Fault diagnosis method and system applied to automobile air conditioner compressor

Abstract

The invention relates to the technical field of fault diagnosis, and discloses a fault diagnosis method and system applied to an automobile air conditioner compressor. The method comprises the following steps: firstly, introducing auxiliary noise to decompose a vibration signal, calculating a decomposition result by using a sample entropy method to obtain a sample entropy, and identifying and removing noise; secondly, variational mode decomposition is carried out on the vibration signals after noise elimination, an improved crayfish optimization algorithm is used for optimizing a decomposition scale and a penalty factor in the variational mode decomposition process, and a final IMF component signal set is obtained; mapping the final IMF component signal set to polar coordinates through a symmetric polar coordinate transformation method to form a fault diagnosis image set; and finally, training to obtain a convolutional neural network fault diagnosis model, identifying a fault diagnosis image set, and outputting a fault diagnosis result of the automobile air conditioner compressor. The vibration signal of the automobile air conditioner compressor is processed and analyzed, the purpose of fault diagnosis is achieved, and the method is objective and accurate.

Description

Technical Field

[0001] The present invention relates to the technical field of fault diagnosis, and in particular to a fault diagnosis method and system applied to an automobile air-conditioning compressor. Background Art

[0002] Chinese patent CN113435111B discloses a reciprocating compressor fault diagnosis method and system, which specifically includes: obtaining dynamic pressure data in the compressor cylinder, performing segmented preprocessing on the dynamic pressure data in the compressor cylinder according to different working processes of the compressor, and obtaining pV diagrams of different working processes; extracting features from the pV diagrams of the different working processes, obtaining features of the different working processes, and combining the features of the different working processes into a comprehensive feature vector; training to obtain a pre-trained neural network, inputting the comprehensive feature vector into the pre-trained neural network, outputting the result, and obtaining the mapping relationship between the comprehensive output result and the compressor state to obtain the compressor state type, thereby realizing compressor fault diagnosis. However, the invention does not further process the acquired data, resulting in slow processing speed and errors in the results.

[0003] Traditional compressor fault diagnosis methods usually rely on the experience of technicians to perform fault detection, which makes it difficult to accurately detect fault problems and affects the effective operation of the compressor. At the same time, due to the lack of the use of technologies such as machine learning, no model is established in the fault diagnosis process of the automotive air-conditioning compressor, and no neural network technology is used. The fault diagnosis speed is slow and maintenance plans cannot be made in a timely manner. Summary of the invention

[0004] In view of the problems in the related art, the present invention provides a fault diagnosis method and system applied to an automobile air-conditioning compressor to overcome the above-mentioned technical problems existing in the existing related art.

[0005] In order to solve the above technical problems, the present invention is achieved through the following technical solutions: The present invention is a fault diagnosis method for an automobile air-conditioning compressor, comprising the following steps: S1. Obtain the vibration signal of the automobile air-conditioning compressor during operation to obtain an initial automobile air-conditioning compressor vibration signal set, introduce auxiliary noise and fuse the sample entropy method, identify and remove the vibration signal noise in the initial automobile air-conditioning compressor vibration signal set, and obtain a processed automobile air-conditioning compressor vibration signal set; S2. performing variational modal decomposition on the processed automobile air-conditioning compressor vibration signal set, combining chaos mapping and dimensional learning, and using an improved crayfish optimization algorithm to optimize the decomposition scale and penalty factor in the variational modal decomposition process to obtain the global optimal solution of the algorithm; S3, using the global optimal solution of the algorithm as the optimized decomposition scale and the optimized penalty factor, and then converting the processed automobile air-conditioning compressor vibration signal set into a final IMF component signal set, and using the symmetric polar coordinate transformation method to convert the final IMF component signal set into a fault diagnosis image set; S4. Train a convolutional neural network fault diagnosis model, combine the fault diagnosis image set and the convolutional neural network fault diagnosis model, and output a fault diagnosis result of the automobile air-conditioning compressor.

[0006] The invention obtains the initial automobile air-conditioning compressor vibration signal set, introduces auxiliary noise for decomposition processing, and then uses the sample entropy method to calculate the sample entropy on the decomposition result, and then divides the initial automobile air-conditioning compressor vibration signal set to eliminate the vibration signal noise; this method can fully decompose the vibration signal. Compared with the direct noise elimination method, it reduces the calculation time by screening the noise, greatly improving the denoising effect and efficiency; secondly, in the variational mode decomposition process, the improved crayfish optimization algorithm is used to optimize the decomposition scale and penalty factor, and the optimal fitness function value is found by imitating the foraging, summer escape and competition behavior of freshwater crayfish to obtain the optimal solution of the model. The algorithm introduces chaotic mapping to increase the randomness of initialization and the algorithm search capability, and introduces dimensional learning to overcome the The traditional optimization algorithm has the disadvantages of slow convergence speed and easy to fall into local optimal solution, and the optimization effect is good; after variational mode decomposition, the final IMF component signal set is obtained, and the final IMF component signal set is mapped to polar coordinates through the symmetric polar coordinate transformation method to form a fault diagnosis image; compared with the direct identification of IMF component signals, this method has obvious advantages, and the information of IMF component signals is intuitively converted to images, which can improve the accuracy of fault identification and diagnosis and facilitate subsequent processing; finally, the convolutional neural network fault diagnosis model is trained to output the fault diagnosis results of the automobile air-conditioning compressor, and complete the fault diagnosis of the automobile air-conditioning compressor; this neural network is suitable for image recognition, and its classification performance is better than other neural networks under the premise of less time consumption, and it can better realize high-precision fault diagnosis.

[0007] Preferably, the S1 comprises the following steps: S11, placing a signal acquisition device on the automobile air-conditioning compressor to be tested, acquiring a vibration signal of the automobile air-conditioning compressor when it is running, and obtaining an initial automobile air-conditioning compressor vibration signal set, wherein the initial automobile air-conditioning compressor vibration signal in the initial automobile air-conditioning compressor vibration signal set is The initial automobile air-conditioning compressor vibration signal subset is truncated to form, and the corresponding signal time series is set according to the vibration signal acquisition time; the initial automobile air-conditioning compressor vibration signal set is decomposed by introducing auxiliary noise to obtain the vibration signal decomposition vector. The specific steps are as follows: S111, adding auxiliary noise to the initial automobile air-conditioning compressor vibration signal set to obtain a mixed vibration signal set, performing empirical mode decomposition on the auxiliary noise of the mixed vibration signal set to obtain a first IMF component signal set; setting the adding times threshold to , add auxiliary noise to the mixed vibration signal set to obtain the second IMF component signal set, until the number of additions reaches the addition number threshold, obtain several IMF component signal sets, solve the average value of several IMF component signal sets, and record it as the first stage IMF component signal; S112, using the initial automobile air-conditioning compressor vibration signal set to subtract the first-stage IMF component signal to obtain the first-stage vibration signal set, adding auxiliary noise to the first-stage vibration signal set, repeating S111, and obtaining the second-stage IMF component signal; iteratively decomposing the vibration signal set until the vibration signal in the vibration signal set is less than or equal to 2, stopping the iteration, and obtaining the vibration signal decomposition vector, the calculation formula is as follows: ; in, represents the initial automobile air-conditioning compressor vibration signal set, represents the IMF component signal of the i-th stage, represents the residual decomposition vector, b represents the number of iterations of the vibration signal set, , the IMF component signals and residual decomposition vectors at different stages constitute the vibration signal decomposition vector; S12, determining the signal time series corresponding to the IMF component signals and the residual decomposition vectors at different stages, respectively recording them as the first signal time series and the second signal time series, converting the first signal time series and the second signal time series into an m-dimensional time vector and an n-dimensional time vector respectively in chronological order, respectively calculating the distances of the time vectors in the m-dimensional time vector and the n-dimensional time vector, setting a distance threshold, counting the number of time vectors whose distances of the time vectors are less than the distance threshold, respectively recording them as the first threshold number and the second threshold number, and then calculating the ratio of the first threshold number to the number of m-dimensional time vectors and the ratio of the second threshold number to the number of n-dimensional time vectors, respectively recording them as the m-dimensional time vector ratio and the n-dimensional time vector ratio; The m+1-dimensional time vector ratio and the n+1-dimensional time vector ratio are calculated in sequence to obtain the sample entropy of the IMF component signal at different stages and the sample entropy of the residual decomposition vector. The calculation formula is as follows: , ; in, Represents the sample entropy of the IMF component signal at different stages, represents the sample entropy of the residual decomposition vector, represents the ratio of the m+1-dimensional time vector, represents the m-dimensional time vector ratio, represents the ratio of n+1-dimensional time vector, Represents the ratio of n-dimensional time vector; S13, using the sample entropy of the IMF component signals at different stages and the sample entropy of the residual decomposition vector to divide the initial automobile air-conditioning compressor vibration signal set to obtain a high-frequency vibration signal component, a vibration signal noise component and a low-frequency vibration signal component, and denoising the vibration signal noise component to eliminate the vibration signal noise to obtain a processed automobile air-conditioning compressor vibration signal set.

[0008] The invention introduces auxiliary noise to perform signal decomposition processing, so as to fully decompose the vibration signal, and then uses the sample entropy method on the decomposition processing result to calculate the sample entropy, identify the vibration signal noise and eliminate it. Compared with the direct noise elimination method, the calculation time is reduced by screening the noise, which greatly improves the denoising effect.

[0009] Preferably, S2 comprises the following steps: S21, performing variational modal decomposition on the processed automobile air-conditioning compressor vibration signal set, and establishing a variational modal decomposition function, as follows: ; in, represents the variational mode decomposition function of the cth mode, represents the instantaneous amplitude, represents the instantaneous phase, c represents the decomposition scale; An initial variational mode decomposition model is established according to the variational mode decomposition function, and the initial variational mode decomposition model is updated using a penalty factor and a Lagrangian function to obtain a variational mode decomposition model; S22, taking the variational mode decomposition model as the fitness function, taking the minimum value of the variational mode decomposition model as the optimal fitness function value, introducing chaos mapping and dimensional learning to improve the crayfish optimization algorithm, obtaining an improved crayfish optimization algorithm, and using the improved crayfish optimization algorithm to optimize the decomposition scale and penalty factor in the variational mode decomposition process, obtaining the global optimal solution of the algorithm, the specific steps are as follows: S221. Set the variational mode decomposition model, i.e., the search space. There is a crayfish population in the search space. The dimension of the crayfish population is j. The individual positions of crayfish in the crayfish population represent the decomposition scale and the penalty factor in the variational mode decomposition model. In the initialization stage of the crayfish population, the crayfish population is initialized using chaotic mapping to obtain the initial positions of the crayfish individuals in the crayfish population. In the summer vacation stage, the current number of iterations is set to d, the maximum number of iterations is D, and the optimal position of the i-th crayfish individual since the dth iteration is , the optimal position of the crayfish population in the dth iteration is , then the location of the summer cave ; Set the position of the i-th crayfish in the d-th iteration as , heat coefficient , Represents a random number between the interval [0, 1], when At the d+1th iteration, the position of the i-th crayfish is The calculation formula is as follows: ; At this time, the current best fitness function value at the best position of the crayfish individual in the d+1th iteration is calculated, and the current best fitness function value corresponds to the d+1th best decomposition scale and the best penalty factor; S222, in the competition stage of the crayfish population, any crayfish individual is selected from the crayfish population and recorded as a competing crayfish individual. The position of the competing crayfish individual is ,when When the i-th crayfish individual updates its position according to the competing crayfish individual, To update, ; After the competition, the crayfish population enters the foraging stage, setting the food position to , food factor is , the food size is , food intake is g; when At this time, the crayfish individuals take turns to eat, Update again, this time When satisfied At this time, the crayfish individuals eat directly, and the position Update again, this time ; Introduce dimensional learning to continue updating the individual positions of crayfish and calculate the current position and The Euclidean distance of the ith crayfish individual position in the dth iteration is taken as the center and the Euclidean distance is taken as the radius to form a neighborhood of the crayfish individual. The positions of the crayfish individuals in the neighborhood of the crayfish individual are randomly updated to obtain a new position. The new position is compared with the position The corresponding fitness function value is retained, the current best fitness function value is retained, and the crayfish population for the next iteration is obtained; the crayfish population for the next iteration repeatedly updates the individual positions of crayfish until the current number of iterations reaches the maximum number of iterations, stops the iteration, obtains the final crayfish population, selects the individual positions of crayfish at the best fitness function, and obtains the global optimal solution of the algorithm.

[0010] This invention uses an improved crayfish optimization algorithm to optimize the decomposition scale and penalty factor in the process of variational mode decomposition, and finds the best fitness function value to solve the optimal solution of the model. The algorithm introduces chaotic mapping, increases the randomness of initialization and the algorithm search capability, and introduces dimensional learning, which overcomes the shortcomings of traditional optimization algorithms such as slow convergence speed and easy falling into local optimal solutions, and has good optimization effect.

[0011] Preferably, S3 comprises the following steps: S31, taking the global optimal solution of the algorithm as the optimized decomposition scale and the optimized penalty factor, using the optimized decomposition scale and the optimized penalty factor to calculate the center frequency and the bandwidth factor, decomposing the final IMF component signal of the processed automobile air-conditioning compressor vibration signal set, setting a decomposition threshold until the difference between adjacent final IMF component signals is less than the decomposition threshold, and obtaining a final IMF component signal set; S32, using a symmetric polar coordinate transformation method, converting the final IMF component signal set into a fault diagnosis image set, the specific steps are as follows: S321, according to the final IMF component signal set, obtain the component signal time series, recorded as ,in represents the kth component signal time series, that is, the component signal time series corresponding to the kth final IMF component signal. Any time point is selected in the component signal time series and recorded as the conversion time point , set the time interval of the component signal time series to , represents the next time point, then the position calculation formula of the conversion time point in polar coordinates is as follows: , , ; in, represents the radius of the conversion time point in polar coordinates, Indicates the counterclockwise angle of the initial line in polar coordinates at the conversion time point, Indicates the clockwise angle of the initial line in polar coordinates at the conversion time point, and Respectively represent the maximum and minimum values ​​of the component signal time series, represents the rotation angle of the symmetry plane, represents the angle magnification factor; S322. Determine the position of the conversion time point on the fault diagnosis image according to the radius of the conversion time point in polar coordinates, the counterclockwise angle of the initial line of the conversion time point in polar coordinates, and the clockwise angle of the initial line of the conversion time point in polar coordinates, calculate the positions of other time points on the fault diagnosis image in turn, form a fault diagnosis image corresponding to the kth final IMF component signal, and convert the final IMF component signal set into a fault diagnosis image set.

[0012] This invention obtains the final IMF component signal set after variational mode decomposition, and maps the final IMF component signal set to polar coordinates through the symmetric polar coordinate transformation method to form a fault diagnosis image. Compared with directly identifying IMF component signals, this invention has obvious advantages. It can intuitively convert the information of IMF component signals into an image, which can improve the accuracy of fault identification and diagnosis and facilitate subsequent processing.

[0013] Preferably, S4 comprises the following steps: S41: labeling the fault diagnosis image set ,in represents the fault type of the lth automobile air-conditioning compressor, and the fault types of the automobile air-conditioning compressor include valve damage, refrigerant leakage, etc.; obtain the vibration signals of the automobile air-conditioning compressor corresponding to different fault types, record them as vibration signal sample sets, convert the vibration signal sample sets into fault diagnosis image sample sets, and label them, use the fault diagnosis image sample sets to train the convolutional neural network, and obtain the convolutional neural network fault diagnosis model, the specific steps are as follows: S411. Set the convolutional neural network to include input layer, convolution layer, pooling layer and output layer, and the initial learning rate is , the maximum number of iterations is p, the fault diagnosis image sample set is randomly divided into a sample training set and a sample test set, the sample training set is input into the convolutional neural network for training, and iterates continuously until the maximum number of iterations is reached or the convolutional neural network converges, to obtain a trained convolutional neural network; S412, inputting the sample test set into the trained convolutional neural network, setting a recognition accuracy threshold, and obtaining a convolutional neural network fault diagnosis model when the output result accuracy is greater than the recognition accuracy threshold, otherwise adjusting the weight until the output result accuracy is greater than the recognition accuracy threshold; S42, inputting the fault diagnosis image set into a convolutional neural network fault diagnosis model, outputting a label value, wherein the label value corresponds to the fault type of the automobile air-conditioning compressor, obtaining a fault diagnosis result of the automobile air-conditioning compressor, and completing the fault diagnosis of the automobile air-conditioning compressor.

[0014] The invention obtains a convolutional neural network fault diagnosis model through training. The neural network is suitable for image recognition and outputs the fault diagnosis results of the automobile air-conditioning compressor. With less time consumption, the classification performance is better than other neural networks, and high-precision fault diagnosis can be better achieved.

[0015] This embodiment also discloses a system for a fault diagnosis method applied to an automobile air-conditioning compressor, which specifically includes: a vibration signal noise removal module, a variational modal decomposition optimization module, a fault diagnosis image generation module and a fault diagnosis module; The vibration signal noise elimination module is used to eliminate the vibration signal noise in the initial automobile air-conditioning compressor vibration signal set; The variational mode decomposition optimization module is used to optimize the parameters in the variational mode decomposition process using an improved crayfish optimization algorithm; The fault diagnosis image generation module is used to convert the final IMF component signal into a fault diagnosis image using a symmetric polar coordinate transformation method; The fault diagnosis module is used to output the fault diagnosis result of the automobile air-conditioning compressor based on the convolutional neural network fault diagnosis model.

[0016] The present invention has the following beneficial effects: 1. The invention introduces auxiliary noise to perform signal decomposition processing, so as to fully decompose the vibration signal, and then uses the sample entropy method on the decomposition processing result to calculate the sample entropy, identify the vibration signal noise and eliminate it. Compared with the direct noise elimination method, the calculation time is reduced by screening the noise, and the denoising effect is greatly improved.

[0017] 2. The invention uses an improved crayfish optimization algorithm to optimize the decomposition scale and penalty factor in the process of variational mode decomposition, and finds the best fitness function value to solve the optimal solution of the model. The algorithm introduces chaotic mapping, increases the randomness of initialization and the algorithm search capability, and introduces dimensional learning, which overcomes the shortcomings of traditional optimization algorithms that have slow convergence speed and are prone to falling into local optimal solutions, and has good optimization effects.

[0018] 3. The present invention obtains the final IMF component signal set after variational mode decomposition, and maps the final IMF component signal set to polar coordinates through the symmetric polar coordinate transformation method to form a fault diagnosis image. Compared with directly identifying the IMF component signal, this invention has obvious advantages. It can intuitively convert the information of the IMF component signal to the image, which can improve the accuracy of fault identification and diagnosis and facilitate subsequent processing.

[0019] 4. The invention obtains a convolutional neural network fault diagnosis model through training. The neural network is suitable for image recognition and outputs the fault diagnosis results of the automobile air-conditioning compressor. Under the premise of less time consumption, the classification performance is better than other neural networks, and high-precision fault diagnosis can be better achieved.

[0020] Of course, any product implementing the present invention does not necessarily need to achieve all of the advantages described above at the same time. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the technical solutions of the embodiments of the invention, the drawings required for describing the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the invention. For ordinary technicians in this field, they can also obtain drawings based on these drawings without paying any creative work.

[0022] Figure 1 The present invention provides a schematic diagram of a fault diagnosis process of a fault diagnosis system applied to an automobile air-conditioning compressor. DETAILED DESCRIPTION

[0023] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0024] In the description of the present invention, it is necessary to understand that the terms "opening", "upper", "lower", "top", "middle", "inside" and the like indicating orientation or positional relationship are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the components or elements referred to must have a specific orientation, be constructed and operate in a specific orientation, and therefore cannot be understood as limiting the invention.

[0025] Example 1 See also Figure 1 This embodiment discloses a fault diagnosis method for an automobile air-conditioning compressor, which specifically includes the following contents: S1. Obtain the vibration signal of the automobile air-conditioning compressor during operation to obtain an initial automobile air-conditioning compressor vibration signal set, introduce auxiliary noise and fuse the sample entropy method, identify and remove the vibration signal noise in the initial automobile air-conditioning compressor vibration signal set, and obtain a processed automobile air-conditioning compressor vibration signal set; The S1 comprises the following steps: S11, placing a signal acquisition device on the automobile air-conditioning compressor to be tested, acquiring a vibration signal of the automobile air-conditioning compressor when it is running, and obtaining an initial automobile air-conditioning compressor vibration signal set, wherein the initial automobile air-conditioning compressor vibration signal in the initial automobile air-conditioning compressor vibration signal set is The initial automobile air-conditioning compressor vibration signal subset is truncated to form, and the corresponding signal time series is set according to the vibration signal acquisition time; the initial automobile air-conditioning compressor vibration signal set is decomposed by introducing auxiliary noise to obtain the vibration signal decomposition vector. The specific steps are as follows: S111, adding auxiliary noise to the initial automobile air-conditioning compressor vibration signal set to obtain a mixed vibration signal set, performing empirical mode decomposition on the auxiliary noise of the mixed vibration signal set to obtain a first IMF component signal set; setting the adding times threshold to , add auxiliary noise to the mixed vibration signal set to obtain the second IMF component signal set, until the number of additions reaches the addition number threshold, obtain several IMF component signal sets, solve the average value of several IMF component signal sets, and record it as the first stage IMF component signal; S112, using the initial automobile air-conditioning compressor vibration signal set to subtract the first-stage IMF component signal to obtain the first-stage vibration signal set, adding auxiliary noise to the first-stage vibration signal set, repeating S111, and obtaining the second-stage IMF component signal; iteratively decomposing the vibration signal set until the vibration signal in the vibration signal set is less than or equal to 2, stopping the iteration, and obtaining the vibration signal decomposition vector, the calculation formula is as follows: ; in, represents the initial automobile air-conditioning compressor vibration signal set, represents the IMF component signal of the i-th stage, represents the residual decomposition vector, b represents the number of iterations of the vibration signal set, , the IMF component signals and residual decomposition vectors at different stages constitute the vibration signal decomposition vector; S12, determining the signal time series corresponding to the IMF component signals and the residual decomposition vectors at different stages, respectively recording them as the first signal time series and the second signal time series, converting the first signal time series and the second signal time series into an m-dimensional time vector and an n-dimensional time vector respectively in chronological order, respectively calculating the distances of the time vectors in the m-dimensional time vector and the n-dimensional time vector, setting a distance threshold, counting the number of time vectors whose distances of the time vectors are less than the distance threshold, respectively recording them as the first threshold number and the second threshold number, and then calculating the ratio of the first threshold number to the number of m-dimensional time vectors and the ratio of the second threshold number to the number of n-dimensional time vectors, respectively recording them as the m-dimensional time vector ratio and the n-dimensional time vector ratio; The m+1-dimensional time vector ratio and the n+1-dimensional time vector ratio are calculated in sequence to obtain the sample entropy of the IMF component signal at different stages and the sample entropy of the residual decomposition vector. The calculation formula is as follows: , ; in, Represents the sample entropy of the IMF component signal at different stages, represents the sample entropy of the residual decomposition vector, represents the ratio of the m+1-dimensional time vector, represents the m-dimensional time vector ratio, represents the ratio of n+1-dimensional time vector, Represents the ratio of n-dimensional time vector; S13, using the sample entropy of the IMF component signals at different stages and the sample entropy of the residual decomposition vector to divide the initial automobile air-conditioning compressor vibration signal set to obtain a high-frequency vibration signal component, a vibration signal noise component and a low-frequency vibration signal component, and performing denoising on the vibration signal noise component to remove the vibration signal noise, thereby obtaining a processed automobile air-conditioning compressor vibration signal set; S2. performing variational modal decomposition on the processed automobile air-conditioning compressor vibration signal set, combining chaos mapping and dimensional learning, and using an improved crayfish optimization algorithm to optimize the decomposition scale and penalty factor in the variational modal decomposition process to obtain the global optimal solution of the algorithm; The S2 comprises the following steps: S21, performing variational modal decomposition on the processed automobile air-conditioning compressor vibration signal set, and establishing a variational modal decomposition function, as follows: ; in, represents the variational mode decomposition function of the cth mode, represents the instantaneous amplitude, represents the instantaneous phase, c represents the decomposition scale; An initial variational mode decomposition model is established according to the variational mode decomposition function, and the initial variational mode decomposition model is updated using a penalty factor and a Lagrangian function to obtain a variational mode decomposition model; S22, taking the variational mode decomposition model as the fitness function, taking the minimum value of the variational mode decomposition model as the optimal fitness function value, introducing chaos mapping and dimensional learning to improve the crayfish optimization algorithm, obtaining an improved crayfish optimization algorithm, and using the improved crayfish optimization algorithm to optimize the decomposition scale and penalty factor in the variational mode decomposition process, obtaining the global optimal solution of the algorithm, the specific steps are as follows: S221. Set the variational mode decomposition model, i.e., the search space. There is a crayfish population in the search space. The dimension of the crayfish population is j. The individual positions of crayfish in the crayfish population represent the decomposition scale and the penalty factor in the variational mode decomposition model. In the initialization stage of the crayfish population, the crayfish population is initialized using chaotic mapping to obtain the initial positions of the crayfish individuals in the crayfish population. In the summer vacation stage, the current number of iterations is set to d, the maximum number of iterations is D, and the optimal position of the i-th crayfish individual since the dth iteration is , the optimal position of the crayfish population in the dth iteration is , then the location of the summer cave ; Set the position of the i-th crayfish in the d-th iteration as , heat coefficient , Represents a random number between the interval [0, 1], when At the d+1th iteration, the position of the i-th crayfish is The calculation formula is as follows: ; At this time, the current best fitness function value at the best position of the crayfish individual in the d+1th iteration is calculated, and the current best fitness function value corresponds to the d+1th best decomposition scale and the best penalty factor; S222, in the competition stage of the crayfish population, any crayfish individual is selected from the crayfish population and recorded as a competing crayfish individual. The position of the competing crayfish individual is ,when When the i-th crayfish individual updates its position according to the competing crayfish individual, To update, ; After the competition, the crayfish population enters the foraging stage, setting the food position to , food factor is , the food size is , food intake is g; when At this time, the crayfish individuals take turns to eat, Update again, this time When satisfied At this time, the crayfish individuals eat directly, and the position Update again, this time ; Introduce dimensional learning to continue updating the individual positions of crayfish and calculate the current position and The Euclidean distance of the ith crayfish individual position in the dth iteration is taken as the center and the Euclidean distance is taken as the radius to form a neighborhood of the crayfish individual. The positions of the crayfish individuals in the neighborhood of the crayfish individual are randomly updated to obtain a new position. The new position is compared with the position The corresponding fitness function value is retained, and the current best fitness function value is obtained, and the crayfish population of the next iteration is obtained; the crayfish population of the next iteration repeatedly updates the individual positions of crayfish until the current number of iterations reaches the maximum number of iterations, stops iteration, obtains the final crayfish population, selects the individual positions of crayfish at the best fitness function, and obtains the global optimal solution of the algorithm; S3, using the global optimal solution of the algorithm as the optimized decomposition scale and the optimized penalty factor, and then converting the processed automobile air-conditioning compressor vibration signal set into a final IMF component signal set, and using the symmetric polar coordinate transformation method to convert the final IMF component signal set into a fault diagnosis image set; The S3 comprises the following steps: S31, taking the global optimal solution of the algorithm as the optimized decomposition scale and the optimized penalty factor, using the optimized decomposition scale and the optimized penalty factor to calculate the center frequency and the bandwidth factor, decomposing the final IMF component signal of the processed automobile air-conditioning compressor vibration signal set, setting a decomposition threshold until the difference between adjacent final IMF component signals is less than the decomposition threshold, and obtaining a final IMF component signal set; S32, using a symmetric polar coordinate transformation method, converting the final IMF component signal set into a fault diagnosis image set, the specific steps are as follows: S321, according to the final IMF component signal set, obtain the component signal time series, recorded as ,in represents the kth component signal time series, that is, the component signal time series corresponding to the kth final IMF component signal. Any time point is selected in the component signal time series and recorded as the conversion time point , set the time interval of the component signal time series to , represents the next time point, then the position calculation formula of the conversion time point in polar coordinates is as follows: , , ; in, represents the radius of the conversion time point in polar coordinates, Indicates the counterclockwise angle of the initial line in polar coordinates at the conversion time point, Indicates the clockwise angle of the initial line in polar coordinates at the conversion time point, and Respectively represent the maximum and minimum values ​​of the component signal time series, represents the rotation angle of the symmetry plane, represents the angle magnification factor; S322, determining the position of the conversion time point on the fault diagnosis image according to the radius of the conversion time point in polar coordinates, the counterclockwise angle of the initial line of the conversion time point in polar coordinates, and the clockwise angle of the initial line of the conversion time point in polar coordinates, and sequentially calculating the positions of other time points on the fault diagnosis image to form a fault diagnosis image corresponding to the kth final IMF component signal, and converting the final IMF component signal set into a fault diagnosis image set; S4, training a convolutional neural network fault diagnosis model, combining the fault diagnosis image set and the convolutional neural network fault diagnosis model, and outputting a fault diagnosis result of the automobile air-conditioning compressor; The S4 comprises the following steps: S41: labeling the fault diagnosis image set ,in represents the fault type of the lth automobile air-conditioning compressor, and the fault types of the automobile air-conditioning compressor include valve damage, refrigerant leakage, etc.; obtain the vibration signals of the automobile air-conditioning compressor corresponding to different fault types, record them as vibration signal sample sets, convert the vibration signal sample sets into fault diagnosis image sample sets, and label them, use the fault diagnosis image sample sets to train the convolutional neural network, and obtain the convolutional neural network fault diagnosis model, the specific steps are as follows: S411. Set the convolutional neural network to include input layer, convolution layer, pooling layer and output layer, and the initial learning rate is , the maximum number of iterations is p, the fault diagnosis image sample set is randomly divided into a sample training set and a sample test set, the sample training set is input into the convolutional neural network for training, and iterates continuously until the maximum number of iterations is reached or the convolutional neural network converges, to obtain a trained convolutional neural network; S412, inputting the sample test set into the trained convolutional neural network, setting a recognition accuracy threshold, and obtaining a convolutional neural network fault diagnosis model when the output result accuracy is greater than the recognition accuracy threshold, otherwise adjusting the weight until the output result accuracy is greater than the recognition accuracy threshold; S42, inputting the fault diagnosis image set into a convolutional neural network fault diagnosis model, outputting a label value, wherein the label value corresponds to the fault type of the automobile air-conditioning compressor, obtaining a fault diagnosis result of the automobile air-conditioning compressor, and completing the fault diagnosis of the automobile air-conditioning compressor.

[0026] Example 2 This embodiment also discloses a system for a fault diagnosis method applied to an automobile air-conditioning compressor, which specifically includes: a vibration signal noise removal module, a variational modal decomposition optimization module, a fault diagnosis image generation module and a fault diagnosis module; The vibration signal noise elimination module is used to eliminate the vibration signal noise in the initial automobile air-conditioning compressor vibration signal set; The variational mode decomposition optimization module is used to optimize the parameters in the variational mode decomposition process using an improved crayfish optimization algorithm; The fault diagnosis image generation module is used to convert the final IMF component signal into a fault diagnosis image using a symmetric polar coordinate transformation method; The fault diagnosis module is used to output the fault diagnosis result of the automobile air-conditioning compressor based on the convolutional neural network fault diagnosis model.

[0027] In the description of this specification, the description with reference to the terms "one embodiment", "example", "specific example", etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0028] The preferred embodiments of the invention disclosed above are only used to help explain the invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to the specific implementation methods described. Obviously, many modifications and changes can be made according to the content of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the invention, so that those skilled in the art can understand and use the invention well.

Claims

1. A fault diagnosis method for an automobile air-conditioning compressor, characterized in that: The steps include: S1. Obtain the vibration signal of the automobile air-conditioning compressor during operation to obtain an initial automobile air-conditioning compressor vibration signal set, introduce auxiliary noise and fuse the sample entropy method, identify and remove the vibration signal noise in the initial automobile air-conditioning compressor vibration signal set, and obtain a processed automobile air-conditioning compressor vibration signal set; S2, performing variational modal decomposition on the processed automobile air-conditioning compressor vibration signal set, combining chaos mapping and dimensional learning, and using an optimization algorithm to optimize the decomposition scale and penalty factor in the variational modal decomposition process to obtain a global optimal solution of the algorithm; S3, using the global optimal solution of the algorithm as the optimized decomposition scale and the optimized penalty factor, and then converting the processed automobile air-conditioning compressor vibration signal set into a final IMF component signal set, and using the symmetric polar coordinate transformation method to convert the final IMF component signal set into a fault diagnosis image set; S4. Train a convolutional neural network fault diagnosis model, combine the fault diagnosis image set and the convolutional neural network fault diagnosis model, and output a fault diagnosis result of the automobile air-conditioning compressor.

2. A fault diagnosis method for automobile air-conditioning compressor according to claim 1, characterized in that: The S1 comprises the following steps: S11, obtaining a vibration signal of the automobile air-conditioning compressor when it is running, and obtaining an initial automobile air-conditioning compressor vibration signal set; introducing auxiliary noise to decompose the initial automobile air-conditioning compressor vibration signal set, and obtaining a vibration signal decomposition vector; S12, the vibration signal decomposition vector includes the signal time series corresponding to the IMF component signals and the residual decomposition vector at different stages, and the sample entropy of the IMF component signals and the sample entropy of the residual decomposition vector at different stages are calculated; S13, after dividing the initial automobile air-conditioning compressor vibration signal set using the sample entropies of the IMF component signals at different stages and the sample entropies of the residual decomposition vector, the vibration signal noise is eliminated to obtain a processed automobile air-conditioning compressor vibration signal set.

3. A fault diagnosis method for automobile air-conditioning compressor according to claim 2, characterized in that: The S11 comprises the following steps: S111, adding auxiliary noise to the initial automobile air-conditioning compressor vibration signal set to obtain a mixed vibration signal set, performing empirical mode decomposition on the auxiliary noise of the mixed vibration signal set to obtain a first-stage IMF component signal; S112. Subtract the first-stage IMF component signal from the initial automobile air-conditioning compressor vibration signal set to obtain a first-stage vibration signal set, add auxiliary noise to the first-stage vibration signal set, iteratively decompose the vibration signal set, and obtain a vibration signal decomposition vector.

4. A fault diagnosis method for automobile air-conditioning compressor according to claim 3, characterized in that: The S2 comprises the following steps: S21, performing variational modal decomposition on the processed automobile air-conditioning compressor vibration signal set, establishing an initial variational modal decomposition model, and updating the initial variational modal decomposition model using a penalty factor and a Lagrangian function to obtain a variational modal decomposition model; S22. Taking the variational mode decomposition model as the fitness function, taking the minimum value of the variational mode decomposition model as the optimal fitness function value, introducing chaotic mapping and dimensional learning, and using the improved crayfish optimization algorithm to optimize the decomposition scale and penalty factor in the variational mode decomposition process, to obtain the global optimal solution of the algorithm.

5. A fault diagnosis method for automobile air-conditioning compressor according to claim 4, characterized in that: The S22 comprises the following steps: S221. Set the variational mode decomposition model, i.e., the search space. There is a crayfish population in the search space. The dimension of the crayfish population is j. The individual positions of crayfish in the crayfish population represent the decomposition scale and the penalty factor in the variational mode decomposition model. In the initialization stage of the crayfish population, the crayfish population is initialized using chaotic mapping to obtain the initial positions of the crayfish individuals in the crayfish population. In the summer vacation stage, the current number of iterations is set to d, the maximum number of iterations is D, and the optimal position of the i-th crayfish individual since the dth iteration is , the optimal position of the crayfish population in the dth iteration is , then the location of the summer cave ; Set the position of the i-th crayfish in the d-th iteration as , heat coefficient , Represents a random number between the interval [0, 1], when At the d+1th iteration, the position of the i-th crayfish is The calculation formula is as follows: ； At this time, the current best fitness function value at the best position of the crayfish individual in the d+1th iteration is calculated, and the current best fitness function value corresponds to the d+1th best decomposition scale and the best penalty factor; S222, in the competition stage of the crayfish population, any crayfish individual is selected from the crayfish population and recorded as a competing crayfish individual. The position of the competing crayfish individual is ,when When the i-th crayfish individual updates its position according to the competing crayfish individual, After the competition, the crayfish population enters the foraging stage, sets the food position, food factor and food size, and continues to adjust the position. Make updates; Introduce dimensional learning to continue updating the individual positions of crayfish and calculate the positions and The Euclidean distance of the ith crayfish individual position in the dth iteration is taken as the center and the Euclidean distance is taken as the radius to form a neighborhood of the crayfish individual. The positions of the crayfish individuals in the neighborhood of the crayfish individual are randomly updated to obtain a new position. The new position is compared with the position The corresponding fitness function value is retained, the current best fitness function value is retained, and the crayfish population for the next iteration is obtained; the crayfish population for the next iteration repeatedly updates the individual positions of crayfish until the current number of iterations reaches the maximum number of iterations, stops the iteration, obtains the final crayfish population, selects the individual positions of crayfish at the best fitness function, and obtains the global optimal solution of the algorithm.

6. A fault diagnosis method for automobile air-conditioning compressor according to claim 5, characterized in that: The S3 comprises the following steps: S31, taking the global optimal solution of the algorithm as the optimized decomposition scale and the optimized penalty factor, using the optimized decomposition scale and the optimized penalty factor to calculate the center frequency and the bandwidth factor, decomposing the processed automobile air-conditioning compressor vibration signal set, and obtaining the final IMF component signal set; S32, using a symmetric polar coordinate transformation method, converting the final IMF component signal set into a fault diagnosis image set.

7. A fault diagnosis method for automobile air-conditioning compressor according to claim 6, characterized in that: The S32 comprises the following steps: S321, obtaining a component signal time series according to the final IMF component signal set, selecting a conversion time point in the component signal time series, and calculating the position of the conversion time point in polar coordinates; S322, determining the position of the conversion time point on the fault diagnosis image, calculating the positions of other time points on the fault diagnosis image in sequence, and converting the final IMF component signal set into a fault diagnosis image set.

8. A fault diagnosis method for automobile air-conditioning compressor according to claim 7, characterized in that: The S4 comprises the following steps: S41, labeling the fault diagnosis image set, obtaining automobile air-conditioning compressor vibration signals corresponding to different fault types, recording them as vibration signal sample sets, converting the vibration signal sample sets into fault diagnosis image sample sets, labeling them, and using the fault diagnosis image sample sets to train a convolutional neural network to obtain a convolutional neural network fault diagnosis model; S42, inputting the fault diagnosis image set into a convolutional neural network fault diagnosis model, outputting a label value, wherein the label value corresponds to the fault type of the automobile air-conditioning compressor, obtaining a fault diagnosis result of the automobile air-conditioning compressor, and completing the fault diagnosis of the automobile air-conditioning compressor.

9. A fault diagnosis method for automobile air-conditioning compressor according to claim 8, characterized in that: The S41 comprises the following steps: S411. Set the convolutional neural network to include input layer, convolution layer, pooling layer and output layer, and the initial learning rate is , the maximum number of iterations is p, the fault diagnosis image sample set is randomly divided into a sample training set and a sample test set, the sample training set is input into the convolutional neural network for training, and iterates continuously until the maximum number of iterations is reached or the convolutional neural network converges, to obtain a trained convolutional neural network; S412. Input the sample test set into the trained convolutional neural network, set the recognition accuracy threshold, and when the output result accuracy is greater than the recognition accuracy threshold, obtain the convolutional neural network fault diagnosis model; otherwise, adjust the weight until the output result accuracy is greater than the recognition accuracy threshold.

10. A system for implementing the fault diagnosis method for automobile air-conditioning compressor according to any one of claims 1 to 9, characterized in that: Specifically include: Vibration signal noise removal module, variational mode decomposition optimization module, fault diagnosis image generation module and fault diagnosis module; The vibration signal noise elimination module is used to eliminate the vibration signal noise in the initial automobile air-conditioning compressor vibration signal set; The variational mode decomposition optimization module is used to optimize the parameters in the variational mode decomposition process using an improved crayfish optimization algorithm; The fault diagnosis image generation module is used to convert the final IMF component signal into a fault diagnosis image using a symmetric polar coordinate transformation method; The fault diagnosis module is used to output the fault diagnosis result of the automobile air-conditioning compressor based on the convolutional neural network fault diagnosis model.

Images