Wheel speed signal anti-interference processing method, system and equipment based on automobile broadband road noise and medium
Through complex signal convolution neural networks, the vehicle's broadband road noise is processed, the spatial and temporal frequency domain characteristics are extracted and the interference suppression model is trained, which solves the problem of low wheel speed signal accuracy when the car is driving at high speed, improves the signal-to-noise ratio and data accuracy, and ensures that Car safety control.
Patent Information
- Application Number
- CN202411142129.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-20
- Publication Date
- 2024-11-26
- Estimated Expiration
- 2044-08-20
AI Technical Summary
When a car is driving at high speed, traditional noise processing methods are difficult to effectively eliminate broadband noise, resulting in a reduced accuracy of wheel speed signal and affecting car safety control.
Using a method based on complex signal convolution neural network, the interference received signals are collected through microphone arrays, overlapping frame processing and short-time Fourier transform, extract spatiotemporal and spatial frequency domain features, train interference suppression models, suppress noise interference, and improve wheel speed Signal-to-noise ratio of the signal.
It significantly improves the signal-to-noise ratio of the wheel speed sensor signal and the accuracy of the calculation data, ensures safety of automobile road driving, and solves the limitations of traditional methods when dealing with wideband road noise in flat spectrum.
Smart Images

Figure CN119030836A_ABST
Abstract
Description
Technical Field
[0001] The present application belongs to the field of automobile road safety control, and specifically relates to a wheel speed signal anti-interference processing method, system, equipment and medium based on automobile broadband road noise. Background Art
[0002] When the car is driving at high speed, the real-time and accurate collection and processing of wheel speed signals is the basis of active safety control systems such as indirect tire pressure monitoring systems, anti-lock braking systems, and drive anti-skid systems. The frequent operation of various inductive circuit devices such as relays, transformers, and electromagnets installed in the vehicle circuit induces noise in the circuit. In addition, various electromagnetic waves in the atmospheric environment will also cause the sensor to induce noise with a wide frequency band and small amplitude. The broadband noise caused by these internal circuit interferences and external environmental interferences on the vehicle makes it difficult to accurately obtain instantaneous wheel speed signals. At the same time, if the wheel speed signal is not accurately collected, the motion parameters such as wheel slip rate / slip rate and yaw angular velocity calculated from the wheel speed signal will also have large errors and large divergence, which will lead to a decrease in the signal-to-noise ratio of the wheel speed sensor signal and the divergence of the data calculated from the wheel speed signal, which is not conducive to the safe and effective control of the car.
[0003] Traditional Kalman filtering, wavelet noise reduction, digital Wiener filtering and other measures can appropriately reduce the noise in the wheel speed signal and reduce the divergence of the vehicle motion parameters calculated based on the wheel speed signal, but they cannot completely eliminate the noise. At present, Kalman filters and genetic iterative algorithms are usually used to suppress sensor signal errors, fault-tolerant methods based on analytical redundancy are used to solve the problem of sensor signal oscillation distortion, and adaptive enhancers based on frequency domain minimum mean square error are used as predictions for wheel speed sensor signals to improve the signal-to-noise ratio. However, the above methods are not suitable for vehicles driving at high speeds, and they are also limited in processing broadband road noise with a relatively flat spectrum. Summary of the invention
[0004] The present invention adopts a wheel speed signal anti-interference processing method based on automobile broadband road noise, and simultaneously extracts features of the interference receiving signal formed by the wheel speed target signal and the broadband noise interference source. Through the complex signal convolution neural network, the intrinsic relationship between the interference receiving signal and the original wheel speed signal is fully learned, and the interference component is suppressed to obtain an interference suppression model, thereby improving the signal-to-noise ratio of the wheel speed sensor signal and the accuracy of the data calculated from the wheel speed signal, thereby ensuring the safety of automobile road driving. It solves the limitation of the current traditional noise active method in processing broadband road noise with a relatively flat spectrum when processing automobiles at high speed, and realizes the improvement of the signal-to-noise ratio of the wheel speed sensor signal and the accuracy of the data calculated from the wheel speed signal. The specific technical scheme is as follows:
[0005] A wheel speed signal anti-interference processing method based on automobile broadband road noise, comprising:
[0006] S1: Obtain the original wheel speed signal data set, collect interference reception signals through the microphone array inside the car, the interference reception signals include wheel speed target signals and broadband noise interference sources, and construct interference signal training sets and interference signal test sets according to the interference reception signals;
[0007] S2: performing overlapping frame processing on the interference signal training set to obtain discrete signal frames, performing short-time Fourier transform on the discrete signal frames to obtain amplitude spectrum and phase spectrum, splicing the amplitude spectrum and phase spectrum to obtain spatiotemporal frequency domain features, and adding training labels to the interference signal training set according to the frequency spectrum features of the original wheel speed signal data set;
[0008] S3: Obtaining an interference suppression model through complex signal convolutional neural network training according to the spatiotemporal frequency domain features and the training labels;
[0009] S4: Performing a performance evaluation test on the interference suppression model using the interference signal test set to obtain a performance evaluation result.
[0010] Preferably, the microphone array is a uniform linear array composed of microphone sensors, and the interference reception signal is collected by the uniform linear array.
[0011] Preferably, the step S1 specifically includes:
[0012] A uniform linear array is equipped, the uniform linear array is composed of isotropic array elements, the interference reception signal is obtained through the uniform linear array, the interference reception signal is incident on the uniform linear array from different directions, the interference reception signal includes a wheel speed target signal and a broadband noise interference source, and an isotropic array element variable is obtained through the isotropic array element according to the interference signal, the isotropic array element variable includes an isotropic array element amplitude change and an isotropic array element delay, and the interference reception signal calculation formula is:
[0013]
[0014] τ m (φ d )=(m-1)Δsinφ d / c,
[0015] ρ m (q)=A / σ m (q),
[0016] μ m (q)=σ m (q) / c,
[0017] Among them, x m To interfere with the received signal, m represents the mth isotropic array element, and the value of m is a positive integer, ω l is the incident frequency of the interference receiving signal, k represents the kth group of isotropic array element variables, the value of k is a positive integer, x m (ω l , k) is the isotropic array element variable of the kth group, the incident frequency ω in the received data of the mth isotropic array element l Corresponding to the received signal, D is the total number of wheel speed target signals, d represents the dth wheel speed target signal, s d is the frequency domain of the wheel speed target signal, s d (ω l , k) represents the frequency domain of the kth group of isotropic array element variables and the dth wheel speed signal, e is the base of the logarithmic function, j is the complex unit, Φ d is the incident angle of the dth wheel speed target signal, τ m (Φ d ) is the relative delay of the wheel speed target signal, Δ is the isotropic array element spacing, c represents the speed of sound, which is 340 m / s, Q is the total number of broadband noise interference sources, q represents the qth broadband noise interference source, ρ is the amplitude of the broadband noise interference source, ρ m (q) is the amplitude when the qth broadband noise interference source reaches the mth isotropic array element, A is a constant, σ m (q) is the distance from the qth broadband noise interference source to the mth isotropic array element, i q (ω l , k) represents the kth group of isotropic array element variables, the frequency signal of the qth broadband noise interference source, μ m (q) represents the propagation delay from the qth broadband noise interference source to the mth isotropic array element.
[0018] Preferably, the step S3 specifically includes:
[0019] S301: The complex signal convolutional neural network is a convolutional neural network for processing complex signals;
[0020] S302: Performing complex convolution on the spatiotemporal frequency domain features to obtain a spatiotemporal frequency domain feature map;
[0021] S303: Perform maximum pooling on the spatiotemporal frequency domain feature map to obtain a spatiotemporal frequency domain pooling feature map;
[0022] S304: Passing the spatiotemporal frequency domain pooling feature map through a fully connected layer to obtain a feature transfer function, where the feature transfer function is a transfer function between the interference reception signal and the isotropic array element, and obtaining a wheel speed test signal according to an optimal solution of the feature transfer function, where the wheel speed test signal is a wheel speed target signal of the isotropic array element;
[0023] S305: Obtaining a wheel speed signal loss value through a loss function according to the wheel speed test signal and the training label. The calculation formula of the loss function is:
[0024]
[0025] Among them, COST is the wheel speed signal loss value, N is the number of spectrum frames input in each batch, and x deNoise,r(w) represents the real part of the wheel speed signal output by the complex signal convolutional neural network, x deNoise,i(w) represents the imaginary part of the wheel speed signal output by the complex signal convolutional neural network, x noNoise,r(w) represents the real part of the training label, x noNoise,i(w) represents the imaginary part of the training label.
[0026] S306: Obtaining weights of each layer of the complex signal convolutional neural network and bias gradients of each layer of the complex signal convolutional neural network through back propagation according to the wheel speed signal loss value;
[0027] S307: The interference suppression model is obtained by updating the complex signal convolutional neural network parameters through the weights of each layer of the complex signal convolutional neural network and the bias gradient of each layer of the complex signal convolutional neural network.
[0028] Preferably, the expression of the complex convolution is:
[0029]
[0030] w=w r +jw i ,
[0031] x=x r +jx i ,
[0032] Where x is a complex matrix, x r is the real part of the interference received signal, x i is the imaginary part of the interference received signal, w is the complex convolution kernel, w r is the real part parameter of the complex convolution kernel, w i is the imaginary parameter of the complex convolution kernel, j is a complex unit, and * represents a convolution operation.
[0033] Preferably, the specific steps of the back propagation are:
[0034] S306-1: Obtaining the loss value gradient of each layer through the sigmoid activation function according to the wheel speed signal loss value;
[0035] S306-2: According to the loss value gradient of each layer, back-propagation layer by layer by applying the chain rule is performed to obtain the weight of each layer of the complex signal convolutional neural network and the bias gradient of each layer of the complex signal convolutional neural network.
[0036] Preferably, the step S4 specifically includes:
[0037] The performance evaluation result is obtained by calculating the signal-to-noise ratio according to the interference received signal and the original wheel speed signal. The specific calculation formula is:
[0038] SNR improvement =SNR after -SNR before ,
[0039]
[0040] x(t) nosiy =x(t) total -x(t) ideal ,
[0041]
[0042] x(t) nosiy =x(t) total -x(t) ideal ,
[0043] Among them, SNR improvement The performance evaluation result, SNR before represents the signal-to-noise ratio of the interference signal training set that has not been processed by the interference suppression model, x(t) ideal represents the original wheel speed signal, x(t) nosiy represents the noise signal, x(t) total represents the interference received signal, SNR after represents the signal-to-noise ratio of the interference signal training set processed by the interference suppression model, x(t) knosiy represents the noise signal in the anti-interference wheel speed signal, x(t) ktotal represents the anti-interference wheel speed signal;
[0044] Determining whether the performance evaluation result is a positive number, if the performance evaluation result is a positive number, the interference suppression model passes the performance evaluation test;
[0045] If the performance evaluation result is a negative number, the interference suppression model fails the performance evaluation test;
[0046] Determine whether the interference suppression model passes the performance evaluation test. If the interference suppression model passes the performance evaluation test, obtain an interference-resistant wheel speed signal through the interference suppression model according to the real-time wheel speed signal. The real-time wheel speed signal is the wheel speed signal collected in real time by the automobile wheel speed sensor, and the anti-interference wheel speed signal is the real-time wheel speed signal with noise removed by the interference suppression model.
[0047] A wheel speed signal anti-interference processing system based on automobile broadband road noise includes a signal acquisition module, a signal preprocessing module, a model training module, and a performance evaluation module, including:
[0048] The signal acquisition module is used to obtain the original wheel speed signal data set, collect interference reception signals through the microphone array inside the car, and the interference reception signals include wheel speed target signals and broadband noise interference sources, and construct interference signal training sets and interference signal test sets according to the interference reception signals;
[0049] The signal preprocessing module is used to perform overlapping frame processing on the interference signal training set to obtain discrete signal frames, perform short-time Fourier transform on the discrete signal frames to obtain amplitude spectrum and phase spectrum, splice the amplitude spectrum and phase spectrum to obtain time-space frequency domain features, and add training labels to the interference signal training set according to the frequency spectrum features of the original wheel speed signal data set;
[0050] The model training module is used to obtain an interference suppression model through complex signal convolution neural network training according to the spatiotemporal frequency domain features and the training labels;
[0051] The performance evaluation module is used to perform a performance evaluation test on the interference suppression model through the interference signal test set to obtain a performance evaluation result.
[0052] An electronic device comprises a memory, a processor and a computer program stored in the memory and executable on the processor. When the processor executes the program, the above-mentioned wheel speed signal anti-interference processing method based on automobile broadband road noise is implemented.
[0053] A storage medium containing computer executable instructions, wherein the computer executable instructions are used to execute the above-mentioned wheel speed signal anti-interference processing method based on automobile broadband road noise when executed by a computer processor.
[0054] The beneficial effects of the present invention are:
[0055] (1) The interference reception signal is segmented on the time axis by overlapping framing to obtain the discrete signal frames. There is partial overlap between frames, which ensures the continuity of information between frames, can more accurately analyze the spectrum characteristics of the signal and reduce the computational complexity;
[0056] (2) The discrete signal frame is converted from the time domain to the frequency domain through short-time Fourier transform, and the spectrum information of the signal in each time period is analyzed through a sliding window to obtain the amplitude spectrum and phase spectrum, which is conducive to the subsequent distinction between the wheel speed signal and broadband noise;
[0057] (3) The amplitude spectrum, phase spectrum, real part and imaginary part of the interference reception signal are processed by complex convolution, which ensures the integrity of the signal and avoids the problem of losing the original intrinsic characteristics of the signal by simply separating the real part and the imaginary part, or processing the amplitude spectrum and the phase spectrum separately, thereby improving the ability to suppress the interference components in the interference reception signal;
[0058] (4) Through the deep complex convolutional neural network, the amplitude and phase information of the signal are taken into consideration at the same time, and the intrinsic relationship between the interference received signal and the original wheel speed signal is fully learned, the interference component in the interference received signal is suppressed, and the anti-interference wheel speed signal is output, thereby improving the signal-to-noise ratio of the wheel speed sensor signal and the accuracy of the wheel speed signal calculation data. BRIEF DESCRIPTION OF THE DRAWINGS
[0059] In order to facilitate understanding by those skilled in the art, the present invention is further described below with reference to the accompanying drawings.
[0060] Figure 1 The present invention is a flowchart of the steps of a wheel speed signal anti-interference processing method based on automobile broadband road noise. DETAILED DESCRIPTION
[0061] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the specific implementation methods, structures, features and effects of the present invention are described in detail below in conjunction with the accompanying drawings and preferred embodiments.
[0062] See also Figure 1 , a wheel speed signal anti-interference processing method based on automobile broadband road noise, comprising:
[0063] S1: Obtain the original wheel speed signal data set, collect interference reception signals through the microphone array inside the car, the interference reception signals include wheel speed target signals and broadband noise interference sources, and construct interference signal training sets and interference signal test sets according to the interference reception signals;
[0064] S2: performing overlapping frame processing on the interference signal training set to obtain discrete signal frames, performing short-time Fourier transform on the discrete signal frames to obtain amplitude spectrum and phase spectrum, splicing the amplitude spectrum and phase spectrum to obtain spatiotemporal frequency domain features, and adding training labels to the interference signal training set according to the frequency spectrum features of the original wheel speed signal data set;
[0065] S3: Obtaining an interference suppression model through complex signal convolutional neural network training according to the spatiotemporal frequency domain features and the training labels;
[0066] S4: Performing a performance evaluation test on the interference suppression model using the interference signal test set to obtain a performance evaluation result.
[0067] In step S1, the microphone array is a uniform linear array composed of microphone sensors, and the interference reception signal is collected by the uniform linear array.
[0068] In this embodiment, a uniform linear array is mounted, the uniform linear array is composed of isotropic array elements, the interference reception signal is obtained through the uniform linear array, the interference reception signal is incident on the uniform linear array from different directions, the interference reception signal includes a wheel speed target signal and a broadband noise interference source, and an isotropic array element variable is obtained through the isotropic array element according to the interference signal, the isotropic array element variable includes an isotropic array element amplitude change and an isotropic array element delay, and the interference reception signal calculation formula is:
[0069]
[0070] τ m (φ d )=(m-1)Δsinφ d / c,
[0071] ρ m (q)=A / σ m (q),
[0072] μ m (q)=σ m (q) / c,
[0073] Among them, x m To interfere with the received signal, m represents the mth isotropic array element, and the value of m is a positive integer, ω l is the incident frequency of the interference receiving signal, k represents the kth group of isotropic array element variables, the value of k is a positive integer, x m (ω l , k) is the isotropic array element variable of the kth group, the incident frequency ω in the received data of the mth isotropic array element lCorresponding to the received signal, D is the total number of wheel speed target signals, d represents the dth wheel speed target signal, s d is the frequency domain of the wheel speed target signal, s d (ω l , k) represents the frequency domain of the kth group of isotropic array element variables and the dth wheel speed signal, e is the base of the logarithmic function, j is the complex unit, Φ d is the incident angle of the dth wheel speed target signal, τ m (Φ d ) is the relative delay of the wheel speed target signal, Δ is the isotropic array element spacing, c represents the speed of sound, which is 340 m / s, Q is the total number of broadband noise interference sources, q represents the qth broadband noise interference source, ρ is the amplitude of the broadband noise interference source, ρ m (q) is the amplitude when the qth broadband noise interference source reaches the mth isotropic array element, A is a constant, σ m (q) is the distance from the qth broadband noise interference source to the mth isotropic array element, i q (ω l , k) represents the kth group of isotropic array element variables, the frequency signal of the qth broadband noise interference source, μ m (q) represents the propagation delay from the qth broadband noise interference source to the mth isotropic array element.
[0074] In step S3, an interference suppression model is obtained by training a complex signal convolutional neural network according to the spatiotemporal frequency domain features and the training labels, which can be specifically achieved by the following steps:
[0075] S301: The complex signal convolutional neural network is a convolutional neural network for processing complex signals;
[0076] S302: Performing complex convolution on the spatiotemporal frequency domain features to obtain a spatiotemporal frequency domain feature map;
[0077] The expression of the complex convolution is:
[0078]
[0079] w=w r +jw i ,
[0080] x=x r +jx i ,
[0081] Where x is a complex matrix, x r is the real part of the interference received signal, x i is the imaginary part of the interference received signal, w is the complex convolution kernel, wr is the real part parameter of the complex convolution kernel, w i is the imaginary parameter of the complex convolution kernel, j is a complex unit, and * represents a convolution operation.
[0082] S303: Perform maximum pooling on the spatiotemporal frequency domain feature map to obtain a spatiotemporal frequency domain pooling feature map;
[0083] S304: Passing the spatiotemporal frequency domain pooling feature map through a fully connected layer to obtain a feature transfer function, where the feature transfer function is a transfer function between the interference reception signal and the isotropic array element, and obtaining a wheel speed test signal according to an optimal solution of the feature transfer function, where the wheel speed test signal is a wheel speed target signal of the isotropic array element;
[0084] S305: Obtaining a wheel speed signal loss value through a loss function according to the wheel speed test signal and the training label. The calculation formula of the loss function is:
[0085]
[0086] Among them, COST is the wheel speed signal loss value, N is the number of spectrum frames input in each batch, and x deNoise,r(w) represents the real part of the wheel speed signal output by the complex signal convolutional neural network, x deNoise,i(w) represents the imaginary part of the wheel speed signal output by the complex signal convolutional neural network, x noNoise,r(w) represents the real part of the training label, x noNoise,i(w) represents the imaginary part of the training label.
[0087] S306: Obtaining weights of each layer of the complex signal convolutional neural network and bias gradients of each layer of the complex signal convolutional neural network through back propagation according to the wheel speed signal loss value;
[0088] S307: The interference suppression model is obtained by updating the complex signal convolutional neural network parameters through the weights of each layer of the complex signal convolutional neural network and the bias gradient of each layer of the complex signal convolutional neural network.
[0089] It should be noted that, in step S301, the complex signal convolutional neural network includes 3 convolutional layers, 3 pooling layers, and 1 fully connected layer; in step S302, the complex signal convolutional neural network uses the amplitude and phase as well as the real and imaginary parts of the spectrum to perform complex convolution operations when processing the space-time frequency domain features to obtain the space-time frequency domain feature map.
[0090] In this embodiment, the weights of each layer of the complex signal convolutional neural network and the bias gradients of each layer of the complex signal convolutional neural network are obtained by back propagation according to the wheel speed signal loss value, which is specifically achieved by the following steps:
[0091] S306-1: Obtaining the loss value gradient of each layer through the sigmoid activation function according to the wheel speed signal loss value;
[0092] S306-2: According to the loss value gradient of each layer, back-propagation layer by layer by applying the chain rule is performed to obtain the weight of each layer of the complex signal convolutional neural network and the bias gradient of each layer of the complex signal convolutional neural network.
[0093] In step S4, the performance evaluation result is obtained by calculating the signal-to-noise ratio according to the interference received signal and the original wheel speed signal. The specific calculation formula is:
[0094] SNR improvement =SNR after -SNR before ,
[0095]
[0096] x(t) nosiy =x(t) total -x(t) ideal ,
[0097]
[0098] x(t) nosiy =x(t) total -x(t) ideal ,
[0099] Among them, SNR improvement The performance evaluation result, SNR before represents the signal-to-noise ratio of the interference signal training set that has not been processed by the interference suppression model, x(t) ideal represents the original wheel speed signal, x(t) nosiy represents the noise signal, x(t) total represents the interference received signal, SNR after represents the signal-to-noise ratio of the interference signal training set processed by the interference suppression model, x(t) knosiy represents the noise signal in the anti-interference wheel speed signal, x(t) ktotal represents the anti-interference wheel speed signal;
[0100] Determining whether the performance evaluation result is a positive number, if the performance evaluation result is a positive number, the interference suppression model passes the performance evaluation test;
[0101] If the performance evaluation result is a negative number, the interference suppression model fails the performance evaluation test;
[0102] Determine whether the interference suppression model passes the performance evaluation test. If the interference suppression model passes the performance evaluation test, obtain an interference-resistant wheel speed signal through the interference suppression model according to the real-time wheel speed signal. The real-time wheel speed signal is the wheel speed signal collected in real time by the automobile wheel speed sensor, and the anti-interference wheel speed signal is the real-time wheel speed signal with noise removed by the interference suppression model.
[0103] A wheel speed signal anti-interference processing system based on automobile broadband road noise includes a signal acquisition module, a signal preprocessing module, a model training module, and a performance evaluation module;
[0104] The signal acquisition module is used to obtain the original wheel speed signal data set, collect interference reception signals through the microphone array inside the car, and the interference reception signals include wheel speed target signals and broadband noise interference sources, and construct interference signal training sets and interference signal test sets according to the interference reception signals;
[0105] The signal preprocessing module is used to perform overlapping frame processing on the interference signal training set to obtain discrete signal frames, perform short-time Fourier transform on the discrete signal frames to obtain amplitude spectrum and phase spectrum, splice the amplitude spectrum and phase spectrum to obtain time-space frequency domain features, and add training labels to the interference signal training set according to the frequency spectrum features of the original wheel speed signal data set;
[0106] The model training module is used to obtain an interference suppression model through complex signal convolution neural network training according to the spatiotemporal frequency domain features and the training labels;
[0107] The performance evaluation module is used to perform a performance evaluation test on the interference suppression model through the interference signal test set to obtain a performance evaluation result.
[0108] The computer storage medium of the embodiment of the present invention may adopt any combination of one or more computer-readable media. The computer-readable medium may be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples (non-exhaustive list) of computer-readable storage media include: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this document, a computer-readable storage medium may be any tangible medium containing or storing a program that may be used by or in combination with an instruction execution system, device or device.
[0109] Computer-readable signal media may include data signals propagated in baseband or as part of a carrier wave, which carry computer-readable program code. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Computer-readable signal media may also be any computer-readable medium other than a computer-readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device.
[0110] The program code included on the computer readable medium can be transmitted with any appropriate medium, including but not limited to wireless, electric wire, optical cable, RF, etc., or any suitable combination of the above. The computer program code for performing the operation of the present invention can be written in one or more programming languages or their combinations, and the programming language includes object-oriented programming languages-such as Java, Smalltalk, C++, and also includes conventional procedural programming languages-such as "C" language or similar programming languages. The program code can be executed completely on the user's computer, partially on the user's computer, as an independent software package, partially on the user's computer and partially on the remote computer, or completely on the remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computer (for example, using an Internet service provider to connect through the Internet).
[0111] The above description is only a preferred embodiment of the present invention and does not limit the present invention in any form. Although the present invention has been disclosed as a preferred embodiment as above, it is not used to limit the present invention. Any technical personnel in this field can make some changes or modify the technical contents disclosed above into equivalent embodiments without departing from the scope of the technical solution of the present invention. However, any brief modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention are still within the scope of the technical solution of the present invention.
Claims
1. A wheel speed signal anti-interference processing method based on automobile broadband road noise, characterized in that: include: S1: Obtain the original wheel speed signal data set, collect interference reception signals through the microphone array inside the car, the interference reception signals include wheel speed target signals and broadband noise interference sources, and construct interference signal training sets and interference signal test sets according to the interference reception signals; S2: performing overlapping frame processing on the interference signal training set to obtain discrete signal frames, performing short-time Fourier transform on the discrete signal frames to obtain amplitude spectrum and phase spectrum, splicing the amplitude spectrum and phase spectrum to obtain spatiotemporal frequency domain features, and adding training labels to the interference signal training set according to the frequency spectrum features of the original wheel speed signal data set; S3: Obtaining an interference suppression model through complex signal convolutional neural network training according to the spatiotemporal frequency domain features and the training labels; S4: Performing a performance evaluation test on the interference suppression model using the interference signal test set to obtain a performance evaluation result.
2. The wheel speed signal anti-interference processing method based on automobile broadband road noise according to claim 1, characterized in that: The microphone array is a uniform linear array composed of microphone sensors, and the interference reception signal is collected by the uniform linear array.
3. The wheel speed signal anti-interference processing method based on automobile broadband road noise according to claim 1, characterized in that: The step S1 specifically includes: A uniform linear array is equipped, the uniform linear array is composed of isotropic array elements, the interference reception signal is obtained through the uniform linear array, the interference reception signal is incident on the uniform linear array from different directions, the interference reception signal includes a wheel speed target signal and a broadband noise interference source, and an isotropic array element variable is obtained through the isotropic array element according to the interference signal, the isotropic array element variable includes an isotropic array element amplitude change and an isotropic array element delay, and the interference reception signal calculation formula is: t m (f d )=(m-1)Δsinφ d / c, r m (q)=A / σ m (q), μ m (q)=σ m (q) / c, Among them, x m To interfere with the received signal, m represents the mth isotropic array element, and the value of m is a positive integer, ω l is the incident frequency of the interference receiving signal, k represents the kth group of isotropic array element variables, the value of k is a positive integer, x m (ω l , k) is the isotropic array element variable of the kth group, the incident frequency ω in the received data of the mth isotropic array element l Corresponding to the received signal, D is the total number of wheel speed target signals, d represents the dth wheel speed target signal, s d is the frequency domain of the wheel speed target signal, s d (ω l , k) represents the frequency domain of the kth group of isotropic array element variables and the dth wheel speed signal, e is the base of the logarithmic function, j is the complex unit, Φ d is the incident angle of the dth wheel speed target signal, Tm(Φd) is the relative delay of the wheel speed target signal, t is the isotropic array element spacing, c represents the speed of sound, which is 340 m / s, Q is the total number of broadband noise interference sources, q represents the qth broadband noise interference source, ρ is the amplitude of the broadband noise interference source, ρm(q) is the amplitude when the qth broadband noise interference source reaches the mth isotropic array element, A is a constant, σ m (q) is the distance from the qth broadband noise interference source to the mth isotropic array element, i q (ω l , k) represents the kth group of isotropic array element variables, the frequency signal of the qth broadband noise interference source, μ m (q) represents the propagation delay from the qth broadband noise interference source to the mth isotropic array element.
4. The wheel speed signal anti-interference processing method based on automobile broadband road noise according to claim 1, characterized in that: The step S3 specifically includes: S301: The complex signal convolutional neural network is a convolutional neural network for processing complex signals; S302: Performing complex convolution on the spatiotemporal frequency domain features to obtain a spatiotemporal frequency domain feature map; S303: Perform maximum pooling on the spatiotemporal frequency domain feature map to obtain a spatiotemporal frequency domain pooling feature map; S304: Passing the spatiotemporal frequency domain pooling feature map through a fully connected layer to obtain a feature transfer function, where the feature transfer function is a transfer function between the interference reception signal and the isotropic array element, and obtaining a wheel speed test signal according to an optimal solution of the feature transfer function, where the wheel speed test signal is a wheel speed target signal of the isotropic array element; S305: Obtaining a wheel speed signal loss value through a loss function according to the wheel speed test signal and the training label. The calculation formula of the loss function is: Among them, COST is the wheel speed signal loss value, N is the number of spectrum frames input in each batch, and x deNoise,r(w) represents the real part of the wheel speed signal output by the complex signal convolutional neural network, x deNoise,i(w) represents the imaginary part of the wheel speed signal output by the complex signal convolutional neural network, x noNoise,r(w) represents the real part of the training label, x noNoise,i(w) represents the imaginary part of the training label. S306: Obtaining weights of each layer of the complex signal convolutional neural network and bias gradients of each layer of the complex signal convolutional neural network through back propagation according to the wheel speed signal loss value; S307: The interference suppression model is obtained by updating the complex signal convolutional neural network parameters through the weights of each layer of the complex signal convolutional neural network and the bias gradient of each layer of the complex signal convolutional neural network.
5. The wheel speed signal anti-interference processing method based on automobile broadband road noise according to claim 4 is characterized in that: The expression of the complex convolution is: w=w r +jw i , x=x r +jx i , Where x is a complex matrix, x r is the real part of the interference received signal, x i is the imaginary part of the interference received signal, w is the complex convolution kernel, w r is the real part parameter of the complex convolution kernel, w i is the imaginary parameter of the complex convolution kernel, j is a complex unit, and * represents a convolution operation.
6. The wheel speed signal anti-interference processing method based on automobile broadband road noise according to claim 4 is characterized in that: The specific steps of the back propagation are: S306-1: Obtaining the loss value gradient of each layer through the sigmoid activation function according to the wheel speed signal loss value; S306-2: According to the loss value gradient of each layer, back-propagation layer by layer by applying the chain rule is performed to obtain the weight of each layer of the complex signal convolutional neural network and the bias gradient of each layer of the complex signal convolutional neural network.
7. The wheel speed signal anti-interference processing method based on automobile broadband road noise according to claim 1, characterized in that: The step S4 specifically includes: The performance evaluation result is obtained by calculating the signal-to-noise ratio according to the interference received signal and the original wheel speed signal. The specific calculation formula is: SNR improvement =SNR after -SNR before , x(t) nosiy =x(t) total -x(t) ideal , x(t) nosiy =x(t) total -x(t) ideal , Among them, SNR improvement The performance evaluation result, SNR before represents the signal-to-noise ratio of the interference signal training set that has not been processed by the interference suppression model, x(t) ideal represents the original wheel speed signal, x(t) nosiy represents the noise signal, x(t) total represents the interference received signal, SNR after represents the signal-to-noise ratio of the interference signal training set processed by the interference suppression model, x(t) knosiy represents the noise signal in the anti-interference wheel speed signal, x(t) ktotal represents the anti-interference wheel speed signal; Determining whether the performance evaluation result is a positive number, if the performance evaluation result is a positive number, the interference suppression model passes the performance evaluation test; If the performance evaluation result is a negative number, the interference suppression model fails the performance evaluation test; Determine whether the interference suppression model passes the performance evaluation test. If the interference suppression model passes the performance evaluation test, obtain an interference-resistant wheel speed signal through the interference suppression model according to the real-time wheel speed signal. The real-time wheel speed signal is the wheel speed signal collected in real time by the automobile wheel speed sensor, and the anti-interference wheel speed signal is the real-time wheel speed signal with noise removed by the interference suppression model.
8. A wheel speed signal anti-interference processing system based on automobile broadband road noise, comprising a signal acquisition module, a signal preprocessing module, a model training module, and a performance evaluation module, characterized in that: include: The signal acquisition module is used to obtain the original wheel speed signal data set, collect interference reception signals through the microphone array inside the car, and the interference reception signals include wheel speed target signals and broadband noise interference sources, and construct interference signal training sets and interference signal test sets according to the interference reception signals; The signal preprocessing module is used to perform overlapping frame processing on the interference signal training set to obtain discrete signal frames, perform short-time Fourier transform on the discrete signal frames to obtain amplitude spectrum and phase spectrum, splice the amplitude spectrum and phase spectrum to obtain time-space frequency domain features, and add training labels to the interference signal training set according to the frequency spectrum features of the original wheel speed signal data set; The model training module is used to obtain an interference suppression model through complex signal convolution neural network training according to the spatiotemporal frequency domain features and the training labels; The performance evaluation module is used to perform a performance evaluation test on the interference suppression model through the interference signal test set to obtain a performance evaluation result.
9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the method for anti-interference processing of wheel speed signals based on automobile broadband road noise as described in any one of claims 1 to 8 is implemented.
10. A storage medium containing computer executable instructions, characterized in that: The computer executable instructions are used to execute the wheel speed signal anti-interference processing method based on automobile broadband road noise as described in any one of claims 1 to 8 when executed by a computer processor.
Citation Information
Patent Citations
Model training method and device and electronic equipment
CN116304658A
Vehicle-mounted noise reduction method and device, vehicle, electronic equipment and storage medium
CN117765913A
Aircraft simulator sound noise reduction method, system, equipment and medium
CN117953911A
Apparatus and method for multiple-microphone speech enhancement
US20210125625A1
