Brake pad service life prediction method and device, electronic equipment, storage medium and vehicle

Abstract

The invention provides a brake pad service life prediction method and device, electronic equipment, a storage medium and a vehicle. The method comprises the steps that inertial sensor data and brake sound data of a to-be-predicted vehicle are acquired; obtaining spectrum features of the brake sound data based on the brake sound data; obtaining sequence features of the inertial sensor data based on the inertial sensor data; performing splicing processing on the spectrum features and the sequence features, and performing feature extraction on a splicing processing result to obtain comprehensive features of the inertial sensor data and the brake sound data; according to the scheme of the invention, the service life of the brake pad is predicted by introducing the sound during braking and the data of the inertial sensor, and no extra detection cost is needed. And the abrasion degree of the brake pad or whether other abnormal conditions such as traffic accidents and collision exist or not can be detected through the sound generated during braking, the method can be suitable for changeable complex environments, and the prediction result is more accurate.

Description

Technical Field

[0001] This disclosure relates to the field of data processing technology, and in particular to a method, apparatus, electronic device, storage medium, and vehicle for predicting brake pad life. Background Technology

[0002] Brake pads are used frequently in the current use of automobiles, and since they are related to vehicle safety, they need to be inspected regularly to determine their lifespan.

[0003] The existing brake pad life prediction scheme utilizes sensor technology to detect signals such as brake temperature, brake fluid pressure, and braking speed, fuses their time-domain characteristics, and studies the local time-frequency features required for brake pad wear state identification. A convolutional neural network is used, first taking the aforementioned time-domain and time-frequency features as input, and then outputting the brake pad wear amount after each braking action. The wear amounts after braking are accumulated to achieve real-time monitoring of the brake pad wear state during braking. However, in practical applications, the force and duration of each braking action are inconsistent, and there may be other interfering factors such as collisions. Therefore, relying solely on material science data collected by sensors to detect brake pad wear is difficult to apply to complex and variable environments, leading to inaccurate prediction results. Summary of the Invention

[0004] This disclosure provides a method, apparatus, electronic device, storage medium, and vehicle for predicting brake pad life.

[0005] According to a first aspect of this disclosure, an information method is provided, comprising: acquiring inertial sensor data and brake sound data of a vehicle to be predicted; obtaining spectral features of the brake sound data based on the brake sound data; obtaining sequence features of the inertial sensor data based on the inertial sensor data; performing concatenation processing on the spectral features and sequence features, and performing feature extraction on the result of the concatenation processing to obtain comprehensive features of the inertial sensor data and brake sound data; and performing multi-classification processing on the comprehensive features to obtain the brake pad life of the vehicle to be predicted.

[0006] In some embodiments, obtaining the spectral features of brake sound data based on brake sound data includes: performing Fourier transform processing on the brake sound data to obtain the spectrum of the brake sound data; performing convolution processing on the spectrum to extract local features of the spectrum; and performing pooling processing on the local features to obtain the spectral features of the brake sound data.

[0007] In some embodiments, obtaining sequence features of inertial sensor data based on inertial sensor data includes: converting the inertial sensor data into a vector sequence; processing data at different positions in the vector sequence through a self-attention mechanism to obtain initial sequence features; performing a linear transformation on the initial sequence features and inputting it into a nonlinear activation function for processing to obtain high-dimensional initial sequence features; and performing residual connection and normalization processing on the high-dimensional initial sequence features to obtain the sequence features of the inertial sensor data.

[0008] In some embodiments, feature extraction is performed on the stitching result to obtain the comprehensive features of inertial sensor data and brake sound data, including: linearly combining the stitching result with a preset weight matrix to obtain a combined result; and inputting the combined result into a nonlinear activation function for processing to obtain the comprehensive features of inertial sensor data and brake sound data.

[0009] In some embodiments, performing multi-classification processing on the comprehensive features to obtain the brake pad life of the vehicle to be predicted includes: using a multi-classification function to obtain the probability distribution of different preset brake pad lifespans corresponding to the comprehensive features; and taking the brake pad lifespan corresponding to the highest probability value in the probability distribution as the brake pad lifespan of the vehicle to be predicted.

[0010] In some embodiments, the method further includes: generating a warning message in response to the brake pad life of the vehicle to be predicted being within a preset brake pad life range, to remind the user to check the brake pads of the vehicle to be predicted.

[0011] According to a second aspect of this disclosure, a brake pad life prediction device is provided. The device includes: an acquisition unit for acquiring inertial sensor data and brake sound data of a vehicle to be predicted; a first feature extraction unit for obtaining spectral features of the brake sound data based on the brake sound data; a second feature extraction unit for obtaining sequential features of the inertial sensor data based on the inertial sensor data; a feature fusion unit for splicing the spectral features and sequential features, and extracting features from the splicing result to obtain comprehensive features of the inertial sensor data and brake sound data; and a prediction unit for performing multi-classification processing on the comprehensive features to obtain the brake pad life of the vehicle to be predicted.

[0012] According to a third aspect of this disclosure, an electronic device is provided, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the method of the first aspect described above.

[0013] According to a fourth aspect of this disclosure, a non-transitory computer-readable storage medium is provided storing computer instructions, wherein the computer instructions are used to cause a computer to perform the method of the first aspect described above.

[0014] According to a fifth aspect of this disclosure, a computer program product is provided, including a computer program that, when executed by a processor, implements the method as described in the first aspect above.

[0015] According to a sixth aspect of this disclosure, a vehicle is provided, the vehicle including the device as described in the second aspect above or the electronic device as described in the third aspect above.

[0016] The present disclosure provides a method, apparatus, electronic device, and storage medium for predicting brake pad life. The method includes: acquiring inertial sensor data and brake sound data of the vehicle to be predicted; obtaining spectral features of the brake sound data based on the brake sound data; obtaining sequence features of the inertial sensor data based on the inertial sensor data; splicing the spectral features and sequence features, and extracting features from the splicing results to obtain comprehensive features of the inertial sensor data and brake sound data; and performing multi-classification processing on the comprehensive features to obtain the brake pad life of the vehicle to be predicted. The present disclosure introduces the sound during braking and inertial sensor data for brake pad life prediction. The sound during braking and inertial sensor data can be acquired and processed in real time without additional detection costs. The sound during braking can be used to detect the wear degree of brake pads or detect the presence of other abnormalities such as car accidents or collisions. It can be applied to complex and variable environments, and the prediction results are more accurate.

[0017] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of this application, nor is it intended to limit the scope of this application. Other features of this application will become readily apparent from the following description. Attached Figure Description

[0018] The accompanying drawings are provided to better understand this solution and do not constitute a limitation of this disclosure. Wherein:

[0019] Figure 1 A schematic flowchart illustrating a brake pad life prediction method provided in this embodiment of the disclosure;

[0020] Figure 2 An example diagram illustrating another brake pad life prediction method provided in this disclosure embodiment;

[0021] Figure 3 A schematic flowchart illustrating a brake pad life prediction method provided in this embodiment of the disclosure;

[0022] Figure 4 An example diagram illustrating a neural network model training process provided in this embodiment of the disclosure;

[0023] Figure 5 An example diagram of a neural network model with a dual-tower structure provided in this embodiment of the disclosure;

[0024] Figure 6 An example diagram illustrating a neural network model for predicting brake pad life, provided in an embodiment of this disclosure;

[0025] Figure 7 This is a schematic diagram of the structure of a brake pad life prediction device provided in an embodiment of the present disclosure;

[0026] Figure 8 A schematic block diagram of an example electronic device 500 provided for embodiments of this disclosure. Detailed Implementation

[0027] The exemplary embodiments of this disclosure are described below with reference to the accompanying drawings, including various details of the embodiments to aid understanding, and should be considered merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of this disclosure. Similarly, for clarity and brevity, descriptions of well-known functions and structures are omitted in the following description.

[0028] The brake pad life prediction method, apparatus, electronic device, and storage medium of this disclosure are described below with reference to the accompanying drawings.

[0029] Figure 1 A method for predicting brake pad life is provided, which is applied to an in-vehicle terminal. Specifically, it can be applied to a cockpit domain controller (HUT Head Unit, HU) or a vehicle domain controller (XCU). This disclosure does not limit the application to this method, as long as it can achieve the desired function.

[0030] like Figure 1 As shown, the method includes:

[0031] Step 101: Obtain inertial sensor data and braking sound data of the vehicle to be predicted.

[0032] In some embodiments, the vehicle's inertial sensor data includes longitudinal (i.e., vehicle driving direction) acceleration data, which mainly includes the vehicle's acceleration during the deceleration phase when braking.

[0033] In some embodiments, brake sound data includes brake sound signals recorded when the vehicle brakes, which can be obtained in real time by the vehicle's microphone.

[0034] In some embodiments, the method can be used in the model training stage for brake pad life prediction, using vehicle inertial sensor data and brake sound data as training data. The label corresponding to the training data can be the remaining service life of the brake pads. Further, the model can be trained using the training data and its label to obtain a trained model for brake pad life prediction, wherein the remaining service life of the brake pads is the difference between the theoretical service life and the service life of the brake pads.

[0035] In some embodiments, a neural network model is used for model training and brake pad life prediction. The neural network model can automatically learn and extract effective features from different training data.

[0036] Step 102: Obtain the spectral characteristics of the brake sound data based on the brake sound data.

[0037] In some embodiments, the spectral features of the brake sound data are obtained by extracting the spectrum of the brake sound data and performing feature extraction on the spectrum.

[0038] Furthermore, in some embodiments, convolutional blocks of a convolutional neural network can be used to extract features from the spectrum.

[0039] Step 103: Obtain the sequence characteristics of the inertial sensor data based on the inertial sensor data.

[0040] In some embodiments, inertial sensor data is converted into a vector sequence, and feature extraction is performed on the vector sequence to extract the sequence features of the inertial sensor data.

[0041] In some embodiments, a neural network model with a dual-tower structure is used to predict brake pad life. This model includes two parallel neural networks that simultaneously extract features from brake sound data and inertial sensor data. During the model training phase, the dual-tower structure allows for independent learning and fitting of the vehicle's inertial sensor data and brake sound data, thereby improving the accuracy and robustness of the model's prediction results.

[0042] In some embodiments, the neural network model includes a sound encoder and an inertial sensor encoder, wherein the sound encoder is used to learn from the inertial sensor data of the vehicle, and the inertial sensor encoder learns from the inertial sensor data.

[0043] In some embodiments, the sound encoder is composed of a convolutional neural network (CNN) model; the inertial sensor encoder is composed of an encoder of a transformer model, wherein the transformer model is a deep learning model based on a self-attention mechanism, which can process sequential information such as sequential information composed of inertial sensor data of a vehicle within a preset time period, and learn the complex dependencies between different data in the sequence.

[0044] It should be noted that artificial intelligence is also widely used in the field of acoustic detection. Common architectures include multilayer perceptron (MLP), convolutional neural network (CNN), recurrent neural network (RNN), etc. However, these models are usually only used to detect whether there is a problem with the brake pads, but cannot predict the life of the brake pads. The neural network model trained based on the method disclosed in this paper can predict the life of the brake pads.

[0045] Step 104: The spectral features and sequence features are spliced ​​together, and the splicing results are used to extract features to obtain the combined features of the inertial sensor data and the brake sound data.

[0046] Step 105: Perform multi-classification processing on the comprehensive features to obtain the brake pad life of the vehicle to be predicted.

[0047] In some embodiments, the method further includes: generating a warning message in response to the brake pad life of the vehicle to be predicted being within a preset brake pad life range, to remind the user to check the brake pads of the vehicle to be predicted.

[0048] In some embodiments, such as Figure 2 As shown, it also includes sending warning information to users. Specifically, the warning information can be broadcast through an audio system or displayed on a vehicle display screen, or it can be uploaded to the cloud and sent to the user's mobile device via the cloud.

[0049] In some embodiments, such as Figure 2 As shown, it also includes: recording each prediction result and generating brake pad life records and brake usage records. For example, each test result can be connected to the cloud, and the brake life prediction result can be uploaded to the cloud to obtain test records for different time periods; the time nodes of the test are recorded, including the user's braking frequency, braking duration, etc., to generate brake usage records. Among them, brake usage records can also be used as user habit information to predict user driving habits, etc.

[0050] In some embodiments, the method further includes: collecting inertial sensor data and brake sound data of the vehicle to be predicted at preset time intervals, predicting the brake pad life based on the inertial sensor data and brake sound data of the vehicle to be predicted, and obtaining the brake pad life of the vehicle to be predicted.

[0051] In other words, based on the actual application of the vehicle to be predicted, the brake pad life of the vehicle to be predicted can be predicted according to the set strategy, such as setting it to predict once a week, which can effectively ensure the safety of the vehicle and the user.

[0052] In some embodiments, brake sound data can be used to detect minor defects and cracks on the surface of brake pads, detect abnormal sound characteristics, detect the wear degree of brake pads or whether there are other abnormalities, achieve a comprehensive assessment of brake pad life, and be applicable to varied and complex environments.

[0053] It is understood that the brake pad life prediction method disclosed herein does not require disassembly or destructive testing of the brake pads, can be tested under actual use conditions, does not require additional components, and has a low cost.

[0054] In summary, the brake pad life prediction method provided in this disclosure acquires inertial sensor data and brake sound data of the vehicle to be predicted; obtains the spectral features of the brake sound data based on the brake sound data; obtains the sequence features of the inertial sensor data based on the inertial sensor data; performs splicing processing on the spectral features and sequence features, and extracts features from the splicing results to obtain comprehensive features of the inertial sensor data and brake sound data; performs multi-classification processing on the comprehensive features to obtain the brake pad life of the vehicle to be predicted. The scheme of this disclosure introduces the sound of braking and inertial sensor data for brake pad life prediction. The sound of braking and inertial sensor data can be acquired and processed in real time without additional detection costs. Moreover, the sound of braking can detect small defects and cracks on the surface of the brake pads, or whether there are other abnormalities such as car accidents. It can be applied to complex and variable environments, and the prediction results are more accurate.

[0055] based on Figure 1 The embodiment shown, Figure 3 The following is a flowchart illustrating a brake pad life prediction method provided in an embodiment of the present disclosure. It is applied to an in-vehicle terminal. Specifically, it can be applied to a cockpit domain controller (HUT Head Unit, HU) or a vehicle domain controller (XCU). The present disclosure does not limit this application, as long as the function can be achieved.

[0056] like Figure 3 As shown, the method includes:

[0057] Step 201: Obtain inertial sensor data and braking sound data of the vehicle to be predicted.

[0058] In some embodiments, the vehicle's inertial sensor data includes longitudinal (i.e., vehicle driving direction) acceleration data, which mainly includes the vehicle's acceleration during the deceleration phase when braking.

[0059] In some embodiments, brake sound data includes brake sound signals recorded when the vehicle brakes, which can be obtained in real time by the vehicle's microphone.

[0060] In some embodiments, the method can be used in the model training stage for predicting brake pad lifespan. Inertial sensor data and brake sound data of the vehicle are used as training data. The label corresponding to the training data can be the remaining service life of the brake pad. Furthermore, the model can be trained using the training data and its label to obtain a trained model for predicting brake pad lifespan. The remaining service life of the brake pad is the difference between the theoretical service life and the service life of the brake pad.

[0061] Step 202: Obtain the spectral characteristics of the brake sound data based on the brake sound data.

[0062] In some embodiments, obtaining the spectral features of brake sound data based on brake sound data includes: performing Fourier transform processing on the brake sound data to obtain the spectrum of the brake sound data; performing convolution processing on the spectrum to extract local features of the spectrum; and performing pooling processing on the local features to obtain the spectral features of the brake sound data.

[0063] In some embodiments, a dual-tower neural network model is proposed for predicting brake pad life. The neural network model includes a sound encoder, an inertial sensor encoder, a fully connected layer (FCLayer), and a multi-classification layer. The dual-tower structure of the neural network model refers to the parallel structure of the sound encoder and the inertial sensor encoder.

[0064] In some embodiments, brake sound data is input into the sound encoder of a neural network model, and the spectral characteristics of the brake sound data can be obtained after model processing, such as... Figure 4 As shown, the neural network model includes a sound encoder (wav_encoder), an inertial sensor encoder (acc_encoder), a fully connected layer (FC), and a multi-classification layer, where the multi-classification layer includes a softmax function. Braking sound data (wav data) is input into the wav_encoder for feature extraction to obtain the spectral features of the braking sound data.

[0065] In some embodiments, the audio encoder is composed of a CNN, which includes a Fourier transform module and convolutional blocks, with each convolutional block containing at least one convolutional layer and a pooling layer.

[0066] Furthermore, in some embodiments, the brake sound data is extracted using a sound encoder to obtain the spectral features of the brake sound data, including: performing Fourier transform processing on the brake sound data through a Fourier transform module to obtain the spectrum of the brake sound data; and inputting the spectrum into a convolutional block for feature extraction to obtain the spectral features of the brake sound data.

[0067] For example, such as Figure 5 As shown, the audio encoder includes a Fourier transform module (FFT) and three convolutional blocks, each containing two convolutional layers, as shown below. Figure 5 The conv3-32 and a pooling layer are as follows Figure 6 In the max-pool, the brake sound data is used as the input of the sound encoder, i.e., wav_inputs. The Fourier transform module performs a Fourier transform operation on the brake sound data to convert the sound into the frequency domain and obtain the sound spectrum. Then, the spectral features are extracted through three convolutional blocks for subsequent prediction.

[0068] Step 203: Obtain the sequence characteristics of the inertial sensor data based on the inertial sensor data.

[0069] In some embodiments, obtaining sequence features of inertial sensor data based on inertial sensor data includes: converting the inertial sensor data into a vector sequence; processing data at different positions in the vector sequence through a self-attention mechanism to obtain initial sequence features; performing a linear transformation on the initial sequence features and inputting it into a nonlinear activation function for processing to obtain high-dimensional initial sequence features; and performing residual connection and normalization processing on the high-dimensional initial sequence features to obtain the sequence features of the inertial sensor data.

[0070] In some embodiments, inertial sensor data is input into an inertial sensor encoder of a neural network model. After model processing, the sequence characteristics of the inertial sensor data can be obtained, such as... Figure 4 As shown, the inertial sensor data, i.e., accdata, is input into the inertial sensor encoder acc_encoder for feature extraction.

[0071] In some embodiments, the inertial sensor encoder consists of a decoder of a transformer model, and includes an input embedding layer, a multi-head attention layer, and a feedforward neural network layer.

[0072] In some embodiments, the input of the transformer model position encoding is removed, and only the sequence input is retained. The sequence input includes a sequence of inertial sensor data within a preset time period. For example, the acceleration data of the vehicle from the inertial sensor from 0 to 1 second is collected, and a data is extracted every 0.1 seconds within the time period of 0 to 1 second to form a sequence.

[0073] Furthermore, in some embodiments, inertial sensor data is converted into a vector sequence through an input embedding layer; a multi-head attention layer is used to obtain the periodicity and correlation of different data in the vector sequence to obtain initial sequence features; and a feedforward neural network layer is used to perform nonlinear transformation on the initial sequence features to obtain the sequence features of the inertial sensor data.

[0074] For example, such as Figure 5 As shown, the inertial sensor encoder consists of a multi-head attention layer and a feedforward neural network layer. acc_inputs is the inertial sensor data, and the input embedding layer converts the inertial sensor data from text to an embedding vector representation. The outputs of each layer of the multi-head attention layer and the feedforward neural network layer also need to undergo residual connection (Add) and normalization (Norm) operations to capture the periodicity and correlation in the sequence and obtain key sequence features.

[0075] Step 204: The spectral features and sequence features are spliced ​​together, and the splicing results are used to extract features to obtain the combined features of the inertial sensor data and the brake sound data.

[0076] In some embodiments, after the data from the inertial sensor and the sound data from the brake pads are processed by two encoders to extract features, the two features are then horizontally spliced ​​together. The calculation formula is x = (a|b), where a and b represent the features of the data from the inertial sensor and the sound data from the brake pads, respectively, and x represents the splicing result.

[0077] In one implementation, such as Figure 4As shown, the neural network model includes a sound encoder (wav_encoder), an inertial sensor encoder (acc_encoder), a fully connected layer (FC), and a multi-classification layer, where the multi-classification layer includes a softmax function. Brake sound data (wav data) is input into the wav_encoder for feature extraction, and inertial sensor data (acc data) is input into the acc_encoder for feature extraction. The extracted features from the inertial sensor data and brake sound data are then input into the fully connected layer (FC) to extract combined features. These combined features are then input into the softmax function to calculate the final output, which is the brake pad life.

[0078] In some embodiments, feature extraction is performed on the stitching result to obtain the comprehensive features of inertial sensor data and brake sound data, including: linearly combining the stitching result with a preset weight matrix to obtain a combined result; and inputting the combined result into a nonlinear activation function for processing to obtain the comprehensive features of inertial sensor data and brake sound data.

[0079] Step 205: Use a multi-classification function to obtain the probability distribution of different brake pad lifespans corresponding to the comprehensive features.

[0080] Step 206: The brake pad life corresponding to the highest probability value in the probability distribution is taken as the brake pad life of the vehicle to be predicted.

[0081] In some embodiments, the multi-classification layer uses the Softmax function to handle multi-classification problems. Softmax is an activation function that can normalize a numerical vector into a probability distribution vector.

[0082] In some embodiments, such as Figure 5 As shown, the splicing result is input into a fully connected layer (FC) to extract comprehensive features, and then input into a softmax layer to calculate the predicted brake pad life.

[0083] In some embodiments, the training phase of the neural network model further includes: constructing a loss function based on the model's prediction results and training labels, updating the parameters of the neural network model using gradient descent backpropagation, until the value of the loss function reaches its minimum, thereby obtaining a trained neural network model.

[0084] In some embodiments, gradient descent is an optimization algorithm used to minimize a loss function. During the training of a neural network, the model parameters can be continuously adjusted using gradient descent to minimize the loss function. Specifically, this process involves calculating the gradient of the loss function with respect to the parameters, and then updating the parameter values ​​in the opposite direction of the gradient to reduce the value of the loss function. Here, the gradient represents the degree of influence or contribution of each parameter to the loss function.

[0085] In some embodiments, the gradient of the parameters can be efficiently calculated through backpropagation, thereby enabling parameter updates in gradient descent.

[0086] In some embodiments, the parameters of the neural network model are updated using gradient descent and backpropagation until the loss function reaches its minimum value, thus obtaining a trained neural network model. This includes: calculating the contributions of the sound encoder, inertial sensor encoder, fully connected layer, and multi-classification layer to the loss function; and adjusting the parameter values ​​of the inertial sensor encoder, fully connected layer, and multi-classification layer according to the contribution values ​​until the loss function reaches its minimum value, thus obtaining a trained neural network model.

[0087] In some embodiments, the learning coefficients of the parameters of the sound encoder, inertial sensor encoder, fully connected layer, and multi-classification layer can be set, and the parameters can be adjusted according to the product of the learning coefficients of the contribution values.

[0088] In some embodiments, the neural network model can continuously learn and adjust by backpropagating and optimizing the training data using gradient descent, thereby improving its ability to predict brake pad life.

[0089] In some embodiments, during the inference phase of the neural network model, such as Figure 6 As shown, the process includes: inputting the collected brake sound data (wav data) and inertial sensor data (acc data) into the sound encoder (wav_encoder) and inertial sensor encoder (acc_encoder) of the trained neural network model, respectively; concatenating the features output by wav_encoder and acc_encoder and inputting them into a fully connected layer (FC) for mapping; and then inputting them into a softmax layer for calculation to obtain the final prediction result, which is the final predicted brake pad life.

[0090] In some embodiments, the method further includes: generating a warning message in response to the brake pad life of the vehicle to be predicted being within a preset brake pad life range, to remind the user to check the brake pads of the vehicle to be predicted.

[0091] In some embodiments, the method further includes sending the warning information to the user. Specifically, the warning information can be broadcast through an audio system or displayed on an in-vehicle display screen, or the warning information can be uploaded to the cloud and sent to the user's mobile device via the cloud.

[0092] In some embodiments, the method further includes: recording each prediction result to generate brake pad life records and brake usage records. For example, each detection result can be interconnected with the cloud, and the brake life prediction result can be uploaded to the cloud to obtain detection records for different time periods; the time points of the recorded detection include the user's braking frequency, braking duration, etc., to generate brake usage records, wherein the brake usage records can also be used as user habit information to predict the user's driving habits, etc.

[0093] In some embodiments, the method further includes: collecting inertial sensor data and brake sound data of the vehicle to be predicted at preset intervals, inputting the inertial sensor data and brake sound data of the vehicle to be predicted into a trained neural network model to obtain the brake pad life of the vehicle to be predicted.

[0094] In summary, the brake pad life prediction method provided in this disclosure uses a dual-tower neural network model to predict brake pad life through brake sound data and inertial sensor data. It can adapt to various environments and emergencies, without causing damage to the brake pads or increasing additional costs. Compared with image-based detection methods, this method does not require the deployment of cameras, has a smaller computational load, and provides more accurate prediction results.

[0095] Figure 7 This is a schematic diagram of the structure of a brake pad life prediction device provided in an embodiment of this disclosure, as shown below. Figure 7 As shown, the device may include:

[0096] The acquisition unit 310 is used to acquire inertial sensor data and brake sound data of the vehicle to be predicted; the first feature extraction unit 320 is used to obtain the spectral features of the brake sound data based on the brake sound data; the second feature extraction unit 330 is used to obtain the sequence features of the inertial sensor data based on the inertial sensor data; the feature fusion unit 340 is used to perform splicing processing on the spectral features and sequence features, and to extract features from the splicing processing result to obtain the comprehensive features of the inertial sensor data and brake sound data; the prediction unit 350 is used to perform multi-classification processing on the comprehensive features to obtain the brake pad life of the vehicle to be predicted.

[0097] In some embodiments, the first feature extraction unit 320 is specifically used to: perform Fourier transform processing on the brake sound data to obtain the spectrum of the brake sound data; perform convolution processing on the spectrum to extract local features of the spectrum; and perform pooling processing on the local features to obtain the spectral features of the brake sound data.

[0098] In some embodiments, the second feature extraction unit 330 is specifically used to: convert inertial sensor data into a vector sequence; process data at different positions in the vector sequence through a self-attention mechanism to obtain initial sequence features; perform a linear transformation on the initial sequence features and input them into a nonlinear activation function for processing to obtain high-dimensional initial sequence features; and perform residual connection and normalization processing on the high-dimensional initial sequence features to obtain sequence features of the inertial sensor data.

[0099] In some embodiments, the feature fusion unit 340 is specifically used to: linearly combine the result of the splicing process with a preset weight matrix to obtain a combined result; and input the combined result into a nonlinear activation function for processing to obtain the integrated features of the inertial sensor data and the brake sound data.

[0100] In some embodiments, the prediction unit 350 is specifically used to: obtain the probability distribution of different brake pad lifespans corresponding to the comprehensive features using a multi-classification function; and take the brake pad lifespan corresponding to the highest probability value in the probability distribution as the brake pad lifespan of the vehicle to be predicted.

[0101] In some embodiments, the device further includes a generation unit for generating a warning message in response to the brake pad life of the vehicle to be predicted being within a preset brake pad life range, to remind the user to check the brake pads of the vehicle to be predicted.

[0102] It should be noted that the foregoing explanation of the method embodiments also applies to the apparatus of this embodiment, and the principle is the same, so it is not limited in this embodiment.

[0103] According to embodiments of this disclosure, this disclosure also provides an electronic device, a readable storage medium, and a computer program product.

[0104] Figure 8 A schematic block diagram of an example electronic device 400 that can be used to implement embodiments of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0105] like Figure 8As shown, device 400 includes a computing unit 401, which can perform various appropriate actions and processes based on a computer program stored in ROM (Read-Only Memory) 402 or a computer program loaded from storage unit 408 into RAM (Random Access Memory) 403. RAM 403 may also store various programs and data required for the operation of device 400. The computing unit 401, ROM 402, and RAM 403 are interconnected via bus 404. I / O (Input / Output) interface 405 is also connected to bus 404.

[0106] Multiple components in device 400 are connected to I / O interface 405, including: input unit 406, such as keyboard, mouse, etc.; output unit 407, such as various types of monitors, speakers, etc.; storage unit 408, such as disk, optical disk, etc.; and communication unit 409, such as network card, modem, wireless transceiver, etc. Communication unit 409 allows device 400 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0107] The computing unit 401 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 401 include, but are not limited to, CPUs (Central Processing Units), GPUs (Graphics Processing Units), various special-purpose AI (Artificial Intelligence) computing chips, various computing units running machine learning model algorithms, DSPs (Digital Signal Processors), and any suitable processor, controller, microcontroller, etc. The computing unit 401 performs the various methods and processes described above, such as the brake pad life prediction method. For example, in some embodiments, the brake pad life prediction method may be implemented as a computer software program tangibly contained in a machine-readable medium, such as storage unit 408. In some embodiments, part or all of the computer program may be loaded and / or installed on device 400 via ROM 402 and / or communication unit 409. When the computer program is loaded into RAM 403 and executed by the computing unit 401, one or more steps of the methods described above may be performed. Alternatively, in other embodiments, the computing unit 401 may be configured to perform the aforementioned brake pad life prediction method by any other suitable means (e.g., by means of firmware).

[0108] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, FPGAs (Field Programmable Gate Arrays), ASICs (Application-Specific Integrated Circuits), ASSPs (Application-Specific Standard Products), SOCs (System-on-Chips), CPLDs (Complex Programmable Logic Devices), computer hardware, firmware, software, and / or combinations thereof. These various implementations may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0109] The program code used to implement the methods of this disclosure may be written in any combination of one or more programming languages. This program code may be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing apparatus, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code may be executed entirely on a machine, partially on a machine, as a standalone software package partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0110] In the context of this disclosure, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, RAM, ROM, EPROM (Electrically Programmable Read-Only Memory) or flash memory, optical fiber, CD-ROM (Compact Disc Read-Only Memory), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0111] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user (e.g., a CRT (Cathode-Ray Tube) or LCD (Liquid Crystal Display) monitor); and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the computer. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0112] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or middleware components (e.g., application servers), or frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication (e.g., communication networks) of any form or medium. Examples of communication networks include LANs (Local Area Networks), WANs (Wide Area Networks), the Internet, and blockchain networks.

[0113] Computer systems can include clients and servers. Clients and servers are generally geographically separated and typically interact via communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. A server can be a cloud server, also known as a cloud computing server or cloud host, a hosting product within the cloud computing service ecosystem, addressing the shortcomings of traditional physical hosts and VPS (Virtual Private Server, or simply "VPS") services, such as high management difficulty and weak business scalability. Servers can also be servers for distributed systems or servers incorporating blockchain technology.

[0114] It's important to note that artificial intelligence (AI) is the study of enabling computers to simulate certain human thought processes and intelligent behaviors (such as learning, reasoning, thinking, and planning). It encompasses both hardware and software technologies. AI hardware technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, and big data processing. AI software technologies primarily include computer vision, brake pad life prediction, natural language processing, machine learning / deep learning, big data processing, and knowledge graph technologies.

[0115] It should be understood that the various forms of processes shown above can be used to rearrange, add, or delete steps. For example, the steps described in this disclosure can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this disclosure can be achieved, and this is not limited herein.

[0116] The specific embodiments described above do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure should be included within the scope of protection of this disclosure.

Claims

1. A method for predicting brake pad life, characterized in that, include: Acquire inertial sensor data and braking sound data of the vehicle to be predicted; The spectral characteristics of the brake sound data are obtained based on the brake sound data; Based on the inertial sensor data, the sequence characteristics of the inertial sensor data are obtained; The spectral features and the sequence features are concatenated, and the result of the concatenation is used to extract features to obtain the combined features of the inertial sensor data and the brake sound data. The brake pad life of the vehicle to be predicted is obtained by performing multi-classification processing on the comprehensive features.

2. The method according to claim 1, characterized in that, The step of obtaining the spectral characteristics of the brake sound data based on the brake sound data includes: The brake sound data is subjected to Fourier transform processing to obtain the spectrum of the brake sound data; The spectrum is convolved to extract local features of the spectrum; The local features are pooled to obtain the spectral features of the brake sound data.

3. The method according to claim 1, characterized in that, The step of obtaining the sequence features of the inertial sensor data based on the inertial sensor data includes: The inertial sensor data is converted into a vector sequence; The initial sequence features are obtained by processing data at different positions in the vector sequence using a self-attention mechanism. The initial sequence features are linearly transformed and then processed by a nonlinear activation function to obtain high-dimensional initial sequence features. The high-dimensional initial sequence features are subjected to residual connection and normalization processing to obtain the sequence features of the inertial sensor data.

4. The method according to claim 1, characterized in that, The process of extracting features from the splicing results to obtain comprehensive features of the inertial sensor data and the brake sound data includes: The result of the splicing process is linearly combined with a preset weight matrix to obtain the combined result; The combined results are input into a nonlinear activation function for processing to obtain the combined features of the inertial sensor data and the brake sound data.

5. The method according to claim 1, characterized in that, The process of performing multi-classification on the comprehensive features to obtain the brake pad life of the vehicle to be predicted includes: The probability distribution of different brake pad lifespans corresponding to the comprehensive features is obtained by using a multi-classification function; The brake pad life corresponding to the highest probability value in the probability distribution is taken as the brake pad life of the vehicle to be predicted.

6. The method according to claim 1, characterized in that, The method further includes: In response to the brake pad life of the vehicle to be predicted being within a preset brake pad life range, a warning message is generated to remind the user to check the brake pads of the vehicle to be predicted.

7. A brake pad life prediction device, characterized in that, include: The acquisition unit is used to acquire inertial sensor data and braking sound data of the vehicle to be predicted; The first feature extraction unit is used to obtain the spectral features of the brake sound data based on the brake sound data; The second feature extraction unit is used to obtain the sequence features of the inertial sensor data based on the inertial sensor data; The feature fusion unit is used to concatenate the spectral features and the sequence features, and to extract features from the concatenation result to obtain the combined features of the inertial sensor data and the brake sound data. The prediction unit is used to perform multi-classification processing on the comprehensive features to obtain the brake pad life of the vehicle to be predicted.

8. An electronic device, characterized in that, include: At least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the method of any one of claims 1 to 6.

9. A non-transitory computer-readable storage medium storing computer instructions, characterized in that, The computer instructions are used to cause the computer to perform the method according to any one of claims 1 to 6.

10. A vehicle, characterized in that, Includes the brake pad life prediction device of claim 7 or the electronic device of claim 8.

Images