Synchronizer gear ring detection device and method

Abstract

The invention discloses a synchronizer gear ring detection device and method, belonging to the technical field of synchronizer gear ring fault detection. The invention adopts a three-axis accelerometer to collect vibration signals of a transmission in three directions, superimposes the three-axis vibration signals in the frequency domain, fully retains the signal components in each direction, obtains an enhanced vibration signal, adopts morphological operation to extract local maximum values ​​and local minimum values ​​of the enhanced vibration signal respectively, reflects the extreme value characteristics of the vibration signal, then constructs a vibration trajectory matrix for the enhanced vibration signal, extracts features, constitutes an approximate feature sequence and a detail feature sequence, reflects the component features of the enhanced vibration signal at different scales, and then adopts a synchronizer gear ring detection system to process the peak sequence, the valley sequence, the approximate feature sequence and the detail feature sequence, comprehensively evaluates the state of the synchronizer gear ring through multi-feature fusion analysis, and improves the accuracy of synchronizer gear ring fault recognition.

Description

Technical Field

[0001] The present invention relates to the technical field of synchronizer gear ring fault detection, and in particular to a synchronizer gear ring detection device and method. Background Art

[0002] With the rapid development of the automobile industry, the performance and reliability of the gearbox, as a key component of the automobile transmission system, directly affect the power, economy and comfort of the vehicle. Among the many parts of the gearbox, the synchronizer ring is an important component to achieve smooth gear shifting. However, due to long-term use and the influence of various factors, the synchronizer ring may be worn, deformed or damaged, resulting in problems such as difficult gear shifting, abnormal noise or gear skipping.

[0003] Traditional synchronizer ring fault detection methods mainly rely on manual disassembly inspection or simple sound judgment. These methods are not only time-consuming and labor-intensive, but also difficult to detect potential fault hazards in a timely manner. With the advancement of science and technology, some fault diagnosis methods based on vibration signal analysis have gradually been applied to the field of gearbox detection. However, due to the complex working environment of the gearbox, the vibration signal often contains a lot of noise and interference, making it difficult to accurately identify the synchronizer ring fault, and there is a problem of low accuracy in synchronizer ring fault identification. Summary of the invention

[0004] In view of the above-mentioned deficiencies in the prior art, a synchronizer gear ring detection device and method provided by the present invention solves the problem of low synchronizer gear ring fault identification accuracy in the prior art.

[0005] In order to achieve the above-mentioned invention object, the technical solution adopted by the present invention is: a synchronizer gear ring detection device, comprising: a three-axis accelerometer, a signal enhancement system, a first sequence construction system, a second sequence construction system and a synchronizer gear ring detection system;

[0006] The triaxial accelerometer is installed on the gearbox to continuously monitor the gearbox vibration during gear shifting and obtain the gearbox triaxial vibration signal;

[0007] The signal enhancement system is used to superimpose the triaxial vibration signals in the frequency domain to obtain an enhanced vibration signal;

[0008] The first sequence construction system is used to extract the local maximum and local minimum of the enhanced vibration signal by using morphological operations, and to construct a peak sequence and a valley sequence;

[0009] The second sequence construction system is used to construct a vibration trajectory matrix for the enhanced vibration signal, extract features from the vibration trajectory matrix, and construct an approximate feature sequence and a detail feature sequence;

[0010] The synchronizer gear ring detection system is used to process the peak sequence, valley sequence, approximate feature sequence and detail feature sequence to obtain the synchronizer gear ring fault value.

[0011] The beneficial effects of the present invention are as follows: the present invention adopts a three-axis accelerometer to collect vibration signals of the transmission in three directions, superimposes the three-axis vibration signals in the frequency domain, fully retains the signal components in all directions, obtains an enhanced vibration signal, adopts morphological operations to extract local maximum values ​​and local minimum values ​​of the enhanced vibration signal respectively, reflects the extreme value characteristics of the vibration signal, then constructs a vibration trajectory matrix for the enhanced vibration signal, and extracts features to form an approximate feature sequence and a detail feature sequence, reflecting the component characteristics of the enhanced vibration signal at different scales, and then adopts a synchronizer gear ring detection system to process the peak sequence, valley sequence, approximate feature sequence and detail feature sequence, and comprehensively evaluates the state of the synchronizer gear ring through multi-feature fusion analysis, thereby improving the accuracy of synchronizer gear ring fault identification.

[0012] Further, the signal enhancement system includes: a Fourier transform unit, a superposition unit and an inverse Fourier transform unit;

[0013] The Fourier transform unit is used to perform Fourier transform on the three-axis vibration signals respectively to obtain an X-axis vibration frequency domain signal, a Y-axis vibration frequency domain signal and a Z-axis vibration frequency domain signal;

[0014] The superposition unit is used to superimpose the X-axis vibration frequency domain signal, the Y-axis vibration frequency domain signal and the Z-axis vibration frequency domain signal to obtain a frequency domain superposition signal;

[0015] The inverse Fourier transform unit is used to perform inverse Fourier transform on the frequency domain superposition signal to obtain an enhanced vibration signal.

[0016] The beneficial effect of the above further scheme is that the present invention can strengthen weak fault signals that may not be obvious in a single axis through frequency domain superposition. Through the synergistic enhancement of signals in three directions, early or minor synchronizer ring faults are easier to detect, and the characteristic frequency components related to synchronizer ring faults are effectively enhanced. By using the frequency domain superposition of three-axis vibration signals, the vibration information in different directions is integrated to ensure that no vibration characteristics that may be related to the fault are missed.

[0017] Further, the first sequence construction system includes: an opening operation unit, a first subtraction unit, a peak sequence construction unit, a closing operation unit, a second subtraction unit and a valley sequence construction unit;

[0018] The opening operation unit is used to perform an opening operation on the enhanced vibration signal;

[0019] The first subtraction unit is used to subtract the opening operation result from the enhanced vibration signal to obtain a peak significant signal;

[0020] The peak sequence construction unit is used to detect the local maximum of the peak significant signal by using the dilation operation, and construct the peak sequence with all the local maximum values ​​as elements;

[0021] The closing operation unit is used to perform a closing operation on the enhanced vibration signal;

[0022] The second subtraction unit is used to subtract the enhanced vibration signal from the closed operation result to obtain a valley significant signal;

[0023] The valley sequence construction unit is used to detect local minima of valley significant signals by using corrosion operations, and construct a valley sequence using all local minima as elements.

[0024] The beneficial effect of the above further scheme is: the present invention improves the accuracy of detecting local maximum and local minimum values ​​by subtracting the open operation result from the enhanced vibration signal and subtracting the closed operation result from the enhanced vibration signal, thereby extracting significant features: local maximum and local minimum values.

[0025] Further, the second sequence construction system includes: a matrix construction unit, a wavelet transform unit, a first calculation unit, a second calculation unit, an approximate feature sequence construction unit and a detail feature sequence construction unit;

[0026] The matrix construction unit is used to construct a vibration trajectory matrix for the enhanced vibration signal;

[0027] The wavelet transform unit is used to perform multi-layer wavelet transform on each row of the vibration trajectory matrix to obtain an approximate coefficient set and a detail coefficient set of each row;

[0028] The first calculation unit is used to calculate the first eigenvalue for each approximation coefficient set;

[0029] The second calculation unit is used to calculate the second eigenvalue for each detail coefficient set;

[0030] The approximate feature sequence construction unit is used to construct an approximate feature sequence by taking each first feature value as an element;

[0031] The detail feature sequence constructing unit is used to construct a detail feature sequence using each second feature value as an element.

[0032] The beneficial effects of the above further scheme are as follows: each row of the vibration trajectory matrix represents the evolution process of the signal at different time scales. Performing wavelet transform on each row can better capture the local characteristics of the signal at different time and frequency scales, which is conducive to discovering the small vibration changes caused by synchronizer gear ring faults. The vibration trajectory matrix retains the time structure information of the original signal. Performing wavelet transform on each row can analyze the changing characteristics of the signal components when the signal changes with time at different scales, which is helpful to identify the dynamic characteristics of synchronizer gear ring faults.

[0033] Furthermore, the calculation formulas for the first eigenvalue and the second eigenvalue are both: , where X is the first eigenvalue or the second eigenvalue, x i is the i-th element in the approximation coefficient set or the detail coefficient set, N is the number of elements in the set, and i is a positive integer.

[0034] The beneficial effect of the above further scheme is: after performing multi-layer wavelet transform on each row of the vibration trajectory matrix, there are multiple approximate coefficients and detail coefficients for each row. The present invention weights the approximate coefficients and detail coefficients of each row respectively, and characterizes the distribution of the approximate coefficients and detail coefficients by the weighted values.

[0035] Further, the synchronizer gear ring detection system includes: a first feature fusion unit, a second feature fusion unit, a first tanh layer, a second tanh layer, a third tanh layer, a Concat layer, an output enhancement layer and a fully connected layer;

[0036] The first input end of the first feature fusion unit is used to input a peak value sequence, and the second input end thereof is used to input a valley value sequence;

[0037] The first input end of the second feature fusion unit is used to input the approximate feature sequence, and the second input end thereof is used to input the detail feature sequence;

[0038] The input end of the first tanh layer is connected to the output end of the first feature fusion unit;

[0039] The input end of the second tanh layer is connected to the output end of the second feature fusion unit;

[0040] The input end of the Concat layer is connected to the output end of the first tanh layer and the output end of the second tanh layer respectively, and its output end is connected to the input end of the third tanh layer;

[0041] The input end of the output enhancement layer is connected to the output end of the third tanh layer, and its output end is connected to the input end of the fully connected layer;

[0042] The output of the fully connected layer is used as the output of the synchronizer ring detection system.

[0043] The beneficial effects of the above further scheme are as follows: the present invention uses convolution layers to process peak sequences, valley sequences, approximate feature sequences and detail feature sequences respectively to obtain four different types of features, uses multiplier M1 to fuse the features corresponding to the peak sequence with the features corresponding to the valley sequence to obtain a first fused feature, uses multiplier M2 to fuse the features corresponding to the approximate feature sequence with the features corresponding to the detail feature sequence to obtain a second fused feature, so that the first fused feature and the second fused feature are processed by the tanh layer respectively, so as to realize separate processing of the first fused feature and the second fused feature, respectively capture nonlinear relationships, and then use the Concat layer for feature splicing, use the third tanh layer to consider each output feature of the Concat layer, and use the output enhancement layer to enhance the output of the third tanh layer to make the significant features more significant.

[0044] Furthermore, the expression of the first feature fusion unit is: , where Y1 is the output of the first feature fusion unit, X p is the peak sequence, X v is the valley value sequence, Conv is the convolution operation, for ;

[0045] The expression of the second feature fusion unit is: , where Y2 is the output of the second feature fusion unit, A is the approximate feature sequence, and D is the detail feature sequence.

[0046] Furthermore, the expressions of the first tanh layer and the second tanh layer are: , where h n is the nth eigenvalue output by the first tanh layer or the second tanh layer, tanh is the hyperbolic tangent activation function, x n is the nth eigenvalue of the first tanh layer or the second tanh layer input, n is a positive integer, ω x,n For x n The weight of b x,n For x n Bias of

[0047] The expression of the third tanh layer is: , where r n is the nth eigenvalue output by the third tanh layer, e n is the nth eigenvalue of the third tanh layer input, ω e,n for e n The weight of b e,n for e n The bias.

[0048] Furthermore, the expression of the output enhancement layer is: , where y n is the nth eigenvalue of the output enhancement layer, r n is the nth eigenvalue output by the third tanh layer, σ is the Sigmoid function, ω r,n For r n The weight of b r,n For r n The bias of , n is a positive integer.

[0049] The beneficial effect of the above further solution is: the present invention assigns weights and biases to each eigenvalue output by the third tanh layer, and uses the Sigmoid function to calculate the enhancement ratio coefficient of the output enhancement layer. , adaptively adjust the eigenvalue size of the effective features.

[0050] A synchronizer gear ring detection method comprises the following steps:

[0051] S1. Install a triaxial accelerometer on the gearbox, continuously monitor the gearbox vibration during gear shifting, and obtain a triaxial vibration signal of the gearbox;

[0052] S2. Superimposing the three-axis vibration signals in the frequency domain to obtain an enhanced vibration signal;

[0053] S3, using morphological operations to extract local maximum and local minimum values ​​of the enhanced vibration signal, respectively, and constructing peak sequence and valley sequence;

[0054] S4, constructing a vibration trajectory matrix for the enhanced vibration signal, extracting features from the vibration trajectory matrix, and constructing an approximate feature sequence and a detail feature sequence;

[0055] S5. According to the synchronizer gear ring detection system, the peak sequence, the valley sequence, the approximate feature sequence and the detail feature sequence are processed to obtain the synchronizer gear ring fault condition. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 The present invention is a system block diagram of a synchronizer gear ring detection device;

[0057] Figure 2 It is a structural schematic diagram of the synchronizer ring detection system. DETAILED DESCRIPTION

[0058] The specific implementation modes of the present invention are described below so that those skilled in the art can understand the present invention. However, it should be clear that the present invention is not limited to the scope of the specific implementation modes. For those of ordinary skill in the art, as long as various changes are within the spirit and scope of the present invention as defined and determined by the attached claims, these changes are obvious, and all inventions and creations utilizing the concept of the present invention are protected.

[0059] Example 1: Figure 1 As shown, a synchronizer gear ring detection device includes: a three-axis accelerometer, a signal enhancement system, a first sequence construction system, a second sequence construction system and a synchronizer gear ring detection system;

[0060] The triaxial accelerometer is installed on the gearbox to continuously monitor the gearbox vibration during gear shifting and obtain the gearbox triaxial vibration signal;

[0061] The signal enhancement system is used to superimpose the triaxial vibration signals in the frequency domain to obtain an enhanced vibration signal;

[0062] The first sequence construction system is used to extract the local maximum and local minimum of the enhanced vibration signal by using morphological operations, and to construct a peak sequence and a valley sequence;

[0063] The second sequence construction system is used to construct a vibration trajectory matrix for the enhanced vibration signal, extract features from the vibration trajectory matrix, and construct an approximate feature sequence and a detail feature sequence;

[0064] The synchronizer gear ring detection system is used to process the peak sequence, valley sequence, approximate feature sequence and detail feature sequence to obtain the synchronizer gear ring fault value.

[0065] The signal enhancement system includes: a Fourier transform unit, a superposition unit and an inverse Fourier transform unit;

[0066] The Fourier transform unit is used to perform Fourier transform on the three-axis vibration signals respectively to obtain an X-axis vibration frequency domain signal, a Y-axis vibration frequency domain signal and a Z-axis vibration frequency domain signal;

[0067] The superposition unit is used to superimpose the X-axis vibration frequency domain signal, the Y-axis vibration frequency domain signal and the Z-axis vibration frequency domain signal to obtain a frequency domain superposition signal;

[0068] The inverse Fourier transform unit is used to perform inverse Fourier transform on the frequency domain superposition signal to obtain an enhanced vibration signal.

[0069] In this embodiment, when the X-axis vibration frequency domain signal, the Y-axis vibration frequency domain signal and the Z-axis vibration frequency domain signal are superimposed, the specific process is: when there are the same frequency components in three directions or two directions of the X-axis, Y-axis and Z-axis, the amplitudes in three directions or two directions are added on the same frequency component, and the other different frequency components are taken as a union. Among the frequency components taken as the union, the corresponding amplitude of each frequency component is kept unchanged, that is, the amplitudes on the same frequency components are added, and the different frequency components are directly combined to obtain a new frequency domain superposition signal, thereby retaining the signal components in each direction.

[0070] The present invention can strengthen weak fault signals that may not be obvious in a single axis through frequency domain superposition. Through the synergistic enhancement of signals in three directions, early or minor synchronizer ring faults are easier to detect, and the characteristic frequency components related to synchronizer ring faults are effectively enhanced. By using the frequency domain superposition of three-axis vibration signals, the vibration information in different directions is integrated to ensure that no vibration characteristics that may be related to the fault are missed.

[0071] The first sequence construction system includes: an opening operation unit, a first subtraction unit, a peak sequence construction unit, a closing operation unit, a second subtraction unit and a valley sequence construction unit;

[0072] The opening operation unit is used to perform an opening operation on the enhanced vibration signal;

[0073] The first subtraction unit is used to subtract the opening operation result from the enhanced vibration signal to obtain a peak significant signal;

[0074] The peak sequence construction unit is used to detect the local maximum of the peak significant signal by using the dilation operation, and construct the peak sequence with all the local maximum values ​​as elements;

[0075] The closing operation unit is used to perform a closing operation on the enhanced vibration signal;

[0076] The second subtraction unit is used to subtract the enhanced vibration signal from the closed operation result to obtain a valley significant signal;

[0077] The valley sequence construction unit is used to detect local minima of valley significant signals by using corrosion operations, and construct a valley sequence using all local minima as elements.

[0078] The present invention improves the accuracy of detecting local maximum and local minimum values ​​by subtracting the opening operation result from the enhanced vibration signal and subtracting the closed operation result from the enhanced vibration signal, thereby extracting the significant features: local maximum and local minimum values.

[0079] The second sequence construction system includes: a matrix construction unit, a wavelet transform unit, a first calculation unit, a second calculation unit, an approximate feature sequence construction unit and a detail feature sequence construction unit;

[0080] The matrix construction unit is used to construct a vibration trajectory matrix for the enhanced vibration signal;

[0081] The wavelet transform unit is used to perform multi-layer wavelet transform on each row of the vibration trajectory matrix to obtain an approximate coefficient set and a detail coefficient set of each row;

[0082] The first calculation unit is used to calculate the first eigenvalue for each approximation coefficient set;

[0083] The second calculation unit is used to calculate the second eigenvalue for each detail coefficient set;

[0084] The approximate feature sequence construction unit is used to construct an approximate feature sequence by taking each first feature value as an element;

[0085] The detail feature sequence constructing unit is used to construct a detail feature sequence using each second feature value as an element.

[0086] Each row of the vibration trajectory matrix represents the evolution of the signal at different time scales. Performing wavelet transform on each row can better capture the local characteristics of the signal at different time and frequency scales, which is conducive to discovering the small vibration changes caused by synchronizer ring faults. The vibration trajectory matrix retains the time structure information of the original signal. Performing wavelet transform on each row can analyze the changing characteristics of the signal components when the signal changes with time at different scales, which is helpful to identify the dynamic characteristics of synchronizer ring faults.

[0087] The calculation formulas for the first eigenvalue and the second eigenvalue are: , where X is the first eigenvalue or the second eigenvalue, x i is the i-th element in the approximation coefficient set or the detail coefficient set, N is the number of elements in the set, and i is a positive integer.

[0088] After performing multi-layer wavelet transform on each row of the vibration trajectory matrix, there are multiple approximate coefficients and detail coefficients for each row. The present invention weights the approximate coefficients and detail coefficients of each row respectively, and characterizes the distribution of the approximate coefficients and detail coefficients by weighted values.

[0089] In this embodiment, the number of wavelet transform layers is 3, 4 or 5.

[0090] like Figure 2 As shown, the synchronizer gear ring detection system includes: a first feature fusion unit, a second feature fusion unit, a first tanh layer, a second tanh layer, a third tanh layer, a Concat layer, an output enhancement layer and a fully connected layer;

[0091] The first input end of the first feature fusion unit is used to input a peak value sequence, and the second input end thereof is used to input a valley value sequence;

[0092] The first input end of the second feature fusion unit is used to input the approximate feature sequence, and the second input end thereof is used to input the detail feature sequence;

[0093] The input end of the first tanh layer is connected to the output end of the first feature fusion unit;

[0094] The input end of the second tanh layer is connected to the output end of the second feature fusion unit;

[0095] The input end of the Concat layer is connected to the output end of the first tanh layer and the output end of the second tanh layer respectively, and its output end is connected to the input end of the third tanh layer;

[0096] The input end of the output enhancement layer is connected to the output end of the third tanh layer, and its output end is connected to the input end of the fully connected layer;

[0097] The output of the fully connected layer is used as the output of the synchronizer ring detection system.

[0098] The present invention adopts a convolution layer to process a peak sequence, a valley sequence, an approximate feature sequence and a detail feature sequence respectively to obtain four different types of features, adopts a multiplier M1 to fuse the features corresponding to the peak sequence with the features corresponding to the valley sequence to obtain a first fused feature, adopts a multiplier M2 to fuse the features corresponding to the approximate feature sequence with the features corresponding to the detail feature sequence to obtain a second fused feature, so that the first fused feature and the second fused feature are processed by a tanh layer respectively, so as to realize separate processing of the first fused feature and the second fused feature, respectively capture nonlinear relationships, and then adopt a Concat layer to perform feature splicing, adopt a third tanh layer to consider each output feature of the Concat layer, and adopt an output enhancement layer to enhance the output of the third tanh layer to make the significant features more significant.

[0099] The expression of the first feature fusion unit is: , where Y1 is the output of the first feature fusion unit, X p is the peak sequence, X v is the valley value sequence, Conv is the convolution operation, for ;

[0100] The expression of the second feature fusion unit is: , where Y2 is the output of the second feature fusion unit, A is the approximate feature sequence, and D is the detail feature sequence.

[0101] like Figure 2 As shown, the first feature fusion unit includes: a first convolutional layer, a second convolutional layer and a multiplier M1, and the second feature fusion unit includes: a third convolutional layer, a fourth convolutional layer and a multiplier M2. The convolution kernel sizes of the first convolutional layer, the second convolutional layer, the third convolutional layer and the fourth convolutional layer are .

[0102] The expressions for the first tanh layer and the second tanh layer are: , where h n is the nth eigenvalue output by the first tanh layer or the second tanh layer, tanh is the hyperbolic tangent activation function, x n is the nth eigenvalue of the first tanh layer or the second tanh layer input, n is a positive integer, ω x,n For x n The weight of b x,n For x n Bias of

[0103] The expression of the third tanh layer is: , where r n is the nth eigenvalue output by the third tanh layer, e n is the nth eigenvalue of the third tanh layer input, ω e,n for e n The weight of b e,n for e n The bias.

[0104] The expression of the output enhancement layer is: , where y n is the nth eigenvalue of the output enhancement layer, r n is the nth eigenvalue output by the third tanh layer, σ is the Sigmoid function, ω r,n For r n The weight of b r,n For r n The bias of , n is a positive integer.

[0105] The present invention assigns weights and biases to each eigenvalue output by the third tanh layer, and uses the Sigmoid function to calculate the enhancement ratio coefficient of the output enhancement layer. , adaptively adjust the eigenvalue size of the effective features.

[0106] In this embodiment, the weights and biases in the synchronizer gear ring detection system are obtained by training using a gradient descent method.

[0107] Embodiment 2, a synchronizer gear ring detection method, comprising the following steps:

[0108] S1. Install a triaxial accelerometer on the gearbox, continuously monitor the gearbox vibration during gear shifting, and obtain a triaxial vibration signal of the gearbox;

[0109] S2. Superimposing the three-axis vibration signals in the frequency domain to obtain an enhanced vibration signal;

[0110] S3, using morphological operations to extract local maximum and local minimum values ​​of the enhanced vibration signal, respectively, and constructing peak sequence and valley sequence;

[0111] S4, constructing a vibration trajectory matrix for the enhanced vibration signal, extracting features from the vibration trajectory matrix, and constructing an approximate feature sequence and a detail feature sequence;

[0112] S5. According to the synchronizer gear ring detection system, the peak sequence, the valley sequence, the approximate feature sequence and the detail feature sequence are processed to obtain the synchronizer gear ring fault condition.

[0113] The specific implementation process of Example 2 is the same as that of Example 1.

[0114] The present invention adopts a three-axis accelerometer to collect vibration signals of a transmission in three directions, superimposes the three-axis vibration signals in the frequency domain, fully retains the signal components in each direction, obtains an enhanced vibration signal, adopts morphological operations to extract local maximum values ​​and local minimum values ​​of the enhanced vibration signal respectively, reflects the extreme value characteristics of the vibration signal, then constructs a vibration trajectory matrix for the enhanced vibration signal, extracts features to form an approximate feature sequence and a detail feature sequence, reflects the component features of the enhanced vibration signal at different scales, and then adopts a synchronizer gear ring detection system to process the peak value sequence, the valley value sequence, the approximate feature sequence and the detail feature sequence, and comprehensively evaluates the state of the synchronizer gear ring through multi-feature fusion analysis, thereby improving the accuracy of synchronizer gear ring fault recognition.

[0115] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A synchronizer gear ring detection device, characterized in that: include: Three-axis accelerometer, signal enhancement system, first sequence building system, second sequence building system and synchronizer ring detection system; The three-axis accelerometer is installed on the gearbox to continuously monitor the gearbox vibration during gear shifting and obtain the gearbox three-axis vibration signal; The signal enhancement system is used to superimpose the triaxial vibration signals in the frequency domain to obtain an enhanced vibration signal; The first sequence construction system is used to extract local maximum values ​​and local minimum values ​​of the enhanced vibration signal by using morphological operations, and construct a peak sequence and a valley sequence; The second sequence construction system is used to construct a vibration trajectory matrix for the enhanced vibration signal, and extract features from the vibration trajectory matrix to construct an approximate feature sequence and a detail feature sequence, wherein each row of the vibration trajectory matrix represents the evolution process of the signal at different time scales; The synchronizer gear ring detection system is used to process the peak sequence, the valley sequence, the approximate characteristic sequence and the detail characteristic sequence to obtain the synchronizer gear ring fault value; The second sequence construction system comprises: a matrix construction unit, a wavelet transform unit, a first calculation unit, a second calculation unit, an approximate feature sequence construction unit and a detail feature sequence construction unit; The matrix construction unit is used to construct a vibration trajectory matrix for the enhanced vibration signal; The wavelet transform unit is used to perform multi-layer wavelet transform on each row of the vibration trajectory matrix to obtain an approximate coefficient set and a detail coefficient set for each row; The first calculation unit is used to calculate a first eigenvalue for each approximation coefficient set; The second calculation unit is used to calculate a second eigenvalue for each detail coefficient set; The approximate feature sequence construction unit is used to construct an approximate feature sequence by taking each first feature value as an element; The detail feature sequence construction unit is used to construct a detail feature sequence by taking each second feature value as an element; The calculation formulas for the first eigenvalue and the second eigenvalue are both: , where X is the first eigenvalue or the second eigenvalue, x i is the i-th element in the approximate coefficient set or the detail coefficient set, N is the number of elements in the set, and i is a positive integer; The synchronizer gear ring detection system comprises: a first feature fusion unit, a second feature fusion unit, a first tanh layer, a second tanh layer, a third tanh layer, a Concat layer, an output enhancement layer and a fully connected layer; The first input end of the first feature fusion unit is used to input a peak value sequence, and the second input end thereof is used to input a valley value sequence; The first input end of the second feature fusion unit is used to input the approximate feature sequence, and the second input end thereof is used to input the detail feature sequence; The input end of the first tanh layer is connected to the output end of the first feature fusion unit; The input end of the second tanh layer is connected to the output end of the second feature fusion unit; The input end of the Concat layer is connected to the output end of the first tanh layer and the output end of the second tanh layer respectively, and the output end thereof is connected to the input end of the third tanh layer; The input end of the output enhancement layer is connected to the output end of the third tanh layer, and the output end of the output enhancement layer is connected to the input end of the fully connected layer; The output end of the fully connected layer is used as the output end of the synchronizer ring gear detection system; The expression of the first feature fusion unit is: , where Y1 is the output of the first feature fusion unit, X p is the peak sequence, X v is the valley value sequence, Conv is the convolution operation, for ; The expression of the second feature fusion unit is: , where Y2 is the output of the second feature fusion unit, A is the approximate feature sequence, and D is the detail feature sequence; The expressions of the first tanh layer and the second tanh layer are both: , where h n is the nth eigenvalue output by the first tanh layer or the second tanh layer, tanh is the hyperbolic tangent activation function, x n is the nth eigenvalue of the first tanh layer or the second tanh layer input, n is a positive integer, ω x,n For x n The weight of b x,n For x n Bias of The expression of the third tanh layer is: , where r n is the nth eigenvalue output by the third tanh layer, e n is the nth eigenvalue of the third tanh layer input, ω e,n for e n The weight of b e,n for e n Bias of The expression of the output enhancement layer is: , where y n is the nth eigenvalue of the output enhancement layer, r n is the nth eigenvalue output by the third tanh layer, σ is the Sigmoid function, ω r,n For r n The weight of b r,n For r n The bias of , n is a positive integer.

2. The synchronizer ring detection device according to claim 1, characterized in that: The signal enhancement system comprises: a Fourier transform unit, a superposition unit and an inverse Fourier transform unit; The Fourier transform unit is used to perform Fourier transform on the three-axis vibration signals respectively to obtain an X-axis vibration frequency domain signal, a Y-axis vibration frequency domain signal and a Z-axis vibration frequency domain signal; The superposition unit is used to superimpose the X-axis vibration frequency domain signal, the Y-axis vibration frequency domain signal and the Z-axis vibration frequency domain signal to obtain a frequency domain superposition signal; The inverse Fourier transform unit is used to perform inverse Fourier transform on the frequency domain superposition signal to obtain an enhanced vibration signal.

3. The synchronizer gear ring detection device according to claim 1, characterized in that: The first sequence construction system comprises: an opening operation unit, a first subtraction unit, a peak sequence construction unit, a closing operation unit, a second subtraction unit and a valley sequence construction unit; The opening operation unit is used to perform an opening operation on the enhanced vibration signal; The first subtraction unit is used to subtract the opening operation result from the enhanced vibration signal to obtain a peak significant signal; The peak sequence construction unit is used to detect the local maximum value of the peak significant signal by using the expansion operation, and construct the peak sequence with all the local maximum values ​​as elements; The closing operation unit is used to perform a closing operation on the enhanced vibration signal; The second subtraction unit is used to subtract the enhanced vibration signal from the closed operation result to obtain a valley significant signal; The valley sequence construction unit is used to detect local minimum values ​​of valley significant signals by using corrosion operations, and construct a valley sequence using all local minimum values ​​as elements.

4. A synchronizer gear ring detection method, implemented based on the synchronizer gear ring detection device according to any one of claims 1 to 3, characterized in that: The following steps are involved: S1. Install a triaxial accelerometer on the gearbox, continuously monitor the gearbox vibration during gear shifting, and obtain a triaxial vibration signal of the gearbox; S2. Superimposing the three-axis vibration signals in the frequency domain to obtain an enhanced vibration signal; S3, using morphological operations to extract local maximum and local minimum values ​​of the enhanced vibration signal, respectively, and constructing peak sequence and valley sequence; S4, constructing a vibration trajectory matrix for the enhanced vibration signal, extracting features from the vibration trajectory matrix, and constructing an approximate feature sequence and a detail feature sequence; S5. According to the synchronizer gear ring detection system, the peak sequence, the valley sequence, the approximate feature sequence and the detail feature sequence are processed to obtain the synchronizer gear ring fault condition.

Images