A method for diagnosing brake pad wear based on fuzzy neural networks

Abstract

This invention belongs to the field of vehicle braking technology, specifically providing a brake pad wear diagnosis method based on a fuzzy neural network. The method includes: determining characteristic variables of brake pad wear thickness as input variables; constructing an initial brake pad wear diagnosis model based on a fuzzy neural network; acquiring training samples and inputting them into the initial brake pad wear diagnosis model, and determining a final brake pad wear diagnosis model using an adaptive particle swarm optimization algorithm; inputting the input variables into the final brake pad wear diagnosis model to calculate the brake pad wear thickness and remaining thickness; determining whether the remaining brake pad thickness is less than a preset minimum brake pad thickness threshold; obtaining the brake pad wear thickness using a brake pad wear prediction model established based on a fuzzy neural network, and optimizing the model parameters using an adaptive particle swarm optimization algorithm; and calculating the remaining thickness based on the wear thickness, thereby achieving real-time monitoring and diagnosis of brake pad wear in the absence of a brake pad thickness sensor.

Description

Technical Field

[0001] This invention belongs to the field of vehicle braking technology, specifically relating to a brake pad wear diagnosis method based on fuzzy neural networks. Background Technology

[0002] Brake pads play a crucial role in vehicle braking and determine driving safety. Vehicles achieve braking through friction between the brake pads, which leads to the brake pads becoming increasingly thin. When the thickness of the brake pads falls below a minimum threshold, braking failure can occur, posing a significant driving hazard. Therefore, it is essential to monitor and diagnose the wear condition of the vehicle's brake pads during use.

[0003] Due to limitations in installation conditions or costs, some existing vehicles do not have brake pad thickness sensors installed, making it impossible to monitor brake pad wear in real time. Therefore, how to monitor brake pad wear in the absence of thickness sensors is an urgent problem to be solved. Summary of the Invention

[0004] To address the existing technical problems, this invention provides a brake pad wear diagnosis method based on a fuzzy neural network, so that the wear condition of vehicle brake pads can be diagnosed even when a brake pad thickness sensor is not installed.

[0005] In a first aspect, the present invention provides a brake pad wear diagnosis method based on a fuzzy neural network, comprising:

[0006] The characteristic variable determining the wear thickness of the brake pad is used as the input variable;

[0007] Construct an initial brake pad wear diagnosis model based on a fuzzy neural network;

[0008] Training samples are acquired and input into the initial brake pad wear diagnosis model, and the final brake pad wear diagnosis model is determined by an adaptive particle swarm optimization algorithm.

[0009] Input the input variables into the final brake pad wear diagnosis model to calculate the brake pad wear thickness and remaining thickness;

[0010] Determine whether the remaining thickness of the brake pad is less than the preset minimum threshold for brake pad thickness.

[0011] Further improvements to this technical solution include the use of principal component analysis to extract characteristic variables, with the following input:

[0012] s'=[D,t,V0,V d [T, L, W, S, B];

[0013] Where s' is a parameter vector related to the wear degree of the brake pads, D is the brake pedal travel, t is the braking time, V0 is the initial braking velocity, and V d Where is the braking end speed, T is the braking ambient temperature, L is the braking load, W is the vehicle weight, S is the braking distance, and B is the number of braking cycles.

[0014] Further improvements to this technical solution include obtaining principal components from all feature variables based on a cumulative contribution rate greater than 85%, thereby extracting key features. The final extracted key feature vector is represented as s.

[0015] Further improvements to this technical solution include an initial diagnostic model comprising an input layer, a fuzzy rule layer, a normalization layer, and an output layer.

[0016] Further improvements to this technical solution include the following: the input layer is s(t), which represents the input vector of the brake pad wear diagnosis model at time t.

[0017] A further improvement to this technical solution is that the fuzzy rule layer consists of j neurons, and its output is represented as:

[0018]

[0019] Where, φ j s(t) is the output of the j-th neuron in the fuzzy rule layer at time t, and K is the dimension of s(t). i (t) represents the i-th input of the input layer, μ ij (t) represents the center of the j-th neuron corresponding to the i-th input in the fuzzy rule layer, σ ij (t) represents the width of the j-th neuron corresponding to the i-th input in the fuzzy rule layer; the value of j ranges from [1, 10], and the value of i ranges from [1, K].

[0020] A further improvement to this technical solution is that the number of neurons in the normalization layer is the same as the number of neurons in the fuzzy rule layer, and the output of the normalization layer is represented as:

[0021]

[0022] in, Let t be the output of the j-th neuron in the normalized layer at time t.

[0023] Further improvements to this technical solution include an output layer as follows:

[0024] y(t)=φ(t)w T (t);

[0025] Where y(t) is the difference in brake pad thickness before and after braking detected at time t, and φ(t) = [φ1(t), ..., φ10 [w(t)] is the output vector of the normalization layer of the fuzzy neural network, w(t) = [w1(t), ..., w2(t)]. 10 [w(t)] represents the weight vector between neurons in the output layer and the normalized layer, and w1(t) represents the weight vector between the first neuron in the normalized layer and the output layer. 10 (t) represents the weights between the tenth neuron in the normalized layer and the output layer.

[0026] Further improvements to this technical solution include determining the final brake pad wear diagnosis model using an adaptive particle swarm optimization algorithm, specifically including:

[0027] The center and width of the neurons in the fuzzy rule layer of the fuzzy neural network, as well as the weights between the neurons in the output layer and the normalization layer, are used as the position vectors of the particle swarm optimization algorithm; the objective of the particle swarm optimization algorithm is to minimize the absolute value of the error between the output value and the actual value of the fuzzy neural network.

[0028] Set the population size N for particle swarm optimization, with a value in the range [100, 500]; set the maximum number of iterations K for particle swarm optimization. max The value range is [100, 200]; the number of population iterations is k, and the value range is [1, K]. max ];

[0029] The population is initialized by randomly generating a population of N particles. The target value for each particle is calculated, and the optimal solution for each particle is obtained.

[0030] p n (1) = x n (1);

[0031] Where, p n (1) represents the optimal solution obtained by the nth particle in the first generation; x n (1) is the position vector of the nth particle in the first generation; the position vector corresponding to the particle with the smallest target value is taken as the global optimal solution g(1).

[0032] Further improvements to this technical solution include the following: the position and velocity update method for particle swarm optimization is as follows:

[0033]

[0034] x n (k+1)=x n (k)+v n (k+1);

[0035] Among them, v n (k+1) is the velocity vector of the nth particle in the (k+1)th generation, v n(k) is the velocity vector of the nth particle in the kth generation, x n (k+1) is the position vector of the nth particle in the (k+1)th generation, x n (k) is the position vector of the nth particle in the kth generation, p n (k) is the individual optimal solution of the nth particle in the kth generation, r1 is a random vector with a value range of [0, 1], r2 is a random vector of social cognition terms with a value range of [0, 1], and g(k) is the global optimal solution of the population in the kth generation.

[0036] b(k) is the inertial weight of the particle in the kth generation, and it is updated as follows:

[0037]

[0038] Where A(k) is the number of times the optimal solution of an individual in the population is updated;

[0039] c1(k) and c2(k) are the individual cognitive factor and social cognitive factor of the particle in the kth generation, respectively, and are updated as follows:

[0040]

[0041]

[0042] The beneficial effect of this invention is that it obtains the brake pad wear thickness and calculates the remaining thickness by using a brake pad wear prediction model based on a fuzzy neural network and an adaptive particle swarm optimization algorithm to optimize the model parameters, thereby realizing real-time monitoring and diagnosis of brake pad wear in the absence of a brake pad thickness sensor.

[0043] Furthermore, the design principle of this invention is reliable, the structure is simple, and it has a very wide range of application prospects. Attached Figure Description

[0044] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0045] Figure 1 This is a schematic flowchart of a method according to an embodiment of the present invention. Detailed Implementation

[0046] To enable those skilled in the art to better understand the technical solutions of this invention, the technical solutions of the embodiments of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this invention, and not all embodiments. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of this invention.

[0047] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention.

[0048] The brake pad wear diagnosis method based on fuzzy neural network provided in this embodiment of the invention is executed by computer equipment. Figure 1 This is a schematic flowchart illustrating a method according to an embodiment of the present invention. The order of the steps in this flowchart may be changed, and some steps may be omitted, depending on different requirements.

[0049] like Figure 1 As shown, the method includes:

[0050] Step 110: Determine the characteristic variable of brake pad wear thickness as the input variable;

[0051] Step 120: Construct an initial brake pad wear diagnosis model based on a fuzzy neural network;

[0052] Step 130: Obtain training samples and input them into the initial brake pad wear diagnosis model, and determine the final brake pad wear diagnosis model through the adaptive particle swarm optimization algorithm;

[0053] Step 140: Input the input variables into the final brake pad wear diagnosis model to calculate the brake pad wear thickness and remaining thickness;

[0054] Step 150: Determine whether the remaining thickness of the brake pad is less than the preset minimum threshold for brake pad thickness.

[0055] To facilitate understanding of the present invention, the following description further illustrates the brake pad wear diagnosis method based on fuzzy neural networks, using the principle of the present invention and the process of diagnosing brake pad wear based on fuzzy neural networks in the embodiments.

[0056] Specifically, the brake pad wear diagnosis method based on fuzzy neural networks includes:

[0057] S1. Determine the characteristic variable of brake pad wear thickness as the input variable.

[0058] Principal component analysis is used to extract feature variables. The input is:

[0059] s'=[D,t,V0,V d [T, L, W, S, B];

[0060] Where s' is a parameter vector related to the wear degree of the brake pads, D is the brake pedal travel, t is the braking time, V0 is the initial braking velocity, and V d Where is the braking end speed, T is the braking ambient temperature, L is the braking load, W is the vehicle weight, S is the braking distance, and B is the number of braking cycles.

[0061] S2. Construct an initial brake pad wear diagnosis model based on a fuzzy neural network.

[0062] The initial diagnostic model based on a fuzzy neural network consists of an input layer, a fuzzy rule layer, a normalization layer, and an output layer. The input layer is s(t), which represents the input vector of the brake pad wear diagnostic model at time t.

[0063] Specifically, the fuzzy rule layer consists of j neurons, and its output is represented as:

[0064]

[0065] Where, φ j s(t) is the output of the j-th neuron in the fuzzy rule layer at time t, and K is the dimension of s(t). i (t) represents the i-th input of the input layer, μ ij (t) represents the center of the j-th neuron corresponding to the i-th input in the fuzzy rule layer, σ ij (t) represents the width of the j-th neuron corresponding to the i-th input in the fuzzy rule layer; the value of j ranges from [1, 10], and the value of i ranges from [1, K].

[0066] The number of neurons in the normalization layer is the same as the number of neurons in the fuzzy rule layer. The output of the normalization layer is represented as:

[0067]

[0068] in, Let t be the output of the j-th neuron in the normalized layer at time t.

[0069] The output layer is:

[0070] y(t)=φ(t)w T (t);

[0071] Where y(t) is the difference in brake pad thickness before and after braking detected at time t, and φ(t) = [φ1(t), ..., φ 10 [w(t)] is the output vector of the normalization layer of the fuzzy neural network, w(t) = [w1(t), ..., w2(t)]. 10 [w(t)] represents the weight vector between neurons in the output layer and the normalized layer, and w1(t) represents the weight vector between the first neuron in the normalized layer and the output layer. 10 (t) represents the weights between the tenth neuron in the normalized layer and the output layer.

[0072] S3. The final brake pad wear diagnosis model is determined by an adaptive particle swarm optimization algorithm.

[0073] To improve the model's accuracy, an adaptive particle swarm optimization algorithm is used to optimize the model parameters. The parameters adjusted are the center and width of neurons in the fuzzy rule layer, as well as the connection weights between neurons in the normalized layer and the output neuron. The specific steps include:

[0074] S131. The center and width of the neurons in the fuzzy rule layer of the fuzzy neural network, as well as the weights between the neurons in the output layer and the normalization layer, are used as the position vectors of the particle swarm optimization algorithm; the objective of the particle swarm optimization algorithm is to minimize the absolute value of the error between the output value and the actual value of the fuzzy neural network.

[0075] S132. Set the population size N for particle swarm optimization, with a value range of [100, 500]; set the maximum number of iterations K for particle swarm optimization. max The value range is [100, 200]; the number of population iterations is k, and the value range is [1, K]. max ];

[0076] S133. Initialize the population by randomly generating a population of N particles. Solve for the target value for each particle to obtain the optimal solution for each individual particle.

[0077] p n (1) = x n (1);

[0078] Where, p n (1) represents the optimal solution obtained by the nth particle in the first generation; x n (1) is the position vector of the nth particle in the first generation; the position vector corresponding to the particle with the smallest target value is taken as the global optimal solution g(1).

[0079] The position and velocity update methods for particle swarm optimization are as follows:

[0080]

[0081] x n (k+1)=x n (k)+v n (k+1);

[0082] Among them, v n (k+1) is the velocity vector of the nth particle in the (k+1)th generation, v n (k) is the velocity vector of the nth particle in the kth generation, x n (k+1) is the position vector of the nth particle in the (k+1)th generation, x n (k) is the position vector of the nth particle in the kth generation, p n (k) is the individual optimal solution of the nth particle in the kth generation, r1 is a random vector with a value range of [0, 1], r2 is a random vector of social cognition terms with a value range of [0, 1], and g(k) is the global optimal solution of the population in the kth generation.

[0083] b(k) is the inertial weight of the particle in the kth generation, and it is updated as follows:

[0084]

[0085] Where A(k) is the number of times the optimal solution of an individual in the population is updated, and N is the number of braking times;

[0086] c1(k) and c2(k) are the individual cognitive factor and social cognitive factor of the particle in the kth generation, respectively, and are updated as follows:

[0087]

[0088]

[0089] The adaptive particle swarm optimization algorithm is used to adjust the model parameters, so as to effectively express the relationship between brake pad wear and feature variables.

[0090] S. Input the input variables into the final brake pad wear diagnosis model to calculate the brake pad wear thickness and remaining thickness.

[0091] Based on the initial thickness of the brake pads, after braking, the wear thickness of the brake pads at each braking time is calculated using the final brake pad wear diagnosis model. The remaining thickness of the brake pads is obtained by subtracting the wear thickness calculated each time from the initial thickness of the new brake pads. Finally, the brake pad wear is diagnosed by combining the preset minimum threshold of brake pad thickness. That is, this invention can obtain the brake pad wear thickness and calculate the remaining thickness by using a brake pad wear prediction model based on a fuzzy neural network and optimizing the model parameters using an adaptive particle swarm optimization algorithm. This enables real-time monitoring and diagnosis of the degree of brake pad wear without a brake pad thickness sensor.

[0092] Although the present invention has been described in detail with reference to the accompanying drawings and preferred embodiments, the invention is not limited thereto. Various equivalent modifications or substitutions can be made to the embodiments of the invention by those skilled in the art without departing from the spirit and essence of the invention, and such modifications or substitutions should all be within the scope of the invention. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the invention should also be covered within the protection scope of the invention. Therefore, the protection scope of the invention should be determined by the scope of the claims.

Claims

1. A brake pad wear diagnosis method based on fuzzy neural networks, characterized in that, include: The characteristic variable determining the wear thickness of the brake pad is used as the input variable; Construct an initial brake pad wear diagnosis model based on a fuzzy neural network; Training samples are acquired and input into the initial brake pad wear diagnosis model, and the final brake pad wear diagnosis model is determined by an adaptive particle swarm optimization algorithm. Input the input variables into the final brake pad wear diagnosis model to calculate the brake pad wear thickness and remaining thickness; Determine whether the remaining thickness of the brake pad is less than the preset minimum threshold for brake pad thickness; The initial diagnostic model consists of an input layer, a fuzzy rule layer, a normalization layer, and an output layer; The final brake pad wear diagnosis model was determined using an adaptive particle swarm optimization algorithm, including: The center and width of the neurons in the fuzzy rule layer of the fuzzy neural network, as well as the weights between the neurons in the output layer and the normalization layer, are used as the position vectors of the particle swarm optimization algorithm; the objective of the particle swarm optimization algorithm is to minimize the absolute value of the error between the output value and the actual value of the fuzzy neural network. The population is initialized by randomly generating a population of N particles. The target value for each particle is calculated, and the optimal solution for each particle is obtained. ； where p n (1) is the individual optimal solution obtained by the nth particle in the first generation; x n (1) is the position vector of the nth particle in the first generation; the position vector corresponding to the particle with the minimum target value is taken as the global optimal solution g(1); The position and velocity update methods for particle swarm optimization are as follows: ； ； Among them, v n (k+1) is the velocity vector of the nth particle in the (k+1)th generation, v n (k) is the velocity vector of the nth particle in the kth generation, x n (k+1) is the position vector of the nth particle in the (k+1)th generation, x n (k) is the position vector of the nth particle in the kth generation, p n (k) is the individual optimal solution of the nth particle in the kth generation, r1 is a random vector with a value range of [0, 1], r2 is a random vector of social cognition terms with a value range of [0, 1], and g(k) is the global optimal solution of the population in the kth generation. b(k) is the inertial weight of the particle in the kth generation, and it is updated as follows: ； Where A(k) is the number of times the optimal solution of an individual in the population is updated; c1(k) and c2(k) are the individual cognitive factor and social cognitive factor of the particle in the kth generation, respectively, and are updated as follows: ； 。 2. The brake pad wear diagnosis method based on fuzzy neural network according to claim 1, characterized in that, Principal component analysis is used to extract feature variables. The input is: ； Where s' is a parameter vector related to the wear degree of the brake pads, D is the brake pedal travel, t is the braking time, V0 is the initial braking velocity, and V d Where is the braking end speed, T is the braking ambient temperature, L is the braking load, W is the vehicle weight, S is the braking distance, and B is the number of braking cycles.

3. The brake pad wear diagnosis method based on fuzzy neural network according to claim 2, characterized in that, Among all feature variables, principal components are obtained based on a cumulative contribution rate greater than 85%, thus enabling the extraction of key features. The final extracted key feature vector is represented as s.

4. The brake pad wear diagnosis method based on fuzzy neural network according to claim 1, characterized in that, in, The input layer is s(t), which represents the input vector of the brake pad wear diagnosis model at time t.

5. The brake pad wear diagnosis method based on fuzzy neural network according to claim 4, characterized in that, The fuzzy rule layer consists of j neurons, and its output is represented as: ； in, Let be the output of the j-th neuron in the fuzzy rule layer at time t, and K be the dimension of s(t). i (t) represents the i-th input of the input layer, μ ij (t) represents the center of the j-th neuron corresponding to the i-th input in the fuzzy rule layer, σ ij (t) represents the width of the j-th neuron corresponding to the i-th input in the fuzzy rule layer; the value of j ranges from [1, 10], and the value of i ranges from [1, K].

6. The brake pad wear diagnosis method based on fuzzy neural network according to claim 5, characterized in that, The number of neurons in the normalization layer is the same as the number of neurons in the fuzzy rule layer. The output of the normalization layer is represented as: ； in, Let t be the output of the j-th neuron in the normalized layer at time t.

7. The brake pad wear diagnosis method based on fuzzy neural network according to claim 6, characterized in that, The output layer is: ； Where y(t) is the difference in brake pad thickness before and after braking detected at time t. This is the output vector of the normalization layer in a fuzzy neural network. Let w1(t) be the weight vector between the neurons in the output layer and the normalized layer, and w1(t) be the weight between the first neuron in the normalized layer and the output layer. 10 (t) represents the weights between the tenth neuron in the normalized layer and the output layer.

8. The brake pad wear diagnosis method based on fuzzy neural network according to claim 1, characterized in that, The final brake pad wear diagnostic model is determined using an adaptive particle swarm optimization algorithm, and also includes: Set the population size N for particle swarm optimization, with a value in the range [100, 500]; set the maximum number of iterations K for particle swarm optimization. max The value range is [100, 200]; the number of population iterations is k, and the value range is [1, K]. max ].

Images