Motor bearing fault diagnosis method and system based on multi-dimensional chromaticity symmetric dot matrix

Abstract

The invention discloses a motor bearing fault diagnosis method and system based on a multi-dimensional chromaticity symmetric dot matrix, and belongs to the technical field of bearing fault diagnosis, and the method comprises the steps: determining a fault mode needing to be diagnosed, and collecting a current signal and a multi-dimensional vibration signal when an electric agricultural machinery equipment drives a motor bearing to operate in each mode; performing time-frequency spectrum analysis on the collected current signal to obtain a rotating speed signal; decoupling the rotating speed signal from the multi-dimensional vibration signal to obtain a multi-dimensional reconstructed vibration signal; performing feature extraction on the obtained multi-dimensional reconstructed vibration signal, and taking a standard deviation as a color feature to obtain a multi-dimensional chromaticity symmetric dot matrix picture; and outputting a prediction result of fault diagnosis by taking the result as an input signal of the SE-CNN neural network. The method does not depend on a rotating speed sensor, and fault features can be accurately extracted in a working environment with strong bottom noise. The robustness and reliability of fault diagnosis are improved, and the application range of fault diagnosis is expanded.

Description

Technical Field

[0001] This invention relates to the technical field of bearing fault diagnosis, and specifically to a method and system for diagnosing motor bearing faults based on a multidimensional chromatic symmetric dot matrix. Background Technology

[0002] Electric agricultural machinery, with its powerful load-bearing capacity, efficient energy utilization, and zero-emission environmental characteristics, has been widely used in various fields such as agricultural operations and field transportation. The drive motor is one of the core components of electric agricultural machinery, converting electrical energy into mechanical energy to power the machinery. Bearings, as a core component, play a crucial role in supporting the rotor, reducing friction, transmitting torque, and maintaining the stable operation of the drive motor. Therefore, the performance of the bearings directly affects the continuity and safety of electric agricultural machinery transportation. However, the complex operating conditions and harsh road conditions of electric agricultural machinery subject the bearings to enormous drive loads and vibration impacts, resulting in a high failure rate of drive motor bearings. This poses a serious challenge to the driving safety and operational continuity of electric agricultural machinery.

[0003] Currently, there are few fault diagnosis methods specifically for electric agricultural machinery, with most focusing on small passenger vehicles. Fault diagnosis research methods for small passenger vehicles are mainly divided into model-based methods and data-driven methods. Model-based methods include digital twin models, dynamic models, and finite element models, while data-driven methods include convolutional neural networks, backpropagation neural networks, and multi-scale entropy. However, small passenger vehicles are mostly used on urban roads and for daily commuting, with relatively simple operating conditions and environments. In contrast, electric agricultural machinery operates under high loads and harsh environments, causing fault characteristic signals to be overwhelmed by noise, making fault diagnosis methods for small passenger vehicles difficult to adapt to electric agricultural machinery. Furthermore, the aforementioned methods are too costly and unsuitable for electric agricultural machinery projects with limited budgets or for retrofitting existing equipment. Therefore, a fault diagnosis solution specifically adapted to the operating characteristics of electric agricultural machinery is needed.

[0004] Current research methods for bearing fault diagnosis under variable speed conditions can be mainly divided into signal transformation-based methods and machine learning-based methods. Signal transformation-based methods include order analysis, wavelet transform, and variational mode decomposition, while machine learning-based methods include deep residual networks, transfer learning, and convolutional neural networks. These methods can flexibly handle bearing faults under variable speed conditions and achieve high accuracy. However, they all rely on high-precision speed sensors to acquire speed signals, which has limitations in practical applications. For budget-constrained electric agricultural machinery projects, this presents economic constraints, and for upgrading existing equipment, it can easily lead to structural compatibility issues. Therefore, a new method for bearing fault diagnosis under variable speed conditions is needed that does not rely on speed sensors. Summary of the Invention

[0005] The purpose of this invention is to provide a fault diagnosis method for motor bearings based on a multidimensional chromatic symmetric dot matrix. Compared with other fault diagnosis methods for drive motor bearings in electric agricultural machinery, this method does not rely on speed sensors and can accurately extract fault features in high-noise operating environments. This improves the robustness and reliability of fault diagnosis and expands its application scope.

[0006] Technical solution: To achieve the above objectives, the technical solution adopted by this invention is as follows:

[0007] A method for diagnosing motor bearing faults based on a multidimensional chromaticity symmetric dot matrix includes the following steps:

[0008] Step S1: Based on the engineering background, determine the fault modes that need to be diagnosed, and use the drive motor test bench to collect the current signal and multi-dimensional vibration signal of the drive motor bearing of the electric agricultural machinery equipment during operation in each mode.

[0009] Step S2: Perform time-frequency analysis on the current signal acquired in step S1 to obtain the rotational speed signal; decouple the rotational speed signal from the multi-dimensional vibration signal to obtain the multi-dimensional reconstructed vibration signal;

[0010] Step S3: Extract features from the multi-dimensional reconstructed vibration signal obtained in step S2, and use the standard deviation as a color feature to obtain a multi-dimensional chromaticity symmetrical dot matrix image;

[0011] Step S4: Use the multidimensional chromaticity symmetric dot matrix image obtained in step S3 as the input signal of the SE-CNN neural network, and output the prediction result of fault diagnosis.

[0012] Preferably, the specific steps of step S1 are as follows:

[0013] Step S101: Based on the analysis of the drive motor bearings of electric agricultural machinery, common fault types include inner ring fault, outer ring fault, and shaft imbalance fault.

[0014] Step S102: Turn on the vibrator on the experimental platform to simulate the background noise of electric agricultural machinery during actual operation. When the motor runs at variable speed, the acquisition card collects the vibration signal and current signal for each mode.

[0015] Preferably, the specific steps of step S2 are as follows:

[0016] Step S201: Perform time-frequency domain analysis on the current signal, extract the frequency components related to the rotational speed, and then calculate the motor speed to obtain the rotational speed signal;

[0017] Step S202: Map the frequency components of the multi-dimensional vibration signal to the order related to the rotational speed change, thereby eliminating the interference of rotational speed fluctuations on the signal and obtaining the multi-dimensional reconstructed vibration signal.

[0018] Preferred method for obtaining the rotational speed signal in step S201:

[0019] The motor's current signal is i M (τ), by performing a short-time Fourier transform, its time-frequency spectrum is obtained as follows:

[0020]

[0021] Among them, I M (t, f) represents the time spectrum of the current signal, where f is the frequency, j is the imaginary unit, t is the time variable, τ is the integration variable, and ω(τ-t) is the window function.

[0022] In the time spectrum, the dominant frequency of each time window is represented as:

[0023]

[0024] Among them, f i (t) is the dominant frequency of the current signal in the i-th time window, |I M (t i f)| is the t-th i The amplitude corresponding to frequency f within a time window;

[0025] Under variable speed operation, the speed of the drive motor changes with time. The speed is related to the motor's current frequency and the number of pole pairs. Therefore, the instantaneous speed of the motor can be calculated using the following formula:

[0026]

[0027] Where, p M n is the number of pole pairs of the motor. M (t) is the rotational speed of the motor in the i-th time window.

[0028] Preferred method: The method for obtaining the multi-dimensional reconstructed vibration signal in step S202 is as follows:

[0029] Motor speed signal n MN (t) Normalize to a standard frequency range, normalize the rotational speed signal n MN The formula for calculating (t) is:

[0030]

[0031] Where, n M-max and n M-minThese are the maximum and minimum values ​​in the speed signal, respectively.

[0032] By normalizing the speed signal n MN (t) can be used to calculate the order signal ω related to the rotational speed. M (t), to complete the reconstruction of the order signal, its calculation formula is:

[0033]

[0034] Where, n base The reference speed represents the standard speed set by the system, and the vibration signal x after order reconstruction. VOR (t) can be expressed as:

[0035] x VOR (t)=x V (t)·cos(2π·ω M (t))

[0036] Where, x VOR (t) represents the vibration signal after order reconstruction.

[0037] Preferably, the specific steps of step S3 are as follows:

[0038] The multi-dimensional reconstructed vibration signal obtained in step S2 is used as the input to the multi-dimensional chromaticity symmetric dot matrix model. Each petal of the multi-dimensional chromaticity symmetric dot matrix image of each mode is transformed from a vibration signal of one dimension. The standard deviation is an effective indicator reflecting the fluctuation amplitude of the vibration signal, so the standard deviation is selected as a new fault feature for colorization. The calculation formula for the multi-dimensional chromaticity symmetric dot matrix image is as follows:

[0039] 1) Algorithm Input

[0040] The input signal of the multidimensional chromaticity symmetric lattice algorithm is defined as v k (t)={x k,1 ,x k,2 ,x k,3 ...x k,j}, v k (t) represents the vibration signal in the k-th dimension, x k,j Let ξ represent the j-th data point in the k-th dimension, where k = 1, 2, 3, ξ is defined as the angle magnification factor, and a is defined as the time delay factor.

[0041] 2) Algorithm output

[0042] In the output signal of the multidimensional chromaticity symmetric lattice algorithm, r k (t) represents x k,j The radius in polar coordinates, ψ k (t) represents x k,jThe clockwise deflection angle in polar coordinates, φ k (t) represents x k,j In polar coordinates, the counterclockwise deflection angle, θ, represents x. k,j The mirror-symmetric rotation angle, c(t) represents x. k,j The color factor in polar coordinates, where X is the set of output images;

[0043] 3) Intermediate variables of the algorithm

[0044] x k,min Indicates v k The minimum value of (t), x k,max Indicates v k The maximum value of (t), σ (m) x represents k,j and the standard deviation of its local region;

[0045] 4) The calculation process for the formation of a multidimensional chromaticity symmetric dot matrix.

[0046] Calculation of shape parameters of multidimensional chromaticity symmetric lattice:

[0047] r k (t)=(x k,j -x k,min ) / (x k,max -x k,min )

[0048] ψ k (t)=θ+(x k,t+a -x k,min )·ξ / (x k,max -x k,min )

[0049] φ k (t)=θ-(x k,t+a -x k,min )·ξ / (x k,max -x k,min )

[0050] Calculation of color parameters for a multidimensional chromaticity symmetric dot matrix:

[0051] e k The set of local regions of size ε (t) is:

[0052] X ε =(x k,j-ε ...x k,j-2 ,x k,j-1 ,x k,j ,x k,j+1 ,x k,j+2 ...x k,j+ε )v k The standard deviation of (t) is:

[0053]

[0054] v k The color value of (t) is:

[0055] c (t) =σ (t)

[0056] Among them, c (t) This represents the color value.

[0057] Preferably, the SE-CNN neural network includes an input layer, a first convolutional layer, a first SE module, a first pooling layer, a second convolutional layer, a second SE module, a second pooling layer, a third convolutional layer, a third SE module, a third pooling layer, a fully connected layer, and a classification output layer, connected in sequence.

[0058] Another objective of this invention is to provide a fault diagnosis system for the drive motor bearings of electric agricultural machinery based on a multidimensional chromaticity symmetric dot matrix, comprising a data acquisition unit, a reconstruction unit, a multidimensional chromaticity symmetric dot matrix model unit, and an SE-CNN neural network unit, wherein:

[0059] The acquisition unit is used to determine the fault modes that need to be diagnosed based on the engineering background, and to collect the current signal and multi-dimensional vibration signal of the electric agricultural machinery equipment drive motor bearing in each mode of operation using the drive motor test bench.

[0060] The reconstruction unit is used to perform time-frequency analysis on the acquired current signal to obtain the rotational speed signal; and to decouple the rotational speed signal from the multi-dimensional vibration signal to obtain the multi-dimensional reconstructed vibration signal.

[0061] The multidimensional chromaticity symmetric dot matrix model unit is used to extract features from the acquired multidimensional reconstructed vibration signal and use the standard deviation as a color feature to obtain a multidimensional chromaticity symmetric dot matrix image.

[0062] The SE-CNN neural network unit is used to take the acquired multidimensional chromaticity symmetrical dot matrix image as the input signal of the SE-CNN neural network and output the prediction result of fault diagnosis.

[0063] Another object of the present invention is to provide an electronic device, comprising: at least one processor, at least one memory, and a communication interface; the processor, memory, and communication interface communicate with each other; the memory stores program instructions that can be executed by the processor, and the processor calls the program instructions to execute the motor bearing fault diagnosis method based on a multidimensional chromaticity symmetric dot matrix.

[0064] Another object of the present invention is to provide a non-transitory computer-readable storage medium storing computer instructions that cause the computer to execute the motor bearing fault diagnosis method based on a multi-dimensional chromaticity symmetric dot matrix.

[0065] Compared with the prior art, the present invention has the following advantages:

[0066] (1) The present invention is a motor bearing fault diagnosis method based on multidimensional chromatic symmetric dot matrix. It is suitable for fault diagnosis of drive motor bearings of electric agricultural machinery equipment, considering the complex working environment and development status of electric agricultural machinery equipment.

[0067] (2) This invention proposes a method for analyzing rotational speed through current signals, which is suitable for electric agricultural machinery equipment projects with limited budgets and the retrofitting and upgrading of existing vehicles.

[0068] (3) This invention proposes a feature extraction method for a multidimensional chromaticity symmetric dot matrix, which enhances the effectiveness of features and significantly improves the efficiency of fault diagnosis. Attached Figure Description

[0069] Figure 1 This is a flowchart of a method according to an embodiment of the present invention;

[0070] Figure 2 This is a schematic diagram of the SE-CNN neural network according to an embodiment of the present invention;

[0071] Figure 3 This is a schematic diagram of the experimental platform structure according to an embodiment of the present invention;

[0072] Figure 4 This is a schematic diagram of a drive motor bearing fault injection according to an embodiment of the present invention;

[0073] Figure 5 This is a schematic diagram of the current signal frequency spectrum according to an embodiment of the present invention;

[0074] Figure 6 This is a schematic diagram of the drive motor speed signal according to an embodiment of the present invention;

[0075] Figure 7 This is a schematic diagram of the order reconstruction of the original vibration data in an embodiment of the present invention;

[0076] Figure 8 This is a schematic diagram comparing the spectra of the original vibration signal and the order-reconstructed vibration signal according to an embodiment of the present invention.

[0077] Figure 9 Schematic diagrams of traditional symmetric lattices and multidimensional chromaticity symmetric lattices for training samples;

[0078] Figure 10This is a schematic diagram of the multidimensional symmetric dot matrix and the multidimensional chromaticity symmetric dot matrix of the training samples in an embodiment of the present invention.

[0079] Figure 11 This is a schematic diagram of the confusion matrix of the SE-CNN neural network. Detailed Implementation

[0080] The present invention will be further illustrated below with reference to the accompanying drawings and specific embodiments. It should be understood that these examples are for illustrative purposes only and are not intended to limit the scope of the invention. After reading this invention, any modifications of the invention in various equivalent forms by those skilled in the art will fall within the scope defined by the appended claims.

[0081] Example 1

[0082] To achieve bearing fault diagnosis under variable speed conditions without relying on a speed sensor, this embodiment provides a motor bearing fault diagnosis method based on a multi-dimensional chromaticity symmetric dot matrix, such as... Figure 1 As shown, it includes the following steps:

[0083] Step S1: Based on the engineering background, determine the fault modes that need to be diagnosed, and use the drive motor test bench to collect the current signal and multi-dimensional vibration signal of the drive motor bearing of the electric agricultural machinery equipment during operation in each mode.

[0084] Step S2: Perform time-frequency analysis on the current signal acquired in step S1 to obtain the rotational speed signal; decouple the rotational speed signal from the multi-dimensional vibration signal to obtain the multi-dimensional reconstructed vibration signal;

[0085] Step S3: Extract features from the multi-dimensional reconstructed vibration signal obtained in step S2, and use the standard deviation as a color feature to obtain a multi-dimensional chromaticity symmetrical dot matrix image;

[0086] Step S4: Use the multidimensional chromaticity symmetric dot matrix image obtained in step S3 as the input signal of the SE-CNN neural network, and output the prediction result of fault diagnosis.

[0087] Compared with other fault diagnosis methods, this invention does not require a large amount of computing resources, which can effectively reduce costs, improve robustness and reliability, and expand the application scope of fault diagnosis.

[0088] Example 2

[0089] This embodiment provides a method for diagnosing motor bearing faults based on a multidimensional chromaticity symmetric dot matrix, such as... Figure 1 As shown, it includes the following steps:

[0090] Step S1: Based on the engineering background, determine the fault modes that need to be diagnosed, and use the drive motor test bench to collect the current signal and multi-dimensional vibration signal of the drive motor bearing of the electric agricultural machinery equipment during operation in each mode.

[0091] Step S2: Perform time-frequency analysis on the current signal acquired in step S1, extract the frequency components related to rotational speed to obtain the rotational speed signal; decouple the rotational speed signal from the multi-dimensional vibration signal to obtain the multi-dimensional reconstructed vibration signal;

[0092] The specific steps of step S2 are as follows:

[0093] Step S201: Perform time-frequency domain analysis on the current signal, extract the frequency components related to the rotational speed, and then calculate the motor speed to obtain the rotational speed signal;

[0094] Electric agricultural machinery, as a type of heavy-duty electric transportation tool, uses a permanent magnet synchronous motor as its drive motor. The frequency converter of this drive motor controls the motor speed by changing the current frequency. To estimate the motor speed, time-frequency domain analysis of the current signal is required to extract the frequency components related to the speed, thereby calculating the motor speed. Assume the motor's current signal is i... M (τ), by performing a short-time Fourier transform, its time-frequency spectrum is obtained as follows:

[0095]

[0096] Among them, I M (t, f) represents the time spectrum of the current signal, where f is the frequency, j is the imaginary unit, t is the time variable, τ is the integral variable, and ω(τ-t) is the window function.

[0097] In the time spectrum, the dominant frequency of each time window is represented as...

[0098]

[0099] Among them, f i (t) is the dominant frequency of the current signal in the i-th time window, |I M (t i f)| is the t-th i The amplitude corresponding to frequency f within a time window.

[0100] Under variable speed operation, the speed of the drive motor changes with time. The speed is related to the motor's current frequency and the number of pole pairs. Therefore, the instantaneous speed of the motor can be calculated using the following formula:

[0101]

[0102] Where, p M n is the number of pole pairs of the motor.M (t) is the motor speed (in rpm) in the i-th time window.

[0103] Step S202: Map the frequency components of the multi-dimensional vibration signal to the order related to the rotational speed change, thereby eliminating the interference of rotational speed fluctuations on the signal and obtaining the multi-dimensional reconstructed vibration signal.

[0104] Order ratio analysis is a method suitable for analyzing signals under variable speed conditions. In the motor system of electric agricultural machinery, the spectrum of vibration signals exhibits characteristics that change with speed due to variations in rotational speed. This makes it difficult for traditional spectrum analysis methods to effectively extract stable fault features under dynamic conditions. Order ratio analysis eliminates the interference of speed fluctuations on the signal by mapping the frequency components of the vibration signal to an order ratio related to the speed change.

[0105] First, the motor speed signal n MN (t) is normalized to a standard frequency range for easier subsequent order calculation. Normalized speed signal n MN The formula for calculating (t) is:

[0106]

[0107] Where, n M-max and n M-min These are the maximum and minimum values ​​in the speed signal, respectively.

[0108] By normalizing the speed signal n MN (t) can be used to calculate the order signal ω related to the rotational speed. M (t), to complete the reconstruction of the order signal, its calculation formula is:

[0109]

[0110] Where, n base This is the reference rotational speed, representing the standard rotational speed set by the system. The vibration signal x after order reconstruction. VOR (t) can be expressed as:

[0111] x VOR (t)=x V (t)·cos(2π·ω M (t))

[0112] The vibration signal is mapped to the order frequency of the rotational speed, effectively eliminating the influence of rotational speed fluctuations. The signal x after order analysis and filtering... VOR (t) has high stability and can be used as input data for subsequent data processing.

[0113] Step S3: Extract features from the multi-dimensional reconstructed vibration signal obtained in step S2, and use the standard deviation as a color feature to obtain a multi-dimensional chromaticity symmetrical dot matrix image;

[0114] Step S4: Use the multidimensional chromaticity symmetric dot matrix image obtained in step S3 as the input signal of the SE-CNN neural network, and output the prediction result of fault diagnosis.

[0115] Example 3

[0116] This embodiment provides a method for diagnosing motor bearing faults based on a multidimensional chromaticity symmetric dot matrix, such as... Figure 1 As shown, it includes the following steps:

[0117] Step S1: Based on the engineering background, determine the fault modes that need to be diagnosed, and use the drive motor test bench to collect the current signal and multi-dimensional vibration signal of the drive motor bearing of the electric agricultural machinery equipment during operation in each mode.

[0118] Step S2: Perform time-frequency analysis on the current signal acquired in step S1 to obtain the rotational speed signal; decouple the rotational speed signal from the multi-dimensional vibration signal to obtain the multi-dimensional reconstructed vibration signal;

[0119] Step S3: Extract features from the multi-dimensional reconstructed vibration signal obtained in step S2, and use the standard deviation as a color feature to obtain a multi-dimensional chromaticity symmetrical dot matrix image;

[0120] Traditional symmetrical dot matrices can only convert one-dimensional signals into two-dimensional images. When the signal features are complex or the features are not significant enough, the effectiveness of the features in the generated symmetrical dot matrices decreases, failing to meet the needs of fault diagnosis. Therefore, to improve the effectiveness of features, feature extraction is performed on the multi-dimensional vibration signal obtained in step S2, and the standard deviation is used as a color feature to further enhance feature sensitivity.

[0121] The formula for calculating a multidimensional chromaticity symmetric dot matrix is:

[0122] 1) Algorithm Input

[0123] The input signal of the multidimensional chromaticity symmetric lattice algorithm is defined as v k (t)={x k,1 ,x k,2 ,x k,3 ...x k,j}, v k (t) represents the vibration signal in the k-th dimension, x k,j Let ξ represent the j-th data point in the k-th dimension, where k = 1, 2, 3, ξ is defined as the angle magnification factor, and a is defined as the time delay factor.

[0124] 2) Algorithm output

[0125] In the output signal of the multidimensional chromaticity symmetric lattice algorithm, r k (t) represents x k,j The radius in polar coordinates, ψ k (t) represents x k,j The clockwise deflection angle in polar coordinates, φ k (t) represents x k,j In polar coordinates, the counterclockwise deflection angle, θ, represents x. k,j The mirror-symmetric rotation angle, c(t) represents x. k,j The color factor in polar coordinates. X represents the set of output images.

[0126] 3) Intermediate variables of the algorithm

[0127] x k,min Indicates v k The minimum value of (t), x k,max Indicates v k The maximum value of (t), σ (m) x represents k,j and the standard deviation of its local area.

[0128] 4) The calculation process for the formation of a multidimensional chromaticity symmetric dot matrix.

[0129] Calculation of shape parameters of multidimensional chromaticity symmetric lattice:

[0130] r k (t)=(x k,j -x k,min ) / (x k,max -x k,min )

[0131] ψ k (t)=θ+(x k,t+a -x k,min )·ξ / (x k,max -x k,min )

[0132] φ k (t)=θ-(x k,t+a -x k,min )·ξ / (x k,max -x k,min )

[0133] Calculation of color parameters for a multidimensional chromaticity symmetric dot matrix:

[0134] e k The set of local regions of size ε (t) is:

[0135] X ε =(x k,j-ε ...x k,j-2 ,xk,j-1 ,x k,j ,x k,j+1 ,x k,j+2 ...x k,j+ε )v k The standard deviation of (t) is:

[0136]

[0137] v k The color value of (t) is:

[0138] c (t) =σ (t)

[0139] Among them, c (t) This represents the color value.

[0140] Step S4: Use the multidimensional chromaticity symmetric dot matrix image obtained in step S3 as the input signal of the SE-CNN neural network, and output the prediction result of fault diagnosis.

[0141] Example 4

[0142] This embodiment provides a method for diagnosing motor bearing faults based on a multidimensional chromaticity symmetric dot matrix, such as... Figure 1 As shown, it includes the following steps:

[0143] Step S1: Based on the engineering background, determine the fault modes that need to be diagnosed, and use the drive motor test bench to collect the current signal and multi-dimensional vibration signal of the drive motor bearing of the electric agricultural machinery equipment during operation in each mode.

[0144] Step S2: Perform time-frequency analysis on the current signal acquired in step S1 to obtain the rotational speed signal; decouple the rotational speed signal from the multi-dimensional vibration signal to obtain the multi-dimensional reconstructed vibration signal;

[0145] Step S3: Extract features from the multi-dimensional reconstructed vibration signal obtained in step S2, and use the standard deviation as a color feature to obtain a multi-dimensional chromaticity symmetrical dot matrix image;

[0146] Step S4: Use the multidimensional chromaticity symmetric dot matrix image obtained in step S3 as the input signal of the SE-CNN neural network, and output the prediction result of fault diagnosis.

[0147] Pattern recognition is performed on the multidimensional chromaticity symmetric dot matrix obtained in step S3. This invention employs SE-CNN for pattern recognition. By embedding an SE module into the traditional CNN, the sensitivity of the neural network to important features is enhanced, and noisy features are suppressed, thereby improving the accuracy of fault diagnosis. The structure of SE-CNN is as follows: Figure 2As shown, the neural network comprises 12 layers, consisting of an input layer, 3 convolutional layers, 3 SE modules, 3 pooling layers, a fully connected layer, and a classification output layer. The SE-CNN neural network includes an input layer, a first convolutional layer, a first SE module, a first pooling layer, a second convolutional layer, a second SE module, a second pooling layer, a third convolutional layer, a third SE module, a third pooling layer, a fully connected layer, and a classification output layer connected in sequence.

[0148] The details of SE-CNN are as follows:

[0149] 1) Input layer

[0150] Function: Input the generated set of multidimensional chromaticity symmetric dot matrix images X.

[0151] 2) Convolutional layer

[0152] Function: Feature extraction layer, which calculates the output by combining the convolution kernel with the local input region.

[0153] The expression for a convolutional layer is:

[0154]

[0155] in, Let i be the i-th input feature map of the li-th layer. The convolutional kernel of layer l connects the i-th input and the j-th output. Let f be the bias term of the j-th output of the l-th layer, and f be the activation function.

[0156] 3) SE module

[0157] Function: Adjust the feature response of each dimension, enhance the feature representation ability of important dimensions, and thus improve the overall performance of the model.

[0158] The SE module consists of a Squeeze phase and an Excitation phase. The expression for the Squeeze phase is:

[0159]

[0160] Among them, z c Let H be the dimension descriptor for the c-th dimension, H be the height of the feature map, W be the width of the feature map, and x be the dimension descriptor for the c-th dimension. c,i,j Let be the feature of the c-th dimension at position (i,j).

[0161] The expression for the Excitation phase is:

[0162] s = σ[W2·RELU(W1·z)]

[0163] Where s is the dimension weight vector, σ is the Sigmoid activation function, W1 and W2 are the weight matrices of the two fully connected layers, ReLU is the ReLU activation function, and z is the output of the Squeeze stage.

[0164] 4) Pooling layer

[0165] Function: Downsampling reduces feature map size and improves the network's generalization ability.

[0166] The expression for the pooling layer is:

[0167]

[0168] in, is the scaling factor, and down is the downsampling function.

[0169] 5) Fully connected layer and output layer

[0170] Function: Maps features to a sample label space for classification.

[0171] The expressions for the fully connected layer and the output layer are:

[0172] y = Wx + b

[0173] Where x is the input vector, W is the weight matrix, and b is the bias vector.

[0174] Example 5

[0175] This embodiment provides a method for diagnosing motor bearing faults based on a multidimensional chromaticity symmetric dot matrix, such as... Figure 1 As shown, it includes the following steps:

[0176] Step S1: Based on the engineering background, determine the fault modes that need to be diagnosed, and use the drive motor test bench to collect the current signal and multi-dimensional vibration signal of the drive motor bearing of the electric agricultural machinery equipment during operation in each mode.

[0177] The specific steps are as follows:

[0178] S101: Based on the analysis of the drive motor bearings of electric agricultural machinery, common fault types include inner ring faults, outer ring faults, and shaft imbalance faults.

[0179] S102: The current signal and multi-dimensional vibration signal of each mode are collected by the drive motor test bench. The vibrator on the test bench is turned on to simulate the background noise of electric agricultural machinery during actual operation. When the motor runs at variable speed, the acquisition card collects 50 sets of vibration signals and current signals for each mode.

[0180] Step S2: Perform time-frequency analysis on the current signal acquired in step S1 to obtain the rotational speed signal; decouple the rotational speed signal from the multi-dimensional vibration signal to obtain the multi-dimensional reconstructed vibration signal;

[0181] The specific steps of step S2 are as follows:

[0182] Step S201: Perform time-frequency domain analysis on the current signal, extract the frequency components related to the rotational speed, and then calculate the motor speed to obtain the rotational speed signal;

[0183] Electric agricultural machinery, as a type of heavy-duty electric transportation tool, uses a permanent magnet synchronous motor as its drive motor. The frequency converter of this drive motor controls the motor speed by changing the current frequency. To estimate the motor speed, time-frequency domain analysis of the current signal is required to extract the frequency components related to the speed, thereby calculating the motor speed. Assume the motor's current signal is i... M (τ), by performing a short-time Fourier transform, its time-frequency spectrum is obtained as follows:

[0184]

[0185] Among them, I M (t, f) represents the time spectrum of the current signal, where f is the frequency, j is the imaginary unit, t is the time variable, τ is the integral variable, and ω(τ-t) is the window function.

[0186] In the time spectrum, the dominant frequency of each time window is represented as...

[0187]

[0188] Among them, f i (t) is the dominant frequency of the current signal in the i-th time window, |I M (t i f)| is the t-th i The amplitude corresponding to frequency f within a time window.

[0189] Under variable speed operation, the speed of the drive motor changes with time. The speed is related to the motor's current frequency and the number of pole pairs. Therefore, the instantaneous speed of the motor can be calculated using the following formula:

[0190]

[0191] Where, p M n is the number of pole pairs of the motor. M (t) is the motor speed (in rpm) in the i-th time window.

[0192] Step S202: Map the frequency components of the multi-dimensional vibration signal to the order related to the rotational speed change, thereby eliminating the interference of rotational speed fluctuations on the signal and obtaining the multi-dimensional reconstructed vibration signal.

[0193] Order ratio analysis is a method suitable for analyzing signals under variable speed conditions. In the motor system of electric agricultural machinery, the spectrum of vibration signals exhibits characteristics that change with speed due to variations in rotational speed. This makes it difficult for traditional spectrum analysis methods to effectively extract stable fault features under dynamic conditions. Order ratio analysis eliminates the interference of speed fluctuations on the signal by mapping the frequency components of the vibration signal to an order ratio related to the speed change.

[0194] First, the motor speed signal n MN (t) is normalized to a standard frequency range for easier subsequent order calculation. Normalized speed signal n MN The formula for calculating (t) is:

[0195]

[0196] Where, n M-max and n M-min These are the maximum and minimum values ​​in the speed signal, respectively.

[0197] By normalizing the speed signal n MN (t) can be used to calculate the order signal ω related to the rotational speed. M (t), to complete the reconstruction of the order signal, its calculation formula is:

[0198]

[0199] Where, n base This is the reference rotational speed, representing the standard rotational speed set by the system. The vibration signal x after order reconstruction. VOR (t) can be expressed as:

[0200] x VOR (t)=x V (t)·cos(2π·ω M (t))

[0201] The vibration signal is mapped to the order frequency of the rotational speed, effectively eliminating the influence of rotational speed fluctuations. The signal x after order analysis and filtering... VOR (t) has high stability and can be used as input data for subsequent data processing.

[0202] Step S3: Extract features from the multi-dimensional reconstructed vibration signal obtained in step S2, and use the standard deviation as a color feature to obtain a multi-dimensional chromaticity symmetrical dot matrix image;

[0203] Traditional symmetrical dot matrices can only convert one-dimensional signals into two-dimensional images. When the signal features are complex or the features are not significant enough, the effectiveness of the features in the generated symmetrical dot matrices decreases, failing to meet the needs of fault diagnosis. Therefore, to improve the effectiveness of features, feature extraction is performed on the multi-dimensional vibration signal obtained in step S2, and the standard deviation is used as a color feature to further enhance feature sensitivity.

[0204] The formula for calculating a multidimensional chromaticity symmetric dot matrix is:

[0205] 1) Algorithm Input

[0206] The input signal of the multidimensional chromaticity symmetric lattice algorithm is defined as v k (t)={x k,1 ,x k,2 ,x k,3 ...x k,j}, v k (t) represents the vibration signal in the k-th dimension, x k,j Let ξ represent the j-th data point in the k-th dimension, where k = 1, 2, 3, ξ is defined as the angle magnification factor, and a is defined as the time delay factor.

[0207] 2) Algorithm output

[0208] In the output signal of the multidimensional chromaticity symmetric lattice algorithm, r k (t) represents x k,j The radius in polar coordinates, ψ k (t) represents x k,j The clockwise deflection angle in polar coordinates, φ k (t) represents x k,j In polar coordinates, the counterclockwise deflection angle, θ, represents x. k,j The mirror-symmetric rotation angle, c(t) represents x. k,j The color factor in polar coordinates. X represents the set of output images.

[0209] 3) Intermediate variables of the algorithm

[0210] x k,min Indicates v k The minimum value of (t), x k,max Indicates v k The maximum value of (t), σ (m) x represents k,j and the standard deviation of its local area.

[0211] 4) The calculation process for the formation of a multidimensional chromaticity symmetric dot matrix.

[0212] Calculation of shape parameters of multidimensional chromaticity symmetric lattice:

[0213] r k(t)=(x k,j -x k,min ) / (x k,max -x k,min )

[0214] ψ k (t)=θ+(x k,t+a -x k,min )·ξ / (x k,max -x k,min )

[0215] φ k (t)=θ-(x k,t+a -x k,min )·ξ / (x k,max -x k,min )

[0216] Calculation of color parameters for a multidimensional chromaticity symmetric dot matrix:

[0217] e k The set of local regions of size ε (t) is:

[0218] X ε =(x k,j-ε ...x k,j-2 ,x k,j-1 ,x k,j ,x k,j+1 ,x k,j+2 ...x k,j+ε )v k The standard deviation of (t) is:

[0219]

[0220] v k The color value of (t) is:

[0221] c (t) =σ (t)

[0222] Among them, c (t) This represents the color value.

[0223] Step S4: Use the multidimensional chromaticity symmetric dot matrix image obtained in step S3 as the input signal of the SE-CNN neural network, and output the prediction result of fault diagnosis.

[0224] Pattern recognition is performed on the multidimensional chromaticity symmetric dot matrix obtained in step S3. This invention employs SE-CNN for pattern recognition. By embedding an SE module into the traditional CNN, the sensitivity of the neural network to important features is enhanced, and noisy features are suppressed, thereby improving the accuracy of fault diagnosis. The structure of SE-CNN is as follows: Figure 2As shown, the neural network comprises 12 layers, consisting of an input layer, 3 convolutional layers, 3 SE modules, 3 pooling layers, a fully connected layer, and a classification output layer. The SE-CNN neural network includes an input layer, a first convolutional layer, a first SE module, a first pooling layer, a second convolutional layer, a second SE module, a second pooling layer, a third convolutional layer, a third SE module, a third pooling layer, a fully connected layer, and a classification output layer connected in sequence.

[0225] The details of SE-CNN are as follows:

[0226] 1) Input layer

[0227] Function: Input the generated set of multidimensional chromaticity symmetric dot matrix images X.

[0228] 2) Convolutional layer

[0229] Function: Feature extraction layer, which calculates the output by combining the convolution kernel with the local input region.

[0230] The expression for a convolutional layer is:

[0231]

[0232] in, Let i be the i-th input feature map of the li-th layer. The convolutional kernel of layer l connects the i-th input and the j-th output. Let f be the bias term of the j-th output of the l-th layer, and f be the activation function.

[0233] 3) SE module

[0234] Function: Adjust the feature response of each dimension, enhance the feature representation ability of important dimensions, and thus improve the overall performance of the model.

[0235] The SE module consists of a Squeeze phase and an Excitation phase. The expression for the Squeeze phase is:

[0236]

[0237] Among them, z c Let H be the dimension descriptor for the c-th dimension, H be the height of the feature map, W be the width of the feature map, and x be the dimension descriptor for the c-th dimension. c,i,j Let be the feature of the c-th dimension at position (i,j).

[0238] The expression for the Excitation phase is:

[0239] s = σ[W2·RELU(W1·z)]

[0240] Where s is the dimension weight vector, σ is the Sigmoid activation function, W1 and W2 are the weight matrices of the two fully connected layers, ReLU is the ReLU activation function, and z is the output of the Squeeze stage.

[0241] 4) Pooling layer

[0242] Function: Downsampling reduces feature map size and improves the network's generalization ability.

[0243] The expression for the pooling layer is:

[0244]

[0245] in, is the scaling factor, and down is the downsampling function.

[0246] 5) Fully connected layer and output layer

[0247] Function: Maps features to a sample label space for classification.

[0248] The expressions for the fully connected layer and the output layer are:

[0249] y = Wx + b

[0250] Where x is the input vector, W is the weight matrix, and b is the bias vector.

[0251] Example 6

[0252] This embodiment provides a fault diagnosis system for the drive motor bearings of electric agricultural machinery based on a multi-dimensional chromatic symmetric dot matrix, such as... Figure 1 As shown, it includes an acquisition unit, a reconstruction unit, a multidimensional chromaticity symmetric dot matrix model unit, and an SE-CNN neural network unit, wherein:

[0253] The acquisition unit is used to determine the fault modes that need to be diagnosed based on the engineering background, and to collect the current signal and multi-dimensional vibration signal of the electric agricultural machinery equipment drive motor bearing in each mode of operation using the drive motor test bench.

[0254] The reconstruction unit is used to perform time-frequency analysis on the acquired current signal to obtain the rotational speed signal; and to decouple the rotational speed signal from the multi-dimensional vibration signal to obtain the multi-dimensional reconstructed vibration signal.

[0255] The multidimensional chromaticity symmetric dot matrix model unit is used to extract features from the acquired multidimensional reconstructed vibration signal and use the standard deviation as a color feature to obtain a multidimensional chromaticity symmetric dot matrix image.

[0256] The SE-CNN neural network unit is used to take the acquired multidimensional chromaticity symmetrical dot matrix image as the input signal of the SE-CNN neural network and output the prediction result of fault diagnosis.

[0257] Example 7

[0258] This embodiment provides an electronic device, including: at least one processor, at least one memory, and a communication interface; the processor, memory, and communication interface communicate with each other; the memory stores program instructions that can be executed by the processor, and the processor calls the program instructions to execute the motor bearing fault diagnosis method based on a multidimensional chromaticity symmetric dot matrix.

[0259] Example 8

[0260] This embodiment provides a non-transitory computer-readable storage medium that stores computer instructions that cause the computer to execute the motor bearing fault diagnosis method based on a multi-dimensional chromaticity symmetric dot matrix.

[0261] Example 9

[0262] This embodiment is a verification example of the present invention. The following description, in conjunction with specific experimental data, illustrates the embodiment of the present invention, employing methods such as... Figure 3 The experimental setup shown is used for data acquisition. A 24V DC power supply powers the drive motor, a vibrator simulates background noise, and a motor control module connects to a host computer to control the drive motor to move at speeds ranging from 1000 rpm to 3000 rpm. The experimental setup uses current and vibration sensors. The data acquisition module connects to the sensors to collect current and vibration data and then uploads it to the host computer. To verify the effectiveness of the proposed method, three faults are introduced into the external bearing of the drive motor: inner loop fault, outer loop fault, and shaft imbalance fault. Figure 4 As shown.

[0263] The data acquisition card collected 50 sets of vibration and current signals for each mode. The sampling frequency was 8kHz, the sampling time was 5s, and the data sample length was 40,000. The modes to be identified in this study were: normal, inner loop fault, outer loop fault, and shaft imbalance fault. For each mode, 40 sets of signals were selected as training samples, for a total of 160 sets. For each mode, 10 sets were selected as test samples, for a total of 40 sets. The detailed experimental dataset is shown in Table 1.

[0264] Table 1 Description of training and test samples

[0265]

[0266] The current-based time-frequency spectrum analysis method for estimating motor speed can effectively extract frequency components related to motor speed changes from the current signal, thereby obtaining the motor speed signal. First, a short-time Fourier transform is performed on the current signal to obtain the time-frequency spectrum. Then, the frequency corresponding to the maximum amplitude in each time window is found, and the speed signal is obtained based on the relationship between frequency and speed. Figure 5 The time-frequency spectrum obtained is used as an example, based on a sample of a motor bearing in normal mode. Figure 6 This is the drive motor speed signal obtained from time-frequency spectrum analysis.

[0267] Raw vibration data and reconstructed vibration data, such as Figure 7 As shown in the image, the blue original vibration signal exhibits complex and highly fluctuating characteristics, reflecting the motor's vibration under variable speed conditions. In contrast, the green reconstructed signal from the order analysis is more stable with smaller fluctuations. This indicates that the order analysis successfully removed the influence of speed variations from the reconstructed signal, resulting in a more regular waveform. This stable signal better matches the signal characteristics under uniform speed conditions, which is helpful for subsequent fault feature extraction and pattern recognition.

[0268] Meanwhile, to verify the advantages of frequency domain reconstructed signals, especially their mitigation of spectral line blurring caused by velocity fluctuations, the frequency spectra of the original vibration signal and the reconstructed vibration signal were also plotted, such as... Figure 8 As shown. Comparison Figure 8 (a) and Figure 8 (b) It can be seen that after the order analysis reconstruction, the clarity of the frequencies in the orange box is not only enhanced, but also the spectral energy caused by velocity fluctuations and noise is reduced.

[0269] To verify the effectiveness of the multidimensional chromaticity symmetric dot matrix for different pattern recognition methods, symmetric dot matrices and multidimensional chromaticity symmetric dot matrices were first extracted from the 1st and 15th training samples of all pattern training sets, respectively. Figure 8 The paper presents training samples of symmetric lattices and multidimensional chromaticity symmetric lattices in four modes. It can be found that the multidimensional chromaticity symmetric lattice shows more obvious differences in shape than the symmetric lattice, but the differences between the normal mode and the imbalance fault are smaller.

[0270] Then, multidimensional chromaticity symmetric lattices were extracted from the 1st and 15th training samples of all modes, respectively. Table 2 shows the multidimensional symmetric lattices and multidimensional chromaticity symmetric lattice training samples for the four modes. Figure 9 The images show obvious differences in shape and color, effectively identifying all patterns. Moreover, the multidimensional chromatic symmetric dot matrix images of the same pattern are almost similar, showing that this feature has a certain stability for different samples.

[0271] This study uses a SE-CNN neural network to classify test samples of multidimensional chromatic symmetric dot matrices. The input image size is 224×224×3, and after processing, the size is 1×1×64. Fully connected layers and classification layers are then used for classification. The activation function is ReLU, the classifier is Softmax, the maximum number of training iterations is 100, and the learning rate is 0.0015.

[0272] The confusion matrix of the test samples after fault diagnosis by the SE-CNN neural network is as follows: Figure 11 As shown, out of 40 test samples, only one sample was incorrectly identified, resulting in a fault diagnosis accuracy rate of 97.5%.

[0273] This invention extracts rotational speed signals based on the time-frequency spectrum of current signals. Simultaneously, it decouples the rotational speed signal from the vibration signal using order ratio analysis. Then, it extracts features based on multi-dimensional vibration signals and uses the standard deviation as a color feature. Finally, it embeds an SE module into a CNN neural network to enhance the network's sensitivity to color features. Based on the SE-CNN neural network, it identifies fault modes in the extracted multi-dimensional chromatic symmetric dot matrix image, enabling fault diagnosis of drive motor bearings in electric agricultural machinery. Compared with other fault diagnosis methods, this invention requires less computational resources, effectively reducing costs, improving robustness and reliability, and expanding the application scope of fault diagnosis.

[0274] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for diagnosing motor bearing faults based on a multidimensional chromaticity symmetric dot matrix, characterized in that, Includes the following steps: Step S1: Based on the engineering background, determine the fault modes that need to be diagnosed, and use the drive motor test bench to collect the current signal and multi-dimensional vibration signal of the drive motor bearing of the electric agricultural machinery equipment during operation in each mode. Step S2: Perform time-frequency analysis on the current signal acquired in step S1 to obtain the rotational speed signal; decouple the rotational speed signal from the multi-dimensional vibration signal to obtain the multi-dimensional reconstructed vibration signal; Step S3: Extract features from the multi-dimensional reconstructed vibration signal obtained in step S2, and use the standard deviation as a color feature to obtain a multi-dimensional chromaticity symmetrical dot matrix image; Step S4: Use the multidimensional chromaticity symmetric dot matrix image obtained in step S3 as the input signal of the SE-CNN neural network, and output the prediction result of fault diagnosis.

2. The motor bearing fault diagnosis method based on a multidimensional chromaticity symmetric dot matrix according to claim 1, characterized in that: The specific steps of step S1 are as follows: Step S101: Based on the analysis of the drive motor bearings of electric agricultural machinery, common fault types include inner ring fault, outer ring fault, and shaft imbalance fault. Step S102: Turn on the vibrator on the experimental platform to simulate the background noise of electric agricultural machinery during actual operation. When the motor runs at variable speed, the acquisition card collects the vibration signal and current signal for each mode.

3. The motor bearing fault diagnosis method based on a multidimensional chromaticity symmetric dot matrix according to claim 1, characterized in that: The specific steps of step S2 are as follows: Step S201: Perform time-frequency domain analysis on the current signal, extract the frequency components related to the rotational speed, and then calculate the motor speed to obtain the rotational speed signal; Step S202: Map the frequency components of the multi-dimensional vibration signal to the order related to the rotational speed change, thereby eliminating the interference of rotational speed fluctuations on the signal and obtaining the multi-dimensional reconstructed vibration signal.

4. The motor bearing fault diagnosis method based on a multidimensional chromaticity symmetric dot matrix according to claim 1, characterized in that: The method for obtaining the rotational speed signal in step S201 is as follows: The motor's current signal is i M (τ), by performing a short-time Fourier transform, its time-frequency spectrum is obtained as follows: Among them, I M (t, f) represents the time spectrum of the current signal, where f is the frequency, j is the imaginary unit, t is the time variable, τ is the integration variable, and ω(τ-t) is the window function. In the time spectrum, the dominant frequency of each time window is represented as: Among them, f i (t) is the dominant frequency of the current signal in the i-th time window, |I M (t i f)| is the t-th i The amplitude corresponding to frequency f within a time window; Under variable speed operation, the speed of the drive motor changes with time. The speed is related to the motor's current frequency and the number of pole pairs. Therefore, the instantaneous speed of the motor can be calculated using the following formula: Where, p M n is the number of pole pairs of the motor. M (t) is the rotational speed of the motor in the i-th time window.

5. The motor bearing fault diagnosis method based on a multidimensional chromaticity symmetric dot matrix according to claim 1, characterized in that: The method for obtaining the multi-dimensional reconstructed vibration signal in step S202 is as follows: Motor speed signal n MN (t) Normalize to a standard frequency range, normalize the rotational speed signal n MN The formula for calculating (t) is: Where, n M-max and n M-min These are the maximum and minimum values ​​in the speed signal, respectively. By normalizing the speed signal n MN (t) can be used to calculate the order signal ω related to the rotational speed. M (t), to complete the reconstruction of the order signal, its calculation formula is: Where, n base The reference speed represents the standard speed set by the system, and the vibration signal x after order reconstruction. VOR (t) can be expressed as: x VOR (t)=x V (t)·cos(2π·ω M (t)) Where, x VOR (t) represents the vibration signal after order reconstruction.

6. The motor bearing fault diagnosis method based on a multidimensional chromaticity symmetric dot matrix according to claim 1, characterized in that: The specific steps of step S3 are as follows: The multi-dimensional reconstructed vibration signal obtained in step S2 is used as the input to the multi-dimensional chromaticity symmetric dot matrix model. Each petal of the multi-dimensional chromaticity symmetric dot matrix image of each mode is transformed from a vibration signal of one dimension. The standard deviation is an effective indicator reflecting the fluctuation amplitude of the vibration signal, so the standard deviation is selected as a new fault feature for colorization. The calculation formula for the multi-dimensional chromaticity symmetric dot matrix image is as follows: 1) Algorithm Input The input signal of the multidimensional chromaticity symmetric lattice algorithm is defined as v k (t)={x k,1 ,x k,2 ,x k,3 ...x k,j }, v k (t) represents the vibration signal in the k-th dimension, x k,j Let ξ represent the j-th data point in the k-th dimension, where k = 1, 2, 3, ξ is defined as the angle magnification factor, and a is defined as the time delay factor. 2) Algorithm output In the output signal of the multidimensional chromaticity symmetric lattice algorithm, r k (t) represents x k,j The radius in polar coordinates, ψ k (t) represents x k,j The clockwise deflection angle in polar coordinates, φ k (t) represents x k,j In polar coordinates, the counterclockwise deflection angle, θ, represents x. k,j The mirror-symmetric rotation angle, c(t) represents x. k,j The color factor in polar coordinates, where X is the set of output images; 3) Intermediate variables of the algorithm x k,min Indicates v k The minimum value of (t), x k,max Indicates v k The maximum value of (t), σ (m) x represents k,j and the standard deviation of its local region; 4) The calculation process for the formation of a multidimensional chromaticity symmetric dot matrix. Calculation of shape parameters of multidimensional chromaticity symmetric lattice: r k (t)=(x k,j -x k,min ) / (x k,max -x k,min ) ψ k (t)=θ+(x k,t+a -x k,min )·ξ / (x k,max -x k,min ) f k (t)=θ-(x k,t+a -x k,min )·ξ / (x k,max -x k,min ) Calculation of color parameters for a multidimensional chromaticity symmetric dot matrix: e k The set of local regions of size ε (t) is: X ε =(x k,j-ε ...x k,j-2 ,x k,j-1 ,x k,j ,x k,j+1 ,x k,j+2 ...x k,j+ε ) v k The standard deviation of (t) is: v k The color value of (t) is: c (t) =s (t) Among them, c (t) This represents the color value.

7. The motor bearing fault diagnosis method based on a multidimensional chromaticity symmetric dot matrix according to claim 1, characterized in that: The SE-CNN neural network includes an input layer, a first convolutional layer, a first SE module, a first pooling layer, a second convolutional layer, a second SE module, a second pooling layer, a third convolutional layer, a third SE module, a third pooling layer, a fully connected layer, and a classification output layer, all connected in sequence.

8. A fault diagnosis system for the drive motor bearings of electric agricultural machinery based on a multidimensional chromatic symmetric dot matrix, characterized in that: It includes an acquisition unit, a reconstruction unit, a multidimensional chromaticity symmetric dot matrix model unit, and an SE-CNN neural network unit, wherein: The acquisition unit is used to determine the fault modes that need to be diagnosed based on the engineering background, and to collect the current signal and multi-dimensional vibration signal of the electric agricultural machinery equipment drive motor bearing in each mode of operation using the drive motor test bench. The reconstruction unit is used to perform time-frequency analysis on the acquired current signal to obtain the rotational speed signal; and to decouple the rotational speed signal from the multi-dimensional vibration signal to obtain the multi-dimensional reconstructed vibration signal. The multidimensional chromaticity symmetric dot matrix model unit is used to extract features from the acquired multidimensional reconstructed vibration signal and use the standard deviation as a color feature to obtain a multidimensional chromaticity symmetric dot matrix image. The SE-CNN neural network unit is used to take the acquired multidimensional chromaticity symmetrical dot matrix image as the input signal of the SE-CNN neural network and output the prediction result of fault diagnosis.

9. An electronic device, characterized in that, include: At least one processor, at least one memory, and a communication interface; The processor, memory, and communication interface communicate with each other; The memory stores program instructions that can be executed by the processor, which calls the program instructions to execute the motor bearing fault diagnosis method based on a multidimensional chromaticity symmetric dot matrix as described in any of claims 1-7.

10. A non-transitory computer-readable storage medium, characterized in that, The non-transitory computer-readable storage medium stores computer instructions that cause the computer to execute the motor bearing fault diagnosis method based on a multidimensional chromaticity symmetric dot matrix as described in any one of claims 1-7.

Images