Drum brake with external rotation center of brake shoe and control method thereof

Abstract

The invention relates to the technical field of brakes, in particular to a drum brake with an external brake shoe rotation center and a control method thereof.The drum brake comprises a back plate, and fixing columns are symmetrically installed at one end of the bottom of the outer side face of the back plate. Two connecting rods rotate to drive the two brake shoes to move relatively, the two brake shoes rotate outside the corresponding cylindrical blocks, so that the two brake shoes move relatively, the two brake pads are attached to the outer ring face of the brake drum, and the braking effect is achieved; the limiting block can be driven to be connected into the limiting groove when the limitation of the reset spring is relieved, so that the brake pad is fixedly mounted on the brake pad mounting plate, and the disassembly and assembly are convenient; therefore, the problems that a traditional drum brake is of a built-in type, the heat dissipation effect is poor, chippings generated during braking can be left in a brake drum, and a brake pad of the built-in drum brake is inconvenient to disassemble, assemble and replace are effectively solved.

Description

Technical Field

[0001] The present invention relates to the technical field of brakes, and specifically to an externally-mounted drum brake with the rotation center of the brake shoe and its control method. Background Technique

[0002] In the field of automotive braking technology, drum brakes have always occupied an important position with their unique structure and high braking efficiency. A drum brake mainly consists of parts such as a brake drum, brake shoes, brake wheel cylinders, return springs, and positioning pins. When the driver steps on the brake pedal, the brake fluid is transmitted through the braking system to the master cylinder of the drum brake. The master cylinder transfers the brake fluid to the brake wheel cylinders inside the brake drum through hydraulic action. The pistons in the brake wheel cylinders move outward under the hydraulic action, pushing the brake shoes to open, so that the friction linings on the brake shoes are pressed against the inner surface of the brake drum, thereby generating a frictional torque to achieve the deceleration or stop of the vehicle. The drum brake has a relatively simple structure and low manufacturing cost, enabling vehicle manufacturers to reduce production costs. Secondly, the braking ability of the drum brake is good. Especially when driving at high speed and in emergency braking, it can provide higher braking force and better braking stability. However, the traditional drum brake is an internal type with poor heat dissipation effect, and the debris generated by braking will remain in the brake drum and is prone to accumulate and block. In addition, it is inconvenient to disassemble, install, and replace the brake pads for the internal drum brake. Therefore, there are certain drawbacks.

[0003] In summary, the present invention solves the existing problems by designing an externally-mounted drum brake with the rotation center of the brake shoe. Summary of the Invention

[0004] The purpose of the present invention is to provide an externally-mounted drum brake with the rotation center of the brake shoe and its control method to solve the problems mentioned in the above background technique.

[0005] To achieve the above purpose, the present invention provides the following technical solutions:

[0006] An externally-mounted drum brake with the rotation center of the brake shoe includes a back plate. At one end of the bottom of the outer side of the back plate, fixed columns are symmetrically installed. On the fixed columns and on the outer side of the back plate, brake shoes are symmetrically installed. At one end of the outer side of each brake shoe, an arc-shaped sliding groove is opened. On the outer side of each brake shoe, several groups of support blocks are installed. At one end of the outer side of each brake shoe, a brake pad mounting plate is installed. On the inner side wall of the brake pad mounting plate, a brake pad is installed. At both ends of the outer arc surface of the brake pad, fixed rods are symmetrically installed. On the outer ring surface of the fixed rod, a limit groove is opened. At both ends of the outer arc surface of the brake pad mounting plate and respectively between two support blocks, fixed blocks are installed. On the outer side surface of the fixed block, a telescopic rod is installed. On the outer ring surface of the telescopic rod, a return spring is installed. The end of the telescopic rod away from the fixed block is installed with a limit block inside the limit groove;

[0007] On the outer side of the back plate and inside both arc-shaped chutes, limiting rods are installed. At both ends of the edge of the outer side of the back plate, fixing arc blocks are symmetrically installed. On the inner side wall of the fixing arc block, a hook spring is installed. At one end of the top of the outer side of the back plate, a rotating rod is installed. At one end of the rotating rod, a rotating block is installed. At both ends of the rotating block, connecting rods are installed. At the end of the connecting rod away from the rotating block, a connecting block is installed. On the outer side of the back plate and between the two brake shoes, a brake drum is installed. On the outer side of the brake drum, a mounting block is installed. On the outer side of the mounting block, a mounting hole is opened. On the outer side of the mounting block and outside the mounting hole, several groups of fixing holes are opened. On the outer side of the back plate and outside the brake drum, a protective cover is installed. On the outer side of the protective cover, several groups of heat dissipation holes are opened. On the outer ring surface of the protective cover, several groups of sewage discharge holes are equidistantly opened.

[0008] As a preferred solution of the present invention, the outer ring surface of the fixing column is rotationally connected to the inside of one end of the brake shoe, and the outer side of the back plate is slidably connected to the outer sides of the two brake shoes.

[0009] As a preferred solution of the present invention, the end of the fixing rod away from the brake pad penetrates through the brake pad mounting plate and extends to its outside. The outer side of the limiting block is slidably connected to the outside of the brake pad mounting plate, and the outside of the limiting block is slidably connected and closely fitted to the inside of the limiting groove.

[0010] As a preferred solution of the present invention, two limiting rods are provided and are respectively located inside the two arc-shaped chutes. One end of the limiting rod is fixedly connected to the back plate, and the outer ring surface of the limiting rod is slidably connected to the inner side wall of the arc-shaped chute.

[0011] As a preferred solution of the present invention, rings are installed on both the fixing arc block and the brake shoe, and the two ends of the hook spring are respectively sleeved and connected to the rings on the fixing arc block and the brake shoe.

[0012] As a preferred solution of the present invention, one end of the rotating rod penetrates through the back plate and is connected to the external vehicle braking system, and the end of the rotating rod away from the back plate is fixedly connected to the outer side of the rotating block.

[0013] As a preferred solution of the present invention, both ends of the rotating block are rotationally connected to the two connecting rods, and the end of the connecting rod away from the rotating block is rotationally connected to one end of the connecting block.

[0014] As a preferred solution of the present invention, the protective cover is fixedly connected to the outside of the back plate, the brake drum is fixedly connected to the external tire, and the debris generated by the friction between the brake pads and the brake drum can be discharged through multiple groups of sewage discharge holes.

[0015] The control method process of the drum brake with an externally placed rotation center of the brake shoe of the present invention includes: Step 1: Installation of the sensor system and data acquisition; Step 2: Modeling of the variational graph autoencoder VGAE; Step 3: Reinforcement learning control strategy; By introducing graph neural network, reinforcement learning and variational graph autoencoder algorithms, the braking process is optimized, and monitoring and control are carried out through the algorithms.

[0016] Compared with the prior art, the beneficial effects of the present invention are:

[0017] 1. In the present invention, through the designed drum brake with an externally placed rotation center of the brake shoe, two connecting rods rotate to drive two brake shoes to move relatively. The two brake shoes rotate outside the corresponding cylindrical blocks respectively, so that the two brake shoes move relatively to make the two brake pads fit the outer ring surface of the brake drum, thus achieving the braking effect. When the connecting rod passes through the brake pad mounting plate, the limit block is released, and the restriction of the return spring is released, which can drive the limit block to be inserted into the inside of the limit groove, thereby fixedly installing the brake pad on the brake pad mounting plate. The disassembly and assembly are convenient, thus effectively solving the problems of poor heat dissipation of the traditional built-in drum brake, the debris generated by braking remaining in the brake drum, and the inconvenience of disassembling, installing and replacing the brake pads of the built-in drum brake.

[0018] 2. Precise monitoring and control: Real-time data acquisition: Data is collected in real time through temperature, pressure and displacement sensors to ensure that the system is always in a monitored state. The sensor data is standardized and normalized to eliminate the magnitude differences of different sensor data, improve the consistency and comparability of the data, and extract high-dimensional features from the complex sensor data through the variational graph autoencoder (VGAE) to capture the potential relationships and patterns inside the system.

[0019] 3. Improve system performance: Optimize the control strategy: Through the reinforcement learning algorithm, the control strategy is dynamically updated to ensure that the system can achieve the best performance under various operating conditions. Based on the real-time feedback, the system can adaptively adjust the braking pressure, the position of the fixed rod and the position of the brake pads to optimize the braking effect and system stability.

[0020] 4. Enhance system stability and reliability: By monitoring multiple variables such as temperature, pressure and displacement, ensure that all parts of the system are in the best state and reduce the failure rate. The reward function designed based on the braking effect and system stability guides the reinforcement learning algorithm to optimize the control strategy to ensure the stable operation of the system.

[0021] 5. Improve efficiency and response speed: Through real-time data acquisition and processing, the system can quickly respond to changes and timely adjust the control parameters to avoid delay and error accumulation. The VGAE model is used for efficient feature extraction to reduce the calculation amount and improve the overall calculation efficiency of the system.

[0022] 6. Self-learning ability: The reinforcement learning algorithm enables the system to have self-learning ability, continuously optimize the control strategy according to historical data and current state, and improve the efficiency and effectiveness of the system's long-term operation. Through the interaction and feedback mechanism of reinforcement learning, the system can accumulate experience and gradually enhance its ability to handle various complex situations.

[0023] 7. Improve safety: By introducing intelligent algorithms, the system can more accurately monitor and control key parameters, reduce the risk of human operation errors, and improve the overall safety of the system. The intelligent algorithm can identify and warn of abnormal states, take timely measures to prevent potential dangers from occurring. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 is a schematic diagram of the overall structure of the present invention;

[0025] Figure 2 is the Figure 1 schematic diagram of the structural stratification of the present invention;

[0026] Figure 3 is the Figure 2 schematic diagram of a partial structure of the present invention;

[0027] Figure 4 is the Figure 3 schematic diagram of a partial structure of the present invention;

[0028] Figure 5 is the Figure 3 schematic diagram of the partial structural stratification of the present invention;

[0029] Figure 6 is the Figure 6 schematic diagram of a partial structure of the present invention;

[0030] Figure 7 is the Figure 6 schematic diagram of the partial structural stratification of the present invention.

[0031] Figure 8 : Flow chart of the intelligent algorithm introduced by the present invention

[0032] In the figure: 1. Backplane; 2. Fixed column; 3. Brake shoe; 301. Arc-shaped chute; 4. Support block; 5. Brake pad mounting plate; 501. Brake pad; 6. Fixed rod; 601. Limit groove; 7. Fixed block; 8. Telescopic rod; 9. Return spring; 10. Limit block; 11. Limit rod; 12. Fixed arc block; 13. Hook spring; 14. Rotating rod; 15. Rotating block; 16. Connecting rod; 17. Connecting block; 18. Brake drum; 19. Mounting block; 1901. Mounting hole; 1902. Fixing hole; 20. Protective cover; 2001. Heat dissipation hole; 2002. Sewage discharge hole. DETAILED DESCRIPTION OF THE INVENTION

[0033] The following will describe the technical solutions in the embodiments of the present invention in a clear and complete manner in combination with the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0034] To facilitate the understanding of the present invention, the present invention will be described more comprehensively below with reference to the relevant drawings. Several embodiments of the present invention are given. However, the present invention can be implemented in many different forms and is not limited to the embodiments described herein. On the contrary, the purpose of providing these embodiments is to make the disclosure of the present invention more thorough and comprehensive.

[0035] It should be noted that when an element is referred to as being "fixed to" another element, it can be directly on the other element or there may also be a middle element. When an element is considered to be "connected" to another element, it can be directly connected to the other element or there may be a middle element at the same time. The terms "vertical", "horizontal", "left", "right" and similar expressions used herein are only for the purpose of illustration.

[0036] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which the present invention belongs. The terms used herein in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The term "and / or" used herein includes any and all combinations of one or more of the related listed items.

[0037] Embodiment, please refer to Figures 1-6 , the present invention provides a technical solution:

[0038] A drum brake with an externally placed rotation center of the brake shoe, comprising a back plate 1. At one end of the bottom of the outer side of the back plate 1, fixing columns 2 are symmetrically installed. On the fixing columns 2 and on the outer side of the back plate 1, brake shoes 3 are symmetrically installed. At one end of the outer side of each brake shoe 3, an arc-shaped sliding groove 301 is formed. On the outer side of the brake shoes 3, several groups of support blocks 4 are installed. At one end of the outer side of the brake shoes 3, a brake pad mounting plate 5 is installed. On the inner side wall of the brake pad mounting plate 5, a brake pad 501 is installed. At both ends of the outer arc surface of the brake pad 501, fixing rods 6 are symmetrically installed. On the outer ring surface of the fixing rods 6, limit grooves 601 are formed. At both ends of the outer arc surface of the brake pad mounting plate 5 and respectively between two support blocks 4, fixing blocks 7 are installed. On the outer side surface of the fixing blocks 7, telescopic rods 8 are installed. On the outer ring surface of the telescopic rods 8, return springs 9 are installed. The end of the telescopic rod 8 away from the fixing block 7 and inside the limit groove 601, a limit block 10 is installed;

[0039] On the outer side of the backboard 1 and inside both of the two arc-shaped chutes 301, limiting rods 11 are installed. At both ends of the outer side edge of the backboard 1, fixed arc blocks 12 are symmetrically installed. On the inner side wall of the fixed arc block 12, a hook spring 13 is installed. At one end of the top of the outer side of the backboard 1, a rotating rod 14 is installed. At one end of the rotating rod 14, a rotating block 15 is installed. At both ends of the rotating block 15, connecting rods 16 are installed. At the end of the connecting rod 16 far from the rotating block 15, a connecting block 17 is installed. On the outer side of the backboard 1 and between the two brake shoes 3, a brake drum 18 is installed. On the outer side of the brake drum 18, a mounting block 19 is installed. On the outer side of the mounting block 19, a mounting hole 1901 is opened. On the outer side of the mounting block 19 and outside the mounting hole 1901, several groups of fixing holes 1902 are opened. On the outer side of the backboard 1 and outside the brake drum 18, a protective cover 20 is installed. On the outer side of the protective cover 20, several groups of heat dissipation holes 2001 are opened. On the outer ring surface of the protective cover 20, several groups of sewage discharge holes 2002 are equidistantly opened.

[0040] Specifically, referring to Figure 2 and Figure 3 , there is a rotational connection between the outer ring surface of the fixed column 2 and the inside of one end of the brake shoe 3, and there is a sliding connection between the outer side of the backboard 1 and the outer sides of the two brake shoes 3, thus ensuring that the two brake shoes 3 rotate outside the corresponding fixed column 2 respectively, so that the two brake shoes 3 move relatively, making the two brake pads 501 fit the outer ring surface of the brake drum 18, thereby achieving the braking effect.

[0041] Furthermore, two limiting rods 11 are provided and are respectively located inside the two arc-shaped chutes 301. One end of the limiting rod 11 is fixedly connected to the backboard 1, and there is a sliding connection between the outer ring surface of the limiting rod 11 and the inner side wall of the arc-shaped chute 301, thus ensuring that by setting the arc-shaped chute 301 to move along the limiting rod 11, the stability of the relative movement and reset of the two brake shoes 3 is increased.

[0042] Furthermore, rings are installed on both the fixed arc block 12 and the brake shoe 3, and both ends of the hook spring 13 are sleeved and connected with the rings on the fixed arc block 12 and the brake shoe 3 respectively, thus ensuring that after braking is completed, the two hook springs 13 will drive the corresponding brake shoes 3 to reset, so that the brake pads 501 on the two brake shoes 3 are far away from the outside of the brake drum 18, and the wheel can rotate normally.

[0043] Furthermore, one end of the rotating rod 14 penetrates through the backboard 1 and is connected to the external vehicle braking system, and the end of the rotating rod 14 far from the backboard 1 is fixedly connected to the outer side surface of the rotating block 15, thus ensuring that when braking is required, the user steps on the brake pedal, and the rotating rod 14 is driven to rotate synchronously with the rotating block 15 through the vehicle braking transmission system.

[0044] Furthermore, both ends of the rotating block 15 are rotatably connected to the two connecting rods 16, and one ends of the connecting rods 16 far from the rotating block 15 are respectively rotatably connected to one ends of the connecting blocks 17, so as to ensure that the rotation of the rotating block 15 drives the two connecting rods 16 to rotate synchronously, and the rotation of the two connecting rods 16 drives the two brake shoes 3 to move relatively.

[0045] Furthermore, the protective cover 20 is fixedly connected to the outside of the back plate 1, the brake drum 18 is fixedly connected to the external tire, and the debris generated by the friction between the brake pads 501 and the brake drum 18 can be discharged through multiple sewage discharge holes 2002, so as to ensure that the inside of the dust-proof cover can be cooled and the sewage can be discharged by providing the heat dissipation holes 2001 and the sewage discharge holes 2002.

[0046] Specifically, referring to Figure 5 , Figure 6 and Figure 7 , one end of the fixing rod 6 far from the brake pad 501 penetrates through the brake pad mounting plate 5 and extends to its outside. The outer side of the limiting block 10 is slidably connected to the outside of the brake pad mounting plate 5, and the outside of the limiting block 10 is slidably connected and closely fitted to the inside of the limiting groove 601. Thus, it is ensured that when the fixing rod 6 passes through the brake pad mounting plate 5 and the limiting block 10 is released, the restriction of the return spring 9 is released, which can drive the limiting block 10 to be inserted into the inside of the limiting groove 601, so as to fixedly mount the brake pad 501 on the brake pad mounting plate 5, and the disassembly and assembly are convenient.

[0047] Workflow of the present invention: When using the drum brake with an externally positioned rotation center of the brake shoe designed by this solution, during the use of this drum brake, first, when braking is required, the user steps on the brake pedal, which drives the rotating rod 14 to rotate through the vehicle brake transmission system, and the synchronous rotating block 15 rotates. The rotation of the rotating block 15 synchronously drives the two connecting rods 16 to rotate. The rotation of the two connecting rods 16 drives the two brake shoes 3 to move relatively. The two brake shoes 3 rotate outside the corresponding fixed columns 2 respectively, so that the two brake shoes 3 move relatively, and the two brake pads 501 are attached to the outer ring surface of the brake drum 18 to achieve the braking effect. By setting the arc-shaped chute 301 to move along the limiting rod 11, the stability of the relative movement and reset of the two brake shoes 3 is increased. When braking is completed, the external brake pedal is released and the rotating block 15 resets. The two hook springs 13 will drive the corresponding brake shoes 3 to reset, so that the brake pads 501 on the two brake shoes 3 are away from the outside of the brake drum 18, and the wheels can rotate normally. By moving the limiting block 10 to synchronously squeeze the telescopic rod 8 and the return spring 9, when the fixed rod 6 passes through the brake pad mounting plate 5 and the limiting block 10 is released, the restriction of the return spring 9 is released, and the limiting block 10 can be driven to access the inside of the limiting groove 601, so as to fixedly install the brake pad 501 on the brake pad mounting plate 5, which is convenient for disassembly and assembly. By setting a protective cover 20 outside the brake drum 18, the internal brake shoes 3 and other components are protected. The debris generated by the friction between the brake pads 501 and the brake drum 18 can be discharged through multiple sewage discharge holes 2002. Since the brake pads 501 are arranged outside the brake drum 18 and the protective cover 20 is provided with heat dissipation holes 2001 on the outside, the heat dissipation performance is better than that of the internal drum brake, and the heat dissipation effect is further enhanced by setting the heat dissipation holes 2001, and the practicability is wider.

[0048] As Figure 8 , for the control method of the drum brake with an externally positioned rotation center of the brake shoe of the present invention, when the braking system is adjusted, a graph neural network, reinforcement learning, and variational graph autoencoder algorithms are introduced to optimize this process of the braking system. Through the algorithms, precise monitoring and control can be carried out during actual operation; the system performance is improved, the stability and reliability of the system are enhanced, the efficiency and response speed are increased, and it has the ability of self-learning and the safety is improved. The specific process is as follows:

[0049] Step 1: Installation of the sensor system and data acquisition

[0050] Sensor installation

[0051] Temperature sensor: Collect temperature data T(t) and install it near the brake drum and brake pads; monitor the temperature change in real time;

[0052] Select a suitable position near the brake drum and brake pads for installation, ensure that the sensor is in close contact, use a thermocouple, a temperature sensor of the RTD (Resistance Temperature Detector) type, and calibrate it to ensure accuracy;

[0053] Pressure sensor: Collect pressure data P(t) and install it in the hydraulic system; Monitor the pressure of the brake fluid;

[0054] Select key positions in the hydraulic system, near the hydraulic pipeline or hydraulic cylinder for installation, use a strain gauge or a piezoelectric type pressure sensor, and calibrate it to ensure accuracy;

[0055] Displacement sensor: Collect displacement data D(t) and install it between the fixed rod and the brake pad mounting plate; Monitor the displacement of the fixed rod and the position change of the brake pads;

[0056] Select a suitable position between the fixed rod and the brake pad mounting plate for installation, use an LVDT (Linear Variable Differential Transformer), an optical displacement sensor, and calibrate it to ensure accuracy;

[0057] Data acquisition

[0058] Data transmission: Sensor data is collected in real time through the embedded control system STM32 microcontroller and transmitted to the central processing unit for processing and storage; The original data vector X is collected

[0059] X = T(t), P(t), D(t)

[0060] Connect the data of each sensor to the STM32 microcontroller, perform data conversion through the ADC (Analog-to-Digital Converter), and transmit the data to the central processing unit using serial communication (UART);

[0061] Data preprocessing

[0062] Perform standardization and normalization processing to eliminate the influence of differences in the magnitudes of data from different sensors;

[0063] Calculate the mean and standard deviation of each variable:

[0064]

[0065] Among them,

[0066] μ(T) is the mean of the temperature variable

[0067] μ(P) is the mean of the pressure variable

[0068] μ(D) is the mean of the displacement variable

[0069] μ(X) is the data mean;

[0070] σ(T) is the standard deviation of the temperature variable

[0071] σ(P) is the standard deviation of the pressure variable

[0072] σ(D) is the standard deviation of the displacement variable

[0073] σ(X) is the data standard deviation:

[0074] Normalization formula:

[0075]

[0076] X norm is the normalized data vector

[0077] X is the original data vector

[0078] μ(X) is the data mean;

[0079] σ(X) is the data standard deviation;

[0080] Feature extraction

[0081] Graph neural network:

[0082] Model: Model the braking system as a graph structure;

[0083] Nodes: Represent different parts of the braking system, brake shoes, fixed posts, brake pads;

[0084] Edges: Represent the connection relationships between nodes, and define the relationships between nodes and edges according to the physical structure of the braking system; Edges can represent the mechanical or hydraulic actions between them;

[0085] Variables: Node feature matrix H(0), edge feature matrix E

[0086] Initialize the node feature matrix H(0) and the edge feature matrix E:

[0087]

[0088] where,

[0089] represents the feature vector of the i-th node at the 0-th layer;

[0090] H (0) is the initial node feature matrix, representing the features of each node at the 0-th layer;

[0091] n is the number of nodes;

[0092] E is the initial edge feature matrix, representing the connection relationships between nodes;

[0093] e ijRepresents the features of the edge between node i and node j;

[0094] i and j represent the indices of the nodes;

[0095] Node feature extraction: Extract the features of each node to form a high-dimensional feature vector;

[0096] Variable representation: Node feature vector H (l+1)

[0097] Use the graph convolutional layer to extract node features:

[0098]

[0099] Among them,

[0100] H (l+1) Represents the node feature vector of the (l + 1)-th layer

[0101] Represents the feature vector of the j-th node in the l-th layer;

[0102] W (l) Represents the weight matrix of the l-th layer, used to transform node features;

[0103] σ represents the activation function, used to introduce non-linear transformation, ReLU function;

[0104] N(i) represents the set of neighbor nodes of node i;

[0105] N(j) represents the set of neighbor nodes of node j;

[0106] |N(i)| represents the number of neighbor nodes of node i;

[0107] |N(j)| represents the number of neighbor nodes of node j;

[0108] In the variational graph autoencoder (VGAE) modeling and reinforcement learning control strategy, the installation of the sensor system and data acquisition through the data obtained by the sensor can provide the necessary input for the VGAE model and provide real-time feedback for the reinforcement learning control strategy;

[0109] Step 2: Variational graph autoencoder (VGAE) modeling

[0110] Encoder: Use the graph neural network (GNN) as the encoder to encode the input graph structure data into a low-dimensional latent variable representation; Through multi-layer graph convolutional operations, the GNN can effectively extract the complex relationships between nodes and their neighbor nodes, thereby obtaining global information;

[0111] Input data: Graph-structured data X, including node feature matrix and edge feature matrix; including temperature T(t), pressure P(t), and displacement D(t) information of components such as brake shoes, fixed columns, and brake pads; Edge features: including connection relationships, distances, and connection strengths between nodes.

[0112] Output latent variable: The encoder outputs the latent variable z, representing the latent features of the graph-structured data, including features related to the health of the brake, including whether there is abnormal wear and abnormal installation of components.

[0113] Formula:

[0114]

[0115] Among them,

[0116] q(z|X): The posterior distribution of the latent variable z;

[0117] q: Represents the posterior distribution;

[0118] z: Latent variable, representing the latent features of the graph-structured data;

[0119] X: Input graph-structured data, including node feature matrix and edge feature matrix;

[0120] N(z|μ(X,diag(σ 2 X)): The normal distribution of the latent variable z;

[0121] N: Represents the normal distribution;

[0122] μ(X): Mean function, calculated by the GNN encoder, representing the mean of the input data;

[0123] μ: Mean, reflecting the central position of the data;

[0124] X: Input graph-structured data;

[0125] diag(σ 2 X): Variance diagonal matrix, calculated by the GNN encoder, representing the variance of the input data;

[0126] σ 2 : Variance, reflecting the degree of dispersion of the data;

[0127] diag: Diagonal matrix, placing the variance values on the diagonal;

[0128] Decoder

[0129] Model: Use the decoder to decode the low-dimensional latent variable z into the original graph-structured data X

[0130] Output data: The decoder outputs the reconstructed graph-structured data X

[0131] Formula:

[0132] p(X|z) = Bernoulli(X|σ(zW T ))

[0133] p(X∣z) is the conditional distribution of the reconstructed data X;

[0134] p: represents the conditional distribution;

[0135] X: the reconstructed graph structure data;

[0136] z: the latent variable, representing the latent features of the graph structure data;

[0137] Bernoulli represents the Bernoulli distribution;

[0138] (zW T ) : the activation function sigmoid function is applied to the product of the latent variable and the weight matrix;

[0139] σ: the activation function, using the sigmoid function to limit the output value between 0 and 1;

[0140] z: the latent variable;

[0141] W is the weight matrix of the decoder;

[0142] Variational inference: Using variational inference to optimize the model parameters, maximizing the variational lower bound (ELBO) to improve the reconstruction ability and generalization performance of the model;

[0143] Formula:

[0144] L ELBO = E q(z|X) [logp(X|z)]-KL(q(z|X)||p(z))

[0145] L ELBO is the variational lower bound, used to optimize the model parameters

[0146] L: represents the objective function;

[0147] ELBO: the variational lower bound (Evidence Lower Bound), used to measure the reconstruction ability and

[0148] generalization performance;

[0149] E q(z|X) [logp（X|z）] is the expectation of the reconstruction loss;

[0150] E: the expected value, representing the estimation of the average value of a random variable;

[0151] q(z∣X): The posterior distribution of the latent variable z;

[0152] [logp(X|z)]: The log-likelihood of the reconstructed data X;

[0153] KL(q(z|X)||p(z)) is the KL divergence between the posterior distribution q(z|X) and the prior distribution p(z);

[0154] KL: Kullback-Leibler divergence, which is used to measure the difference between two probability distributions;

[0155] q(z∣X): The posterior distribution of the latent variable z;

[0156] p(z): The prior distribution of the latent variable z;

[0157] Step 3: Reinforcement learning control strategy

[0158] State space: Describes the current state of the braking system, including the temperature, pressure, and displacement data collected by sensors and the high-dimensional feature vector extracted by the VGAE model;

[0159] Temperature T(t): The real-time temperature of each component in the braking system; The data collected by the temperature sensor is used to monitor and adjust the system temperature;

[0160] Pressure P(t): The real-time pressure of the brake fluid; The data provided by the pressure sensor ensures that the hydraulic system is in the best working state;

[0161] Displacement D(t): The position change of the fixed rod and the brake pad; The data recorded by the displacement sensor is used to adjust the position of the braking system components;

[0162] High-dimensional feature vector z: The latent features of the graph structure data extracted by the VGAE model; A low-dimensional representation containing the complex internal relationships of the system, including abnormal wear and abnormal installation of components here;

[0163] Specific judgment is as follows:

[0164] 1. The relationship between temperature T(t) and abnormal wear, abnormal installation

[0165] Abnormal wear: When the brake is working, heat is generated due to friction. If the contact between the brake shoe and the brake drum is uneven or there are foreign objects, it will cause excessive friction, temperature rise, and then uneven wear. For example, local overheating may cause the surfaces of the brake shoe and the brake drum to deform, further aggravating the wear and forming a "hot spot" area. Abnormal wear usually shows a rapid increase in temperature and may be uneven.

[0166] Numerical case:

[0167] Normal wear: Under normal working conditions, the temperature during the operation of the brake is generally between 80°C and 120°C.

[0168] Abnormal wear: If abnormal wear occurs in a certain part of the brake, the temperature may rise rapidly. For example, the local temperature can reach 150°C - 300°C. At this time, melting or uneven wear may occur on the contact surface between the brake shoe and the brake drum, affecting the braking performance.

[0169] Abnormal installation: If the brake is not installed correctly, it may cause poor contact between the contact surface of the brake shoe and the brake drum, resulting in excessive friction or premature wear. Improper installation usually leads to uneven temperature distribution. For example, if the brake shoe cannot fully contact the brake drum, local overheating may occur.

[0170] Numerical case:

[0171] Normal installation: Under normal circumstances, the temperature change of the brake is stable, and the temperature change curves of each component are roughly similar.

[0172] Abnormal installation: If the installation angle of the brake shoe is improper, the temperature may be uneven, and the temperature of some parts may be too high, 170°C - 200°C, while other parts may remain at a lower temperature.

[0173] 2. Relationship between pressure P(t) and abnormal wear, abnormal installation

[0174] Abnormal wear: The pressure of the brake mainly comes from the hydraulic system. When the brake hydraulic pressure is abnormal, it may indicate uneven contact force between the brake shoe and the brake drum. Abnormal wear usually leads to an increase or decrease in the contact area between the brake shoe and the brake drum, thus affecting the braking pressure. If the wear is uneven, the hydraulic system may require a higher pressure to achieve the same braking effect.

[0175] Numerical case:

[0176] Normal wear: During normal braking, the working pressure of the hydraulic system is usually between 50 - 250 kPa.

[0177] Abnormal wear: If severe local wear occurs, the braking pressure may increase to 300 - 1000 kPa to compensate for the reduction in braking effect caused by wear.

[0178] Abnormal installation: If the brake is installed improperly, it may cause the brake shoes to apply pressure unevenly. For example, if the brake shoes deviate from the correct position, it may lead to uneven pressure distribution in the hydraulic system, and thus uneven braking force.

[0179] Numerical case:

[0180] Normal installation: Under standard conditions, the readings of the pressure sensor should be uniform, and the pressure varies within the range of 50 - 250 kPa.

[0181] Abnormal installation: If the brake shoes are not installed parallel, extreme pressure variations may occur. The pressure in some parts may rise to 300 - 1700 kPa, resulting in local overload.

[0182] 3. Relationship between displacement D(t) and abnormal wear, abnormal installation (which can be judged by the position changes of the fixed rod and brake pads)

[0183] Abnormal wear: The displacement of the brake shoes is closely related to the frictional force applied during the braking process. Under normal circumstances, the brake shoes will uniformly push the brake drum, resulting in uniform displacement changes. However, if the brake shoes experience abnormal wear (such as local wear), it may lead to non-uniform displacement changes. The more severely worn parts may not function properly, resulting in a mismatch between the displacement change and the braking effect.

[0184] Numerical case:

[0185] Normal wear: The change in displacement is usually stable, and the position changes of the fixed rod and brake pads are 0.5 mm - 1 mm.

[0186] Abnormal wear: If the brake shoes experience uneven wear, the displacement of the fixed rod and brake pads may be abnormal. For example, the local displacement may increase to 6 mm - 8 mm, while the displacement in other areas is smaller, showing uneven pressing force.

[0187] Abnormal installation: If the brake is not installed correctly (such as the fixed rod and brake pads not being properly aligned), it may lead to non-uniform displacement of the brake shoes during the braking process, thereby affecting the braking performance. Improper installation may cause some components not to come into correct contact during braking, resulting in the brake slipping or making premature contact, thereby affecting the displacement and braking effect.

[0188] Numerical case:

[0189] Normal installation: Under normal circumstances, the position changes of the fixed rod and brake pads should be uniform, usually within the range of 0.5 mm - 1 mm.

[0190] Abnormal installation: If installed improperly, the displacement of some components may be abnormal. For example, the displacement increases to 2 mm - 5 mm, affecting the normal operation of the brake.

[0191] State space variable: s = (T(t), P(t), D(t), z)

[0192] Action space: Executable control actions used to adjust various parameters of the braking system to optimize performance;

[0193] Adjust the braking pressure: change the pressure of the brake fluid; increase or decrease the pressure by adjusting the pressure valve;

[0194] Move the fixed rod: adjust the position of the fixed rod; change the position of the fixed rod through a motor or a hydraulic system;

[0195] Adjust the position of the brake pads: change the relative position of the brake pads; move the brake pads through a robotic arm or other device;

[0196] Action space variable: a = [a 1 , a 2 ,..., a n

[0197] where a i represents the i-th specific control action, increasing pressure, decreasing displacement;

[0198] Reward function: used to evaluate the effect of each action of the braking system and guide the reinforcement learning algorithm to optimize the control strategy; the reward function is designed based on the braking effect and system stability;

[0199] Formula:

[0200] r = -(T(t) - T opt ) 2 -(P(t) - P opt ) 2 -(D(t) - D opt ) 2

[0201] Parameter meaning:

[0202] r: reward value, reflecting the effect after performing a certain action in the current state;

[0203] T(t): temperature at the current moment;

[0204] T opt : optimal temperature value, the reference temperature of the system in the best working state;

[0205] P(t): pressure at the current moment;

[0206] P opt : optimal pressure value, ensuring that the brake fluid works within the best pressure range;

[0207] D(t): displacement at the current moment;

[0208] D opt : optimal displacement value, ensuring that the positions of the fixed rod and the brake pads are in the best state;

[0209] ​Policy Update: Through the Q-learning reinforcement learning algorithm, the control policy is updated through interaction and feedback to optimize the performance of the braking system;

[0210] Q-learning Update Formula

[0211] Q(s,a) ← Q(s,a) + α t [r + γmax a `Q(s`,a`)]] + βΔQ(s,a)

[0212] Parameter Meanings:

[0213] Q(s,a): The Q-value of performing action a in state s, representing the expected return after taking a specific action in a specific state;

[0214] α t : The learning rate set at time t, controlling the speed of Q-value update, with the value range 0 ≤ α ≤ 10

[0215] r: The current reward value, reflecting the immediate return brought by the current action;

[0216] γ: The discount factor, measuring the importance of future rewards, with the value range 0 ≤ γ ≤ 10

[0217] max a `Q(s`,a`): The maximum Q-value among all possible actions a` in the new state s`, representing the best expected return in the new state;

[0218] s`: The new state after performing action a;

[0219] β: The momentum term coefficient, helping to smooth the Q-value update and avoid drastic fluctuations, with the value range 0 ≤ β ≤ 10

[0220] ΔQ(s,a): The increment of the previous Q-value update, used to calculate the momentum term.

[0221] Numerical Case Analysis

[0222] Step 1: Installation of the Sensor System and Data Acquisition

[0223] Temperature Sensor

[0224] Location: Near the brake drum and brake pads

[0225] Type: Thermocouple Temperature Sensor

[0226] Reading: 75°C

[0227] Calibrated Accuracy: ±0.5°C

[0228] Pressure Sensor

[0229] Location: Hydraulic pipeline

[0230] Type: Piezoelectric pressure sensor

[0231] Reading: 200 kPa

[0232] Accuracy after calibration: ±2 kPa

[0233] Displacement sensor

[0234] Location: Between the fixed rod and the brake pad mounting plate

[0235] Type: LVDT displacement sensor

[0236] Reading: 5 mm

[0237] Accuracy after calibration: ±0.1 mm

[0238] Data acquisition

[0239] Transmission: Data conversion is performed through the ADC of the STM32 microcontroller and transmitted to the central processing unit using UART.

[0240] Data preprocessing

[0241] Mean temperature (μ T ): 72.8 °C

[0242] Mean pressure (μ P ): 202 kPa

[0243] Mean displacement (μ D ): 4.7 mm

[0244] Standard deviation of temperature (σ T ): 1.72 °C

[0245] Standard deviation of pressure (σ P ): 5.48 kPa

[0246] Standard deviation of displacement (σ D ): 0.42 mm

[0247] Data after standardization:

[0248] Temperature: Pressure: Displacement: Step 2: Modeling with Variational Graph Autoencoder (VGAE)

[0249] Node feature matrix (H (0) )

[0250]

[0251] Edge feature matrix (E)

[0252]

[0253] Node feature extraction (after one layer of graph convolution)

[0254] Assume that the weight matrix W and the activation function σ are ReLU.

[0255] The encoder outputs the latent variable (z)

[0256] Assume that the mean μ(X) and the variance diagonal matrix σ 2 (X) is

[0257]

[0258] Step 3: Reinforcement learning control strategy

[0259] State space

[0260] Temperature T(t) = 75 °C

[0261] Pressure P(t) = 200 kPa

[0262] Displacement D(t) = 5 mm

[0263] High-dimensional feature vector z = [0.1, 0.5, -0.3]

[0264] Action space

[0265] Increase braking pressure: +10 kPa

[0266] Decrease the displacement of the fixed rod: -1 mm

[0267] Adjust the position of the brake pad: +2 mm

[0268] Reward function

[0269] Optimal state: T opt = 70 °C, P opt = 210 kPa, D opt = 4 mm

[0270] Calculation of the current state reward value:

[0271] r = -(75 - 70)^2 - (200 - 210)^2 - (5 - 4)^2 = -25 - 100 - 1 = -126

[0272] Q-learning update formula

[0273] Current Q value Q(s, a) = 50

[0274] Learning rate α = 0.1

[0275] Discount factor γ = 0.9

[0276] The maximum Q value max in the new state a `Q(s`,a`) = 60

[0277] Update formula:

[0278] Q(s,a) ← 50 + 0.1[-126 + 0.9×60 - 50]

[0279] Q(s,a) ← 50 + 0.1[-126 + 54 - 50]

[0280] Q(s,a) ← 50 + 0.1[-122]

[0281] Q(s,a) ← 50 - 12.2 = 37.8

[0282] Although embodiments of the present invention have been shown and described, it will be understood by those of ordinary skill in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A drum brake with an external brake shoe rotation center, comprising a back plate (1), characterized in that: A fixing column (2) is symmetrically mounted on one end of the bottom of the outer side surface of the back plate (1); a brake shoe (3) is symmetrically mounted on the fixing column (2) and located on the outer side surface of the back plate (1); an arc-shaped slide groove (301) is provided on one end of the outer side surface of the brake shoe (3); a plurality of groups of support blocks (4) are mounted on the outer side surface of the brake shoe (3); a brake pad mounting plate (5) is mounted on one end of the outer side surface of the brake shoe (3); a brake pad (501) is mounted on the inner side wall of the brake pad mounting plate (5); and the brake pad (50 1), fixed rods (6) are symmetrically mounted at both ends of the outer arc surface, a limiting groove (601) is arranged on the outer ring surface of the fixed rod (6), fixed blocks (7) are mounted at both ends of the outer arc surface of the brake pad mounting plate (5) and are respectively located between the two support blocks (4), a telescopic rod (8) is mounted on the outer side surface of the fixed block (7), a return spring (9) is mounted on the outer ring surface of the telescopic rod (8), and a limiting block (10) is mounted on one end of the telescopic rod (8) away from the fixed block (7) and located inside the limiting groove (601).

2. The brake shoe rotation center external drum brake according to claim 1, characterized in that: Limit rods (11) are installed on the outer side of the back plate (1) and inside the two arc-shaped sliding grooves (301); fixed arc blocks (12) are symmetrically installed at both ends of the edge of the outer side of the back plate (1); a hook spring (13) is installed on the inner wall of the fixed arc block (12); a rotating rod (14) is installed at one end of the top of the outer side of the back plate (1); a rotating block (15) is installed at one end of the rotating rod (14); connecting rods (16) are installed at both ends of the rotating block (15); a connecting block (17) is installed at the end of the connecting rod (16) away from the rotating block (15); and the outer side of the back plate (1) and the inner wall of the fixed arc block (12) are symmetrically installed at both ends of the outer side of the back plate (1). A brake drum (18) is installed between the two brake shoes (3), a mounting block (19) is installed on the outer side surface of the brake drum (18), a mounting hole (1901) is opened on the outer side surface of the mounting block (19), a plurality of fixing holes (1902) are opened on the outer side surface of the mounting block (19) and located outside the mounting hole (1901), a protective cover (20) is installed on the outer side surface of the back plate (1) and located outside the brake drum (18), a plurality of heat dissipation holes (2001) are opened on the outer side surface of the protective cover (20), and a plurality of drainage holes (2002) are opened at equal intervals on the outer annular surface of the protective cover (20).

3. The brake shoe rotation center external drum brake according to claim 1, characterized in that: The outer annular surface of the fixing column (2) is rotationally connected to the interior of one end of the brake shoe (3), and the outer side surface of the back plate (1) is slidingly connected to the outer side surfaces of the two brake shoes (3).

4. The brake shoe rotation center external drum brake according to claim 1, characterized in that: One end of the fixing rod (6) away from the brake pad (501) passes through the brake pad mounting plate (5) and then extends to the outside thereof; the outer side surface of the limiting block (10) is slidably connected to the outside of the brake pad mounting plate (5), and the outside of the limiting block (10) is slidably connected to the inside of the limiting groove (601) and fits tightly.

5. The brake shoe rotation center external drum brake according to claim 1, characterized in that: The limiting rod (11) is provided with two portions respectively located inside two arc-shaped sliding grooves (301); one end of the limiting rod (11) is fixedly connected to the back plate (1), and the outer ring surface of the limiting rod (11) is slidably connected to the inner side wall of the arc-shaped sliding groove (301).

6. The brake shoe rotation center external drum brake according to claim 1, characterized in that: The fixed arc block (12) and the brake shoe (3) are both provided with circular rings, and the two ends of the hook spring (13) are respectively sleeved and connected between the circular rings on the fixed arc block (12) and the brake shoe (3).

7. The brake shoe rotation center external drum brake according to claim 1, characterized in that: One end of the rotating rod (14) passes through the back plate (1) and is connected to an external vehicle brake system, and one end of the rotating rod (14) away from the back plate (1) is fixedly connected to the outer side surface of the rotating block (15).

8. The brake shoe rotation center external drum brake according to claim 1, characterized in that: Both ends of the rotating block (15) are rotatably connected to two connecting rods (16), and one end of the connecting rod (16) away from the rotating block (15) is rotatably connected to one end of the connecting block (17).

9. The brake shoe rotation center external drum brake according to claim 1, characterized in that: The protective cover (20) is fixedly connected to the outside of the back plate (1), the brake drum (18) is fixedly connected to the external tire, and the debris generated by the friction between the brake pad (501) and the brake drum (18) can be discharged through multiple sets of drainage holes (2002).

10. The control method of a drum brake with an external brake shoe rotation center according to claim 1, characterized in that: Graph neural network, reinforcement learning and variational graph autoencoder algorithm are introduced to optimize the braking process, and the algorithm is used for monitoring and control. The specific process is as follows: Step 1: Sensor system installation and data collection Sensor Installation Temperature sensor: collects temperature data T(t) and is installed near the brake drum and brake pad; monitors temperature changes in real time; Choose a suitable location near the brake drum and brake pad for installation, ensure the sensor is in close contact, use thermocouples, resistance temperature detectors (RTD) type temperature sensors, and calibrate them to ensure accuracy; Pressure sensor: collects pressure data P(t) and is installed in the hydraulic system; monitors the pressure of the brake fluid; Select the key position of the hydraulic system, install it near the hydraulic pipeline or hydraulic cylinder, use strain gauge or piezoelectric type pressure sensor, and calibrate it to ensure accuracy; Displacement sensor: collects displacement data D(t) and is installed between the fixed rod and the brake pad mounting plate; monitors the displacement of the fixed rod and the position change of the brake pad; Select the appropriate position between the fixing rod and the brake pad mounting plate for installation, use the linear variable differential transformer LVDT, optical displacement sensor, and calibrate to ensure accuracy; Data collection Data transmission: Sensor data is collected in real time by the embedded control system STM32 microcontroller and transmitted to the central processing unit for processing and storage; the raw data vector X is collected: X=T(t),P(t),D(t) The data from each sensor is connected to the STM32 microcontroller, the data is converted through the analog-to-digital converter ADC, and the data is transmitted to the central processing unit using the serial communication UART; Data preprocessing Standardization and normalization processing to eliminate the impact of differences in the magnitude of data from different sensors; Compute the mean and standard deviation of each variable: in, μ(T) is the mean of the temperature variable; μ(P) is the mean of the pressure variable; μ(D) is the mean of the displacement variable; μ(X) is the data mean; σ(T) is the standard deviation of the temperature variable; σ(P) is the standard deviation of the pressure variable; σ(D) is the standard deviation of the displacement variable; σ(X) is the data standard deviation; Standardization formula: X is the normalized data vector; X is the original data vector; μ(X) is the data vector. Mean; Feature extraction Graph Neural Networks: Model: Model the braking system as a graph structure; Nodes: represent different parts of the brake system, brake shoes, fixed columns, brake pads; Edge: represents the connection between nodes. The relationship between nodes and edges is defined according to the physical structure of the braking system. Variables: node feature matrix H(0), edge feature matrix E; Initialize the node feature matrix H(0) and edge feature matrix E: And=[and ij ] in, represents the feature vector of the i-th node in layer 0; H (0) is the initial node feature matrix, which represents the features of each node at layer 0; n is the number of nodes; E is the initial edge feature matrix, which represents the connection relationship between nodes; e ij Represents the features of the edge between node i and node j; i and j represent the index of the node; node feature extraction: extract the features of each node to form a high-dimensional feature vector; variable representation: node feature vector H(l+1); Use graph convolutional layers to extract node features: Among them, H (l+1) Represents the node feature vector of the l+1th layer; represents the feature vector of the jth node in the lth layer; W (l) represents the weight matrix of the lth layer, which is used to transform node features; σ represents the activation function, which is used to introduce nonlinear transformation, ReLU function; N(i) represents the set of neighbor nodes of node i; N(j) represents the set of neighbor nodes of node j; |N(i)| represents the number of neighbor nodes of node i; |N(j)| represents the number of neighbor nodes of node j; In the variational graph autoencoder VGAE modeling and reinforcement learning control strategy, the installation of the sensor system and data acquisition can provide the necessary input for the VGAE model through the data obtained by the sensor, and provide real-time feedback for the reinforcement learning control strategy; Step 2: Variational Graph Autoencoder VGAE Modeling Encoder: Use graph neural network GNN as encoder to encode the input graph structure data into low-dimensional latent variable representation; through multi-layer graph convolution operations, GNN can effectively extract the complex relationship between nodes and their neighboring nodes, thereby obtaining global information; Input data: graph structure data X, including node feature matrix and edge feature matrix; including temperature T(t), pressure P(t) and displacement D(t) information of brake shoes, fixed columns, brake pads and other components; edge features: including connection relationship, distance and connection strength between nodes; Output latent variables: The encoder outputs latent variables z, which represent the potential features of the graph structure data and contain features related to brake health, including whether the components are abnormally worn and installed abnormally; formula: Where, q(z|X): posterior distribution of latent variable z; q: represents posterior distribution; z: latent variable, representing the potential features of graph structure data; X: input graph structure data, including node feature matrix and edge feature matrix; N(z|μ(X), diag(σ 2 (X))): Normal distribution of latent variable z; N: represents normal distribution; μ(X): ​​mean function, calculated by GNN encoder, represents the mean of input data; diag(σ 2 (X)): variance diagonal matrix, calculated by the GNN encoder, representing the variance of the input data; σ 2 : Variance, reflecting the degree of dispersion of data; Decoder Model: Use the decoder to decode the low-dimensional latent variable z into the original graph structure data Output data: The decoder outputs the reconstructed graph structure data formula: p(X|z)=Bernoulli(X|σ(zW T )) p(X|z) is the conditional distribution of the reconstructed data X; p: represents the conditional distribution; X: the reconstructed graph structure data; z: hidden variable, representing the potential characteristics of the graph structure data; Bernoulli represents the Bernoulli distribution; σ(zW T ): Activation function sigmoid function applied to the product of latent variables and weight matrix; σ: Activation function, using sigmoid function to limit the output value between 0 and 1; W is the weight matrix of the decoder; Variational inference: Use variational inference to optimize model parameters, maximize the variational lower bound ELBO, and improve the model's reconstruction ability and generalization performance; formula: L ELBO =E q(z|X) [log p(X|z)]-KL(q(z|X)||p(z)) Among them, L ELBO is the variational lower bound, used to optimize the model parameters; L: represents the objective function; ELBO: variational lower bound, used to measure the reconstruction ability and generalization performance of the model; E q(z|X) [log p(X|z)] is the expectation of the reconstruction loss; E: expected value, which means the estimation of the mean value of the random variable; q(z|X): the posterior distribution of the latent variable z; [log p(X|z)]: the log likelihood of the reconstructed data X; KL(q(z|X)||p(z)) is the KL divergence between the posterior distribution q(z|X) and the prior distribution p(z); KL: Kullback-Leibler divergence, which is used to measure the difference between two probability distributions; q(z|X): the posterior distribution of the latent variable z; p(z): the prior distribution of the latent variable z; Step 3: Reinforcement Learning Control Strategy State space: describes the current state of the braking system, including temperature, pressure and displacement data collected by sensors and high-dimensional feature vectors extracted by the VGAE model; Temperature T(t): real-time temperature of each component in the brake system; data collected by the temperature sensor, used to monitor and adjust the system temperature; Pressure P(t): real-time pressure of brake fluid; data provided by the pressure sensor to ensure that the hydraulic system is in the best working condition; Displacement D(t): change in position of the fixing rod and brake pad; data recorded by the displacement sensor and used to adjust the position of the brake system components; High-dimensional feature vector z: latent features of graph-structured data extracted by the VGAE model; contains low-dimensional representations of complex relationships within the system, including abnormal wear and abnormal installation of components; State space variables: s = (T(t), P(t), D(t), z) Action space: executable control actions used to adjust various parameters of the braking system to optimize performance; Adjust brake pressure: change the pressure of the brake fluid; increase or decrease pressure by adjusting the pressure valve; Move the fixed rod: adjust the position of the fixed rod; change the position of the fixed rod through an electric motor or a hydraulic system; Adjust the brake pad position: change the relative position of the brake pad; move the brake pad by a mechanical arm or other device; Action space variables: a=[a1,a2,…,a n ] Among them, a i represents the i-th specific control action, increasing pressure and reducing displacement; Reward function: used to evaluate the effect of each action of the braking system and guide the reinforcement learning algorithm to optimize the control strategy; the reward function is designed based on the braking effect and system stability; formula: r=-(T(t)-T opt ) 2 -(P(t)-P opt ) 2 -(D(t)-D opt ) 2 Parameter meaning: r: reward value, reflecting the effect of performing an action in the current state; T(t): temperature at the current moment; T opt : Optimal temperature value, the reference temperature of the system in the best working state; P(t): pressure at the current moment; P opt : optimal pressure value, ensuring that the brake fluid works within the optimal pressure range; D(t): displacement at the current moment; D opt : Optimal displacement value, ensuring the position of the fixing rod and the brake pad is in the best state; Strategy update: Through the reinforcement learning algorithm Q-learning, the control strategy is updated through interaction and feedback to optimize the performance of the braking system; Q-learning update formula Q(s,a)←Q(s,a)+α t [r+γmax a `Q(s`,a`)]+βΔQ(s,a) Parameter meaning: Q(s,a): The Q value of executing action a in state s, which represents the expected benefit after taking a specific action in a specific state; α t : The learning rate set at time t controls the speed of Q value update, and the value range is 0≤α≤10; r: The current reward value, reflecting the immediate benefit brought by the current action; γ: The discount factor, which measures the importance of future rewards, and the value range is 0≤γ≤10; max a `Q(s`,a`): the maximum Q value among all possible actions a` under the new state s`, which represents the best expected return under the new state; s`: the new state after executing action a; β: the momentum term coefficient, which helps smooth the Q value update and avoid violent fluctuations, with a value range of 0≤β≤10; ΔQ(s,a): the increment of the last Q value update, used to calculate the momentum term.

Images