Motor electric control brake system clamping force self-adaptive correction method based on deep learning

Abstract

The application discloses a motor electric control brake system clamping force self-adaptive correction method based on deep learning, belongs to the chassis control technical field, and fully considers the brake pad wear condition in the use process. In view of the nonlinear change problem caused by the brake pad wear, a deep learning neural network optimization control strategy is introduced, a complex nonlinear relationship between the estimated clamping force of the motor rotation angle and the actual clamping force is modeled as a predictable probability distribution, and self-adaptive correction is carried out, so that the clamping force control is optimized under the condition of brake pad wear, the dynamic change caused by the brake pad wear is effectively coped with, the real-time performance of system control is enhanced, the prediction accuracy and response speed of the system are greatly improved, the stability, reliability and safety of the brake system are comprehensively enhanced, and meanwhile, real-time clamping force monitoring by using a large number of sensors is avoided.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of chassis control, and particularly relates to a motor electric control brake system clamping force self-adaptive correction method based on deep learning. BACKGROUND

[0002] Motor electric control brake systems (EMBs) achieve rapid and precise control of brake force through electrical signals. With their efficient, adjustable, and precise characteristics, they are gradually replacing traditional hydraulic brake systems and becoming important systems and key components in future automotive chassis. Compared with traditional mechanical or hydraulic brake systems, motor electric control brake systems can more accurately adjust brake force distribution, not only improving brake response speed, but also greatly improving brake system stability and safety. The main function of an electronic mechanical brake (EMB) system is to control the target clamping force. Clamping force control is essentially feedback control of the motor. Common feedback control algorithms include proportional-integral-derivative (PID) control algorithms, linear unconstrained model predictive control (MPC) algorithms, and sliding mode variable structure control (SMC) algorithms. EMB systems are composed of motors, torque-increasing speed reducers, motion conversion devices, and brake components. These systems usually use pressure sensors to monitor brake pressure in real time, ensuring accurate control during braking. However, these systems often require a large number of sensors and have a shorter lifespan in high-temperature environments, which limits their application in harsh conditions.

[0003] Currently, one method widely studied in the industry and academia is to assume that key physical components of the brake system, particularly the thickness and texture of the brake pads, remain consistent and unchanged throughout the entire usage cycle. This assumption simplifies the brake force control model because the physical properties of the brake pads directly affect the efficiency and accuracy of brake force transmission. However, in actual applications, brake pads undergo physical wear during continuous use, particularly a gradual reduction in thickness. The wear of the brake pads not only changes their contact characteristics but also indirectly increases the overall physical gap in the brake system. The increase in the gap leads to a critical problem: the estimation of clamping force based on motor rotation angle begins to deviate because the motor must rotate a larger angle to further push the brake pad against the brake disc, thereby compensating for the increased gap. In this case, if the motor angle remains unchanged, the actual applied clamping force will decrease due to the increased physical gap, resulting in actual clamping force lower than the estimated clamping force from the motor angle, thereby affecting the accuracy of clamping force control.

[0004] In summary, although there has been some progress in the research and application of motor electric control brake systems, the existing brake force control technology still has some shortcomings:

[0005] 1) Existing sensors have limited performance in high temperature and harsh environments, resulting in shortened service life or failure, affecting the accurate control of brake force;

[0006] 2) Traditional control methods usually assume that the physical properties of the brake pad remain unchanged, but the brake pad will wear during use, causing changes in contact characteristics and physical clearance, which makes the relationship between the estimated clamping force of the motor angle and the actual clamping force complex and nonlinear;

[0007] 3) Traditional linear control models are difficult to cope with the effects of brake pad wear and complex dynamic changes, and cannot accurately predict the clamping force, affecting the response speed and stability of the brake system;

[0008] 4) Existing systems usually require a large number of sensors, which not only increases production costs, but also brings difficulties in maintenance and calibration, limiting its wide application;

[0009] 5) The existing brake system has a long response time, which may not be able to produce enough braking force in time in emergency braking or rapid response scenarios, affecting the braking effect and safety;

[0010] Therefore, how to improve the robustness, precision and response speed of the system, overcome these challenges, has become the key to the development of motor electric control brake system. SUMMARY

[0011] In order to solve the problems of the prior art, effectively cope with the dynamic changes caused by brake pad wear, greatly improve the prediction accuracy and response speed of the brake system, and comprehensively enhance the stability, reliability and safety of the brake system, the technical scheme adopted by the present application is as follows:

[0012] The motor electric control brake system clamping force self-adaptive correction method based on deep learning comprises the following steps:

[0013] Step 1: Construct a mathematical model of the brake system;

[0014] Based on the brake signal, the motor rotates and outputs torque, converting the rotary motion into linear motion of the brake pad, so that the brake pad clamps the brake disc to brake; the displacement of linear motion is obtained according to the motor rotation angle, the caliper deformation of the brake pad clamping the brake disc is calculated through the initial position and the displacement, and the nonlinear relationship between the deformation and the clamping force is constructed to establish the relationship between the motor rotation angle and the clamping force;

[0015] Step 2: Construct a neural network model;

[0016] Obtain the predicted clamping force of the mathematical model, generate the target clamping force through the neural network model, and correct the actual clamping force through the target clamping force.

[0017] Further, in step 1, the relationship between the motor rotation angle and the screw nut displacement is modeled by a planetary gear reducer and a ball screw, and is as follows:

[0018]

[0019] wherein P represents the pitch, θ represents the motor rotation angle (unit: radian), and i represents the transmission ratio of the reducer.

[0020] Further, the deformation is obtained according to the difference between the screw nut displacement s and the initial brake gap s0.

[0021] Δs = s - s0

[0022] The clamping force F is constructed n and the nonlinear relationship between the deformation Δs is as follows:

[0023] F n = K f1 Δs 3 + K f2 Δs 2 + K f3 Δs + K f4

[0024] wherein K f1 , K f2 , K f3 , K f4 represent the coefficients of the clamping force.

[0025] Further, the motor rotates and outputs torque, and then the low torque output by the motor is converted into high torque by a torque multiplication device, so as to construct the relationship between the clamping force and the motor rotation torque.

[0026]

[0027] wherein T L represents the motor rotation torque, F n represents the clamping force, L represents the lead of the ball screw, η s represents the transmission efficiency of the ball screw, i p represents the transmission ratio of the planetary gear reducer, and η p represents the efficiency of the transmission system.

[0028] Further, the loss function of the neural network model of step 2 adopts the mean square error (MSE), and the calculation formula is as follows:

[0029]

[0030] wherein y represents the true value, m represents the sample number, and represents the predicted value.

[0031] Further, the neural network model of step 2 adopts a multi-layer perceptron, uses a ReLU activation function to introduce a nonlinear characteristic, thereby enhancing the learning ability of the model, and the ReLU function is defined as follows:

[0032] σ(z) = max(0, z)

[0033] wherein z represents the neuron input of the multi-layer perceptron, and σ(z) represents the ReLU activation function, which sets all negative values to zero while keeping all non-negative values unchanged; the output of each layer is calculated as follows:

[0034] z (i) = W (i) a (i) +b (i)

[0035] a (i+1) = σ(z (i) )

[0036] wherein W (i) and b (i) represent the weights and biases of the previous layer, respectively, a (i) represents the output of the previous layer, a (i+1) represents the output of the current layer as the input of the next layer, and z (i) represents the neuron input of the previous layer.

[0037] Further, the neural network model adjusts the weights W and biases b to minimize the loss function by training the multi-layer perceptron, and uses stochastic gradient descent (SGD) as the optimization algorithm, and the update rule of stochastic gradient descent (SGD) is as follows:

[0038]

[0039] wherein θ represents the parameters of the neural network model, including weights and biases, η represents the learning rate, and represents the gradient of the loss function L with respect to the parameters θ.

[0040] The motor electric control brake system based on deep learning includes physical hardware components and a feedback control system, the feedback control system includes a clamping force loop, the clamping force loop adopts the clamping force self-adaptive correction method based on the deep learning motor electric control brake system, and the clamping force is corrected in real time to improve the accuracy of clamping force control.

[0041] Further, the feedback control system further includes a speed loop and a current loop, the speed loop obtains speed data through an encoder installed on the motor for speed feedback control, and the current loop obtains current data through the encoder for current regulation.

[0042] Further, the physical hardware components include a permanent magnet synchronous motor, a planetary gear set, a ball screw and a caliper brake disc; the permanent magnet synchronous motor is used to provide a power source; the planetary gear set is used to increase the output torque; the ball screw converts the rotary motion into linear motion, so that the braking process is completed through the caliper brake disc.

[0043] The advantages and beneficial effects of the present application are that:

[0044] The present application uses a neural network model to replace the traditional clamping force sensor, reduces the dependence of the brake system on hardware sensors, has stronger adaptability, and considers the loss of clamping force caused by brake pad wear after long-term use in the modeling process; the present application proposes a self-adaptive clamping force correction strategy, which uses a multilayer perceptron network, uses the estimated clamping force of the motor rotation angle as input, uses the actual clamping force as training target, predicts the probability distribution of the clamping force, and estimates the actual clamping force by calculating the expected value, compensates for the clamping force estimation error, effectively compensates for the clamping force estimation deviation caused by brake pad wear, significantly improves the control performance of the brake system, and ensures the reliability and accuracy of the system during long-term operation; the feedback control method based on the neural network provided by the present application successfully captures the complex nonlinear relationship between the motor rotation angle and the actual clamping force, the network model predicts the probability distribution of the clamping force and calculates the expected value, realizes real-time correction of the clamping force, and ensures that the brake system can quickly respond to the target value. BRIEF DESCRIPTION OF DRAWINGS

[0045] Figure 1 is a flowchart of the method in the embodiment of the present application.

[0046] Figure 2 is a flowchart of the braking process in the embodiment of the present application.

[0047] Figure 3 is a structural schematic diagram of the brake system in the embodiment of the present application.

[0048] Figure 4 is a diagram of the estimated clamping force of the motor and the actual clamping force when the target brake force is 8000N in the embodiment of the present application.

[0049] Figure 5 is a diagram of the estimated clamping force of the motor and the actual clamping force when the target brake force is 6000N in the embodiment of the present application.

[0050] Figure 6 is a diagram of the estimated clamping force of the motor and the actual clamping force when the target brake force is 4000N in the embodiment of the present application.

[0051] Figure 7is a comparison chart of the estimated clamping force after correction by the neural network and the actual clamping force when the target clamping force is 8000N in the embodiment of the present application.

[0052] Figure 8 is a comparison chart of the estimated clamping force after correction by the neural network and the actual clamping force when the target clamping force is 6000N in the embodiment of the present application.

[0053] Figure 9 is a comparison chart of the estimated clamping force after correction by the neural network and the actual clamping force when the target clamping force is 4000N in the embodiment of the present application. DETAILED DESCRIPTION

[0054] The specific embodiments of the present application will be described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are only for illustration and explanation of the present application, and are not intended to limit the present application.

[0055] As shown in Figure 1 , the deep learning-based motor electric control brake system clamping force adaptive correction method of the present application first models the electronic mechanical brake system mathematically, then proposes an innovative adaptive clamping force correction strategy based on neural network, optimizes the performance of the motor electric control brake system in depth, through the carefully constructed multi-layer perceptron network model, combined with multiple hidden layer structures, the estimated clamping force of the motor rotation angle as input, the actual measured clamping force as training label, realizes the efficient modeling and accurate prediction of the complex nonlinear relationship between the clamping force and the motor angle, the model generates a probability distribution in the output layer by using the Softmax function, so as to effectively capture the potential trend of the clamping force under different input conditions, through the accurate calculation of the expected value of the probability distribution, the method can accurately predict the actual clamping force in complex dynamic environment, and overcome the limitations of traditional control methods, this innovative method based on probability prediction not only greatly improves the accuracy of prediction, but also enhances the reliability and accuracy of brake force control. Specifically includes the following steps:

[0056] Step 1: Selecting and modeling the electronic mechanical brake;

[0057] As shown in Figure 2 , after receiving the brake signal, the electronic mechanical brake (EMB) system will apply brake pressure. The motor starts to rotate and output torque; the torque multiplier converts the low torque output by the motor into high enough torque; the motion conversion device converts the rotary motion of the motor into linear motion of the brake pad, the brake pad clamps the brake disc, thereby realizing the brake.

[0058] The planetary gear reducer and ball screw are modeled. During braking, the deformation of the brake pad when clamping the brake disc will generate clamping force. The deformation of the caliper Δs is defined as the difference between the nut displacement s and the initial brake gap s0:

[0059] Δs = s - s0 (1)

[0060] The unit of deformation is millimeter (mm), and there is a cubic non-linear relationship between the generated clamping force and the deformation:

[0061] F n = K f1 Δs 3 + K f2 Δs 2 + K f3 Δs + K f4 (2)

[0062] Where K f1 , K f2 , K f3 , K f4 represent the coefficients of clamping force. Through the size of the clamping force, the load torque on the EMB motor can be calculated:

[0063]

[0064] Where i p is the transmission ratio of the planetary gear reducer, η s is the transmission efficiency of the ball screw, L is the lead of the ball screw (unit: millimeter), and η p is the efficiency of the transmission system.

[0065] The relationship between the motor rotation angle and the nut displacement is:

[0066]

[0067] Where P is the pitch, θ is the motor rotation angle (unit: radian), and i is the transmission ratio of the reducer.

[0068] According to formulas (1), (2) and (4), the current clamping force can be estimated. Through curve fitting with test bench data, the characteristic curve between caliper deformation and clamping force is obtained. Further, using formula (3), a function is established to describe the relationship between motor rotation angle and clamping force (angle unit is radian). Based on the existing test bench data and literature modeling, there is a non-linear relationship between the clamping force F N and the displacement Δs:

[0069]

[0070] Step 2: Neural network model construction;

[0071] The caliper deformation variable Δs is the difference between the nut displacement s and the brake clearance s0. As the brake is used continuously, the brake clearance s0 increases, resulting in a smaller Δs. This difference leads to a deviation between the estimated clamping force and the actual clamping force. In addition, the relationship between the estimated clamping force and the actual clamping force is a complex nonlinearity that is initially calibrated.

[0072] The present invention employs a multilayer perceptron to solve the regression problem, including network architecture selection, activation function application, loss function determination, and optimization algorithm selection.

[0073] Input layer: receives one input feature, which is the pre-processed predicted clamping force value.

[0074] Hidden layers: composed of three layers of multilayer perceptron. The first layer contains 64 neurons, the second layer also contains 64 neurons, and the third layer contains 128 neurons. The ReLU activation function σ(z) is used to introduce nonlinearity, enhancing the learning ability of the model. The ReLU function is defined as follows:

[0075] σ(z) = max(0, z) (6)

[0076] where z is the input of the neuron. The role of the ReLU function is to set all negative values to zero, while keeping all non-negative values unchanged. The output of each layer is calculated as follows:

[0077] z (i) = W (i) a (i) +b (i) (7)

[0078] a (i+1) = σ(z (i) ) (8)

[0079] where W (i) and b (i) are the weights and biases of the previous layer, a (i) is the output of the previous layer, and a (i+1) is the output of the current layer, serving as the input of the next layer.

[0080] Output layer: the output layer is composed of 2000 neurons, which use the softmax function to represent the distribution of the target clamping force within a specified range. Then, the expected value of this probability distribution is calculated to obtain the corrected actual clamping force.

[0081] Loss function: to train the network, the mean squared error (MSE) is used as the loss function. The calculation formula is as follows:

[0082]

[0083] where y represents the true value, m is the number of samples, is the predicted value.

[0084] Training optimization: Training the MLP involves adjusting the weights W and biases b to minimize the loss function. Stochastic gradient descent (SGD) is employed as the optimization algorithm. The update rule for SGD is as follows:

[0085]

[0086] where θ represents the parameters of the model, including weights and biases, η is the learning rate, is the gradient of the loss function with respect to the parameters.

[0087] As shown in Figure 3 , the electromechanical brake system and its control architecture include physical hardware components and a feedback control system.

[0088] The physical hardware components mainly include a permanent magnet synchronous motor providing the main power, a planetary gear set increasing the output torque, a ball screw converting rotary motion into linear motion, and a caliper brake disc effectively performing the braking action. This configuration achieves efficient power transmission and precise control of braking force. The permanent magnet synchronous motor serves as the main power source, the planetary gear set is used to increase the output torque, and the ball screw converts the rotary motion of the motor into linear motion of the brake pad, thereby completing the braking process through the caliper brake disc.

[0089] The feedback control system adopts a three-loop feedback control structure, including a clamping force loop, a speed loop, and a current loop.

[0090] The clamping force loop is adjusted in real time by accurately calculating the motor rotation angle and combining the neural network constructed based on the deep learning-based motor electric brake system clamping force adaptive correction method to improve the accuracy of clamping force control.

[0091] The speed loop obtains speed data through the encoder installed on the motor for speed feedback control.

[0092] The current loop obtains current data through the encoder for current regulation.

[0093] In terms of control strategy, the clamping force loop and the speed loop mainly adopt PID feedback control technology, and the current loop adopts model predictive control (MPC) algorithm to improve performance verification effect.

[0094] Model training and performance evaluation: In model training, the cross-validation method is used to evaluate the accuracy and generalization ability of the model. The evaluation indicators include: 1) Mean Squared Error (MSE); 2) Root Mean Squared Error (RMSE); 3) Mean Absolute Error (MAE); 4) Determination coefficient (R2 );

[0095] System performance evaluation: The main evaluation indicators of system performance include: 1) steady-state error: measures the deviation between the output value and the target set value after the system reaches a steady state, reflecting the accuracy and reliability of the system in long-term operation; 2) transient response: describes the process from the initial state to the steady state, including response speed and dynamic characteristics in the transition process.

[0096] This architecture combines efficient hardware design with intelligent feedback control strategies, not only improving the control accuracy of the brake system, but also enhancing its running stability and reliability in complex environments.

[0097] Embodiment:

[0098] The purpose of this experiment is to verify the effectiveness of the adaptive clamping force correction strategy based on multi-layer perceptron (MLP), especially its performance under simulated brake wear conditions. In the experiment, in order to construct and test the electromechanical brake system and its control strategy, we choose to use the SIMULINK 2024a platform to create a comprehensive model of the brake system. The model includes components such as motors, sensors and control algorithms, which can visually demonstrate the real-time response and adjustment process of the system. In addition, in order to implement the adaptive clamping force correction strategy based on neural networks, we develop and train a model on the PyTorch platform. After the training phase is completed, the neural network model is exported in ONNX (Open Neural Network Exchange) format and imported into the SIMULINK environment. This integration helps to realize the real-time application and verification of the model in the experimental setup.

[0099] Data collection: brake actuator wear simulation. We simulate the wear of the brake actuator and collect data on the estimated clamping force through the motor shaft angle and the actual clamping force. Assuming that the initial brake clearance is 0.7mm, the brake pad wear increases to 0.75mm.

[0100] Network training: neural network training using collected data. The collected data is used to train the neural network. The input of the network is the motor shaft angle, and the label is the actual measured clamping force. The network architecture consists of an input layer, several hidden layers and an output layer. The output layer uses the softmax function to predict the probability distribution of the clamping force, and determines the clamping force by calculating the expected value of this distribution.

[0101] Experimental results:

[0102] Influence of brake pad wear on clamping force control: influence of brake gap increase. Assuming an initial brake gap of 0.7mm, the gap increases to 0.75mm after brake pad wear. When the target brake force is set to 4000N, 6000N and 8000N respectively, the clamping force estimated by motor rotation angle and the actual clamping force are as shown in Figures 4 to 6 The red line represents the clamping force estimated by the motor, and the yellow line represents the actual clamping force. Although the clamping force estimated by the motor rotation angle can reach the target value, the actual clamping force has a lag phenomenon, and the value is lower than the estimated value.

[0103] Model training and regression performance index evaluation. The model performance is shown in Table 1.

[0104] Table 1 Model fitting results table

[0105]

[0106] The model is further integrated into the brake control system to adapt to and accurately predict the wear state of the brake pad. The predicted clamping force and the actual clamping force are as shown in Figures 7 to 9 It can be seen that the model successfully learns the relationship between the clamping force estimated by the motor rotation angle and the actual clamping force, and shows good prediction accuracy. After adjustment, the clamping force curve estimated by the motor rotation angle and the actual clamping force curve are highly consistent, and can quickly and stably approach the desired target value.

[0107] The above examples are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing examples, those skilled in the art should understand that they can still modify the technical solutions recorded in the foregoing examples, or make equivalent replacement for part or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application.

Claims

1. A deep learning-based adaptive correction method for the clamping force of an electric motor-controlled brake system, characterized by The steps include: Step 1: Construct a mathematical model of the brake system; Based on the brake signal, the motor rotates and outputs torque, converting the rotational motion into linear motion of the brake pad, causing the brake pad to clamp the brake disc to brake; The linear motion displacement is obtained according to the motor rotation angle, and the deformation of the caliper when the brake pad clamps the brake disc is calculated based on the initial position and the displacement. A nonlinear relationship between the deformation of the caliper and the clamping force is constructed to establish the relationship between the motor rotation angle and the clamping force. Modeled by a planetary gear reducer and a ball screw, the relationship between the motor rotation angle and the displacement of the screw nut is: in, Indicates the pitch, Indicates the motor rotation angle, Indicates the transmission ratio of the reducer; The deformation is obtained according to the difference between the nut displacement s and the initial brake clearance s0: Building clamping force and deformation The nonlinear relationship between: in, A coefficient that represents the clamping force; Step 2: Build a neural network model; Obtain the clamping force predicted by the mathematical model, generate the target clamping force through the neural network model, and correct the actual clamping force based on the target clamping force; The loss function of the neural network model uses mean square error, and the calculation formula is as follows: in, represents the true value, represents the number of samples, represents the predicted value; The neural network model uses a multi-layer perceptron and uses an activation function to introduce nonlinear characteristics. The activation function is defined as follows: Where z represents the neuron input of the multilayer perceptron, represents the activation function, which is used to set all negative values ​​to zero and keep all non-negative values ​​unchanged; the output of each layer is calculated as follows: in, and Represent the weight and bias of the previous layer respectively, represents the output of the previous layer, Represents the output of the current layer, which serves as the input of the next layer. Represents the neuron input of the previous layer; The neural network model trains a multi-layer perceptron, adjusts weights and biases to minimize the loss function, and uses stochastic gradient descent as the optimization algorithm. The update rule of stochastic gradient descent is as follows: in, Represents the parameters of the neural network model, including weights and biases, represents the learning rate, Represents the loss function Relative to the parameter gradient.

2. The method for adaptively correcting the clamping force of a motor-controlled brake system based on deep learning according to claim 1, characterized in that: The motor rotates and outputs torque, and the torque multiplier converts the low torque output by the motor into high torque to establish the relationship between the clamping force and the motor rotation torque: in, Indicates the motor rotation torque, Indicates the clamping force, Indicates the lead of the ball screw, Indicates the transmission efficiency of the ball screw, Indicates the transmission ratio of the planetary gear reducer, Indicates the efficiency of the transmission system.

3. A deep learning-based motor-controlled brake system, including physical hardware components and a feedback control system, characterized by: The feedback control system includes a clamping force loop, which adopts the deep learning-based adaptive correction method for the clamping force of the motor-controlled brake system according to claim 1 to correct the clamping force in real time.

4. The deep learning-based motor-controlled brake system according to claim 3, characterized in that: The feedback control system further includes a speed loop and a current loop. The speed loop obtains speed data through an encoder installed on the motor for speed feedback control; the current loop obtains current data through the encoder for current regulation.

5. The motor-controlled brake system based on deep learning according to claim 3, characterized in that: The physical hardware components include a permanent magnet synchronous motor, a planetary gear set, a ball screw and a caliper brake disc; the permanent magnet synchronous motor is used to provide a power source; the planetary gear set is used to increase the output torque; the ball screw converts rotational motion into linear motion, thereby completing the braking process through the caliper brake disc.

Images