Electromechanical brake caliper clamping force estimation method based on neural network model

Abstract

The invention discloses an electronic mechanical brake caliper clamping force estimation method based on a neural network. The method specifically comprises the following steps: (1) constructing clamping force estimation models of different types of electronic mechanical brake calipers based on the neural network; and (2) reading the characteristic parameters at the current moment from the electronic mechanical structure caliper body controller, inputting the characteristic parameters into the clamping force pre-estimation model of the electronic mechanical brake caliper of the corresponding model, and outputting the clamping force at the current moment by the clamping force pre-estimation model. The estimated clamping force changes in real time according to the action state of the electro-mechanical brake caliper body and the environment change, and the clamping force in the caliper body action process is accurately fed back; and the clamping force can be obtained without a clamping force sensor, so that the system cost is effectively reduced.

Description

Technical Field

[0001] The present invention belongs to the field of brake technology, and more specifically, the present invention relates to a method for estimating the clamping force of an electronic mechanical brake caliper based on a neural network model. Background Art

[0002] With the advancement of electrification, intelligence, and wire-controlled vehicles, the demand for wire-controlled chassis actuation systems is also increasing. The advantages of wire-controlled electromechanical brakes, such as flexible layout, environmental friendliness, higher efficiency, faster response, and sufficient redundancy, are gradually emerging and becoming a new trend in automotive chassis development. Electromechanical brake (EMB) systems, directly driven by wheel-end motors, use a motion conversion mechanism to convert the motor's torque and rotational motion into thrust and translational motion of the connector, pushing the brake pads to clamp the brake disc, thereby generating braking force. Furthermore, as an emerging braking system, the EMB system eliminates bulky components such as vacuum boosters and hydraulic lines, making the overall vehicle chassis layout simpler and more flexible. It also offers fast and precise pressure regulation, significantly improving the vehicle's braking performance.

[0003] In the existing electromechanical brake (EMB) caliper technology solution, the brake transmission mechanism requires a clamping force sensor to detect the clamping force signal during the electromechanical braking process, so that the brake caliper can respond accurately and promptly to the target clamping force request. Figure 1 As shown, the electronic mechanical brake caliper (with a clamping force sensor solution) mainly includes the following components: caliper inlay (1), friction plate (2), brake disc (3), transmission mechanism (4a, 4b, 4c), clamping force sensor (5), motor (6). However, the solution with a clamping force sensor has the following disadvantages:

[0004] (1) The cost of the clamping force sensor is relatively high, accounting for a relatively large proportion of the product cost, which causes the cost of the electronic mechanical brake system to remain high, seriously affecting the competitiveness of the product;

[0005] (2) When installing the clamping force sensor of the electronic mechanical brake caliper, there will be installation errors, which will cause detection errors in the clamping force detection. Summary of the Invention

[0006] The present invention provides an electronic mechanical brake caliper clamping force estimation method based on a neural network model, aiming to improve the above-mentioned problem.

[0007] The present invention is implemented as follows: a method for estimating the clamping force of an electromechanical brake caliper based on a neural network model, the method being specifically as follows:

[0008] (1) Construct a clamping force estimation model for different types of electronic mechanical brake calipers based on a neural network model;

[0009] (2) Reading characteristic parameters at the current moment from the electronic mechanical structure caliper controller, inputting the characteristic parameters into the clamping force estimation model of the corresponding electronic mechanical brake caliper model, and the clamping force estimation model outputs the clamping force at the current moment.

[0010] Furthermore, the neural network model is trained offline, and after the training is completed, a clamping force estimation model is formed. The clamping force estimation model is integrated into the electronic brake caliper controller to perform clamping force estimation online.

[0011] Furthermore, the clamp stroke in the electromechanical structure clamp controller is combined with at least one of the motor current, the motor angle, and the motor acceleration as a characteristic parameter.

[0012] Furthermore, sample data of different types of electronic mechanical brake calipers are used to independently train the neural network model to form a clamping force estimation model of different types of electronic mechanical brake calipers.

[0013] Furthermore, the sample data acquisition process of different models of electronic mechanical brake calipers is as follows:

[0014] By simulating different working conditions through bench tests, characteristic parameters and the clamping force of the electronic mechanical brake caliper corresponding to the characteristic parameters are obtained from different types of electronic mechanical structure caliper controllers to form sample data of different types of mechanical brake calipers.

[0015] Furthermore, the neural network training process is as follows:

[0016] The weight parameters and bias parameters of the neural network are trained using samples of the current model of mechanical brake calipers until the clamping force estimation accuracy of the neural network reaches the set standard. The weight parameters and bias parameters of the current neural network are determined as the clamping force estimation model parameters of the current model of electronic mechanical brake calipers.

[0017] Furthermore, the neural network model is a BP neural network, and the structure of the BP neural network includes three layers, namely, an input layer, a hidden layer, and an output layer, wherein: the input layer is used to input feature parameters; the hidden layer assigns a weight to the input of each input layer, and sums it with the deviation to form the input of the neuron transfer function, and the transfer function adopts a sigmoid transfer function; the output layer is composed of linear neurons and outputs a clamping force signal.

[0018] Furthermore, it is characterized in that a neural network using the LM algorithm is used for training.

[0019] Furthermore, the model of the current electronic mechanical brake caliper is determined based on the torque parameter of the motor, the friction of the transmission structure, the operating temperature of the electronic mechanical brake system, and the thickness of the brake friction pad.

[0020] This paper proposes a neural network-based method for estimating the clamping force of an electromechanical brake caliper. This method estimates the clamping force in real time by training a neural network model. The estimated clamping force changes in real time based on the operating state of the electromechanical brake caliper and environmental changes, providing accurate feedback on the clamping force during caliper operation. This method eliminates the need for a clamping force sensor, effectively reducing system costs. Furthermore, this method reduces tangential force measurement errors and installation errors. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 A schematic structural diagram of an electromechanical brake caliper provided in an embodiment of the present invention;

[0022] Figure 2 A flowchart of a method for estimating the clamping force of an electromechanical brake caliper based on a neural network according to an embodiment of the present invention;

[0023] Figure 3 A schematic diagram of the BP neural network model structure provided by an embodiment of the present invention;

[0024] Figure 4 A schematic diagram of the structure of a hidden layer neural network provided by an embodiment of the present invention;

[0025] Figure 5 A schematic diagram of the output layer network structure provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0026] The specific implementation methods of the present invention will be further explained in detail below by describing the embodiments with reference to the accompanying drawings, so as to help those skilled in the art to have a more complete, accurate and in-depth understanding of the inventive concept and technical solution of the present invention.

[0027] Figure 2 A flowchart of a method for estimating the clamping force of an electronic mechanical brake caliper based on a neural network model provided in an embodiment of the present invention is provided. The method for estimating the clamping force of an electronic mechanical brake caliper based on a neural network model mainly consists of two parts: the first part is the training of the neural network model; the second part is the use of the trained neural network model (clamping force estimation model). The method specifically includes the following steps:

[0028] (1) Construct a clamping force estimation model for different types of electronic mechanical brake calipers based on a neural network model;

[0029] In an embodiment of the present invention, the neural network model in step (1) can be a BP neural network model or an ANN neural network model. The neural network model is trained offline to form a clamping force estimation model after training. The clamping force estimation model is integrated into the electronic brake caliper controller to perform clamping force estimation offline.

[0030] In the embodiment of the present invention, the structure of the BP neural network includes three layers, namely, the input layer, the hidden layer, and the output layer, as shown in the figure. Figure 3 As shown, the input layer is used to input feature parameters; the hidden layer assigns a weight to the input of each input layer, and sums it with the deviation to form the input of the neuron transfer function. The transfer function adopts the sigmoid transfer function, as shown in the figure. Figure 4 As shown, IW1.1 is the weight, b1 is the bias value, and the output layer is composed of linear neurons, as shown in the figure. Figure 5 As shown, IW2.1 is the weight, b2 is the deviation value, and the final output is the clamping force signal.

[0031] In an embodiment of the present invention, to improve the clamping force estimation accuracy of the clamping force estimation model, the clamping force estimation models for different types of electronic mechanical brake calipers are trained independently. If there are m types of electronic mechanical brake calipers on the market, then m clamping force estimation models are generated. The offline training method of the neural network model is described in detail below. The offline training process of the neural network model consists of two parts: the sample construction process and the sample-based neural network model training process. The above two processes are described below:

[0032] Sample construction process: Through bench tests to simulate different working conditions, characteristic parameters and the corresponding clamping force of the electronic mechanical brake caliper are read from the electronic mechanical structure caliper controller to form sample data. Sample data of different models of mechanical brake calipers are obtained.

[0033] The training process of the neural network model: The weight parameters and bias parameters of the neural network model are trained through samples of the current model of mechanical brake calipers until the clamping force estimation accuracy of the neural network model reaches the set standard. The set standard can be understood as the minimum value of the estimation accuracy required for the vehicle. The weight parameters and bias parameters of the current neural network model are determined as the clamping force estimation model parameters of the current model of electronic mechanical brake calipers, and the clamping force estimation models of different models of mechanical brake calipers are obtained.

[0034] The neural network model is continuously trained and iterated based on sample data to correct the weights and bias values in the hidden layer neurons and output layer neurons. The learning of the neural network model adopts supervised learning.

[0035] (2) Reading characteristic parameters at the current moment from the electronic mechanical structure caliper controller, inputting the characteristic parameters into the clamping force estimation model of the corresponding electronic mechanical brake caliper model, and the clamping force estimation model outputs the clamping force at the current moment.

[0036] In an embodiment of the present invention, due to the strong correlation between the caliper stroke and the clamping force estimation, the caliper stroke is used as the characteristic parameter of the sample, and the correlation between the motor current, motor angle, motor acceleration and the clamping force estimation is relatively weak. Therefore, the caliper stroke in the electronic mechanical structure caliper controller is combined with at least one of the motor current, motor angle, and motor acceleration as the characteristic parameter. Therefore, when obtaining sample data through bench tests, it is necessary to collect the above-mentioned characteristic data and the corresponding clamping force of different models of mechanical brake calipers under different working conditions. During online use, the above-mentioned characteristic parameters at the current moment are read from the electronic mechanical structure caliper controller in real time, and the corresponding clamping force estimation model is input. The clamping force estimation model predicts the clamping force corresponding to the mechanical brake caliper corresponding to the above-mentioned characteristic parameters at the current moment.

[0037] In an embodiment of the present invention, the model of the current electronic mechanical brake caliper is determined based on the torque parameters of the motor, the friction of the transmission structure, the operating temperature of the electronic mechanical brake system, and the thickness of the brake friction pad. If the above parameters are inconsistent, it is determined that the corresponding electronic mechanical brake calipers are different.

[0038] The neural network-based electromechanical brake caliper clamping force estimation method provided by the present invention has the following advantages:

[0039] (1) The real-time detection of the system clamping force is achieved through the neural network algorithm estimation method, replacing the original caliper clamping force sensor, effectively reducing the system cost and increasing the competitiveness of the product;

[0040] (2) The clamping force estimation model of the present invention adopts a method of separating model training and model operation. The model training is performed using a working computer or a cloud computing unit, and the trained parameters are updated to the controller algorithm model, which separates the training computing power and the operation computing power, reduces the excessive requirements of the model training on the controller operation unit, and effectively reduces the controller cost.

[0041] The present invention has been described exemplarily. Obviously, the specific implementation of the present invention is not limited to the above-mentioned method. As long as various non-substantial improvements are made using the method concept and technical solution of the present invention, or the concept and technical solution of the present invention are directly applied to other occasions without improvement, they are all within the scope of protection of the present invention.

Claims

1. A method for estimating the clamping force of an electromechanical brake caliper based on a neural network model, characterized in that: The method is specifically as follows: (1) Construct a clamping force estimation model for different types of electronic mechanical brake calipers based on a neural network model; (2) Reading characteristic parameters at the current moment from the electronic mechanical structure caliper controller, inputting the characteristic parameters into the clamping force estimation model of the corresponding electronic mechanical brake caliper model, and the clamping force estimation model outputs the clamping force at the current moment.

2. The method for estimating the clamping force of an electromechanical brake caliper based on a neural network model according to claim 1, wherein: The neural network model is trained offline, and a clamping force estimation model is formed after the training is completed. The clamping force estimation model is integrated into the electronic brake caliper controller to perform clamping force estimation online.

3. The method for estimating the clamping force of an electromechanical brake caliper based on a neural network model as claimed in claim 2, wherein: The clamp body stroke in the electronic mechanical structure clamp body controller is combined with at least one of the motor current, the motor angle, and the motor acceleration as a characteristic parameter.

4. The method for estimating the clamping force of an electromechanical brake caliper based on a neural network model as claimed in claim 2, wherein: The neural network model is trained independently using sample data of different types of electronic mechanical brake calipers to form a clamping force estimation model for different types of electronic mechanical brake calipers.

5. The method for estimating the clamping force of an electromechanical brake caliper based on a neural network model as claimed in claim 4, wherein: The sample data acquisition process for different models of electronic mechanical brake calipers is as follows: By simulating different working conditions through bench tests, characteristic parameters and the clamping force of the electronic mechanical brake caliper corresponding to the characteristic parameters are obtained from different models of electronic mechanical structure caliper controllers to form sample data of different models of mechanical brake calipers.

6. The method for estimating the clamping force of an electromechanical brake caliper based on a neural network model as claimed in claim 4, wherein: The neural network training process is as follows: The weight parameters and bias parameters of the neural network are trained using samples of the current model of mechanical brake calipers until the clamping force estimation accuracy of the neural network reaches the set standard. The weight parameters and bias parameters of the current neural network are determined as the clamping force estimation model parameters of the current model of electronic mechanical brake calipers.

7. The method for estimating the clamping force of an electromechanical brake caliper based on a neural network model according to any one of claims 1 to 6, characterized in that: The neural network model is a BP neural network. The structure of the BP neural network includes three layers, namely, an input layer, a hidden layer, and an output layer. Among them, the input layer is used to input feature parameters; the hidden layer assigns a weight to the input of each input layer, and sums it with the deviation to form the input of the neuron transfer function. The transfer function adopts a sigmoid transfer function; the output layer is composed of linear neurons and outputs a clamping force signal.

8. The method for estimating the clamping force of an electromechanical brake caliper based on a neural network model according to any one of claims 1 to 6, characterized in that: The neural network is trained using the LM algorithm.

9. The method for estimating the clamping force of an electromechanical brake caliper based on a neural network model according to any one of claims 1 to 6, characterized in that: The model of the current electronic mechanical brake caliper is determined based on the torque parameters of the motor, the friction of the transmission structure, the operating temperature of the electronic mechanical brake system, and the thickness of the brake friction pad.

Images