the selection step a) comprises a substep of classifying the features and a substep of choosing among said features so as to select the selected features, at least said classification step being carried out by an artificial intelligence algorithm.

3. A manufacturing method according to claim 1 or 2, wherein in step b), the second artificial intelligence algorithm is trained from the selected features and measured clamping forces, so as to obtain a trained artificial intelligence algorithm considered satisfactory.

4. A manufacturing method according to claim 3, wherein in step b) a polynomial function is extracted from the second trained artificial intelligence algorithm considered satisfactory for use by the processor of the estimation system to calculate the estimated clamping force.

5. Manufacturing method according to claim 3, in step b) the calculation function is the second trained artificial intelligence algorithm considered satisfactory.

6. A manufacturing method according to any one of claims 1 to 5, wherein at least the second artificial intelligence algorithm is implemented by a neural network.

7. A manufacturing method according to any one of claims 1 to 6, wherein step b), and possibly also step a), is (are) repeated at several wear levels up to a wear level considered sufficient.

8. Electric brake for motor vehicle comprising means for applying a clamping force to an element attached to a wheel of the motor vehicle, and a system for estimating the clamping force of an electric brake obtained by the method according to any one of the preceding claims and means for collecting selected characteristics, in particular without using a sensor for direct measurement of said clamping force, said collection means comprising sensors and / or means for evaluating all or part of the selected characteristics.

9. Electric brake according to claim 8 in combination with claim 4, said polynomial function being stored in the storage area.

10. Electric brake according to claim 8 in combination with claim 5, wherein the trained artificial intelligence algorithm is stored in the storage area, in particular wherein the processor is a graphics processor.

11. Electric brake according to claim 10, comprising means for continuous learning of the trained artificial intelligence algorithm, said means ensuring at least one detection of a drift of the trained artificial intelligence algorithm.

12. Electric brake according to claim 11, wherein the continuous learning means comprise a backpropagation loop configured to continue training the trained artificial intelligence algorithm during the use of the brake on the motor vehicle.

13. A method for learning the electric brake according to claim 11, wherein the electric brake is mounted on the given motor vehicle, said vehicle comprising means for measuring or estimating the deceleration of said vehicle, said method comprising: - comparison of the estimated clamping force to the measured or estimated deceleration, - monitoring of the variation of the deceleration over time as a function of the clamping force, - verification of the appearance of a drift in the estimated clamping force.

14. A learning method according to claim 13, wherein, when the occurrence of a drift is detected, said method includes a step of learning the trained artificial intelligence algorithm taking into account the drift.

15. Learning method according to claim 13 or 14 taking place during the entire use of the brake.

Images