Motor fault detection method and system based on multi-branch neural network

Abstract

The invention discloses a motor fault detection method and system based on a multi-branch neural network, and the multi-branch neural network in the method is obtained through the following steps: obtaining a multi-dimensional data set from a motor, and carrying out the preprocessing of the multi-dimensional data set, and obtaining a motor training set; inputting the motor training set into a local branch framework to obtain a first branch feature set; inputting the motor training set into a global branch framework to obtain a second branch feature set; inputting the motor training set into the time sequence branch framework to obtain a third branch feature set; according to the first branch feature set, attention fusion is carried out on the second branch feature set and the third branch feature set, and a fusion data set is obtained; and according to the fusion data set, performing parameter updating on the initialized multi-branch neural network to obtain a trained multi-branch neural network. The detection method can effectively improve the detection efficiency and detection accuracy of motor faults. The invention relates to the technical field of deep learning.

Description

Technical Field

[0001] The present invention relates to the field of deep learning technology, and in particular to a motor fault detection method and system based on a multi-branch neural network. Background Art

[0002] As a typical rotating machine, the motor often runs at high speed and high load conditions. Timely fault detection of the motor is conducive to ensuring the stability and safety of industrial production.

[0003] At present, traditional motor fault detection methods mainly use support vector machines (SVM), random forest models, etc. to realize motor fault diagnosis. These methods often rely on manually extracted features and have low efficiency. In addition, the actual industrial environment is usually accompanied by non-stationary working conditions such as start-stop, speed change, and impact load. These non-stationary working conditions can easily lead to misdiagnosis of motor diagnosis and low accuracy.

[0004] Therefore, the problems existing in the prior art still need to be solved and optimized. Summary of the invention

[0005] The purpose of the present invention is to solve one of the technical problems existing in the related art to at least a certain extent.

[0006] To this end, an object of an embodiment of the present invention is to provide a motor fault detection method based on a multi-branch neural network, which can effectively improve the detection efficiency and detection accuracy of motor faults.

[0007] Another object of an embodiment of the present application is to provide a motor fault detection system based on a multi-branch neural network.

[0008] In order to achieve the above technical objectives, the technical solutions adopted by the embodiments of the present application include:

[0009] In a first aspect, an embodiment of the present application provides a motor fault detection method based on a multi-branch neural network, comprising:

[0010] Get motor signal data;

[0011] Inputting the motor signal data into a trained multi-branch neural network to obtain a motor fault detection result;

[0012] Wherein, the multi-branch neural network includes a local branch framework, a global branch framework and a time series branch framework, and the local branch framework, the global branch framework and the time series branch framework have a parallel relationship, and the trained multi-branch neural network is obtained by the following steps:

[0013] Acquire a multidimensional data set from a motor, and preprocess the multidimensional data set to obtain a motor training set;

[0014] Inputting the motor training set into the local branch framework to obtain a first branch feature set, wherein the first branch feature set is used to characterize a set of local feature information corresponding to the motor training set;

[0015] Inputting the motor training set into a global branch framework to obtain a second branch feature set, wherein the second branch feature set is used to characterize a set of global context information corresponding to the motor training set;

[0016] Inputting the motor training set into the time series branch framework to obtain a third branch feature set, wherein the third branch feature set is used to characterize a set of time series information corresponding to the motor training set;

[0017] According to the first branch feature set, performing attention fusion on the second branch feature set and the third branch feature set to obtain a fused data set;

[0018] According to the fused data set, the parameters of the initialized multi-branch neural network are updated to obtain the trained multi-branch neural network.

[0019] In addition, the motor fault detection method according to the above embodiment of the present application may also have the following additional technical features:

[0020] Further, in one embodiment of the present application, the multidimensional data set includes several sensor data subsets, each of the sensor data subsets corresponds to a different vibration sensor, and the multidimensional data set is preprocessed to obtain a motor training set, including:

[0021] According to the diffusion entropy calculation function, weight calculation is performed on the sensor data subsets to obtain a sensor weight corresponding to each of the sensor data subsets;

[0022] According to the sensor weights, data fusion is performed on all the sensor data subsets to obtain a fused data set;

[0023] Noise is added to the fused data set to obtain the motor training set.

[0024] Further, in one embodiment of the present application, the global branch framework includes a plurality of global context shrinking networks, the global context shrinking networks are in a series relationship, and the global context shrinking networks are used to perform the following steps:

[0025] Acquire an input data set, and perform nonlinear transformation on the input data set to obtain a linear transformation data set;

[0026] Performing global context feature extraction on the linear transformation data set to obtain a global data set;

[0027] According to the linear transformation data set, the global data set is subjected to denoising extraction to obtain an output data set corresponding to the global context shrinkage network.

[0028] Further, in one embodiment of the present application, global context feature extraction is performed on the linear transformation data set to obtain a global data set:

[0029] Performing feature combination on the linear transformation data set to obtain a combined data set;

[0030] Performing a first bottleneck transformation on the combined data set to obtain a first bottleneck data set, and performing a second bottleneck transformation on the combined data set to obtain a second bottleneck data set;

[0031] Performing soft thresholding processing on the second bottleneck data set to obtain a first soft threshold data set;

[0032] The first bottleneck data set is element-wise multiplied according to the first soft threshold data set to obtain the global data set.

[0033] Further, in one embodiment of the present application, performing denoising extraction on the global data set according to the linear transformation data set to obtain an output data set corresponding to the global context shrinkage network includes:

[0034] According to the global data set, performing soft threshold extraction on the linear transformation data set to obtain a second soft threshold data set;

[0035] According to the linear transformation data set, data splicing is performed on the second soft threshold data set to obtain the output data set.

[0036] Further, in an embodiment of the present application, the attention fusion is performed on the second branch feature set and the third branch feature set according to the first branch feature set to obtain a fused data set, including:

[0037] According to the third branch feature set, feature splicing is performed on the second branch feature set to obtain a spliced ​​feature set;

[0038] According to the concatenated feature set, the first branch feature set is subjected to attention weighted fusion to obtain the fused data set.

[0039] In a second aspect, an embodiment of the present application provides a motor fault detection system based on a multi-branch neural network, comprising:

[0040] An acquisition module, used for acquiring motor signal data;

[0041] A processing module, used for inputting the motor signal data into a trained multi-branch neural network to obtain a motor fault detection result;

[0042] Wherein, the multi-branch neural network includes a local branch framework, a global branch framework and a time series branch framework, and the local branch framework, the global branch framework and the time series branch framework have a parallel relationship, and the trained multi-branch neural network is obtained by the following steps:

[0043] Acquire a multidimensional data set from a motor, and preprocess the multidimensional data set to obtain a motor training set;

[0044] Inputting the motor training set into the local branch framework to obtain a first branch feature set, wherein the first branch feature set is used to characterize a set of local feature information corresponding to the motor training set;

[0045] Inputting the motor training set into a global branch framework to obtain a second branch feature set, wherein the second branch feature set is used to characterize a set of global context information corresponding to the motor training set;

[0046] Inputting the motor training set into the time series branch framework to obtain a third branch feature set, wherein the third branch feature set is used to characterize a set of time series information corresponding to the motor training set;

[0047] According to the first branch feature set, performing attention fusion on the second branch feature set and the third branch feature set to obtain a fused data set;

[0048] According to the fused data set, the parameters of the initialized multi-branch neural network are updated to obtain the trained multi-branch neural network.

[0049] Furthermore, in an embodiment of the present application, the global branch framework includes a plurality of global context shrinking networks, the global context shrinking networks are in a series relationship, and the global context shrinking networks are used to perform the following steps:

[0050] Acquire an input data set, and perform nonlinear transformation on the input data set to obtain a linear transformation data set;

[0051] Performing global context feature extraction on the linear transformation data set to obtain a global data set;

[0052] According to the linear transformation data set, the global data set is subjected to denoising extraction to obtain an output data set corresponding to the global context shrinkage network.

[0053] In a third aspect, an embodiment of the present application further provides an electronic device, including:

[0054] at least one processor;

[0055] at least one memory for storing at least one program;

[0056] When the at least one program is executed by the at least one processor, the at least one processor implements the method of the first aspect described above.

[0057] In a fourth aspect, an embodiment of the present application further provides a computer-readable storage medium, which stores a program executable by a processor, and the program executable by the processor is used to implement the method of the first aspect when executed by the processor.

[0058] The advantages and benefits of the present application will be partially given in the following description, and partially become apparent from the following description, or be understood through the practice of the present application:

[0059] The embodiment of the present application discloses a motor fault detection method and system based on a multi-branch neural network, wherein the detection method obtains motor signal data; inputs the motor signal data into a trained multi-branch neural network to obtain a motor fault detection result; wherein the multi-branch neural network includes a local branch framework, a global branch framework and a time series branch framework, and the local branch framework, the global branch framework and the time series branch framework have a parallel relationship, and the trained multi-branch neural network is obtained by the following steps: obtaining a multidimensional data set from a motor, and preprocessing the multidimensional data set to obtain a motor training set; inputting the motor training set into the local branch framework to obtain a first branch feature set, the first branch feature set A branch feature set is used to characterize the set of local feature information corresponding to the motor training set; the motor training set is input into the global branch framework to obtain a second branch feature set, the second branch feature set is used to characterize the set of global context information corresponding to the motor training set; the motor training set is input into the time series branch framework to obtain a third branch feature set, the third branch feature set is used to characterize the set of time series information corresponding to the motor training set; according to the first branch feature set, the second branch feature set and the third branch feature set are subjected to attention fusion to obtain a fused data set; according to the fused data set, the parameters of the initialized multi-branch neural network are updated to obtain the trained multi-branch neural network. The detection method is based on three parallel local branch frameworks, a global branch framework, and a time series branch framework, and can extract accurate and reliable feature information more quickly, and by performing attention fusion on the feature information output by the three branch frameworks, the detection efficiency and accuracy of motor faults are effectively improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the embodiments of the present application or the drawings of the related technical solutions in the prior art are introduced below. It should be understood that the drawings introduced below are only for the convenience of clearly expressing some embodiments of the technical solutions of the present application. For those skilled in the art, other drawings can be obtained based on these drawings without creative work.

[0061] Figure 1 A schematic diagram of a motor fault detection method based on a multi-branch neural network provided in an embodiment of the present application;

[0062] Figure 2 A schematic diagram of a training process of a multi-branch neural network provided in an embodiment of the present application;

[0063] Figure 3 A schematic diagram of the structure of a motor fault detection system based on a multi-branch neural network provided in an embodiment of the present application;

[0064] Figure 4 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0065] Embodiments of the present application are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present application, and are not to be construed as limitations on the present application. For the step numbers in the following embodiments, they are only provided for the convenience of explanation, and the order between the steps is not limited in any way, and the execution order of each step in the embodiment can be adaptively adjusted according to the understanding of those skilled in the art.

[0066] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which this application belongs. The terms used herein are only for the purpose of describing the embodiments of this application and are not intended to limit this application.

[0067] At present, traditional motor fault detection methods mainly use support vector machines (SVM) and random forest models to realize motor fault diagnosis. These methods often rely on manually extracted features and are inefficient. In addition, the actual industrial environment is usually accompanied by non-stationary working conditions such as start-stop, speed change, and impact load, which makes the collected motor signals mixed with a lot of noise. These methods have limited performance in processing strong noise signals, which easily leads to misdiagnosis of motor diagnosis and low accuracy. In addition, since a single sensor signal may be interfered by noise and the integrity of the information is affected, the accuracy and credibility of the results of motor fault diagnosis based on a single sensor are often unsatisfactory.

[0068] In view of this, an embodiment of the present invention provides a motor fault detection method based on a multi-branch neural network. The detection method is based on the preprocessing of a multidimensional data set. Specifically, the sensor weight corresponding to each sensor is calculated through a diffuse entropy calculation function, and data fusion is performed according to the sensor weight. The correlation and complementarity of data collected by multiple sensors can be fully utilized, and the reliability of sensor data fusion can be significantly improved. In addition, the detection method is based on parallel training and learning of a local branch framework, a global branch framework, and a time series branch framework, and based on an attention mechanism, the feature sets output by the three branch frameworks are fused. It can still provide relatively accurate and reliable diagnostic information under the influence of a high noise environment, and effectively improve the accuracy and detection efficiency of motor fault detection.

[0069] Reference Figure 1 and Figure 2 In an embodiment of the present application, a motor fault detection method based on a multi-branch neural network includes:

[0070] Step 110, obtaining motor signal data;

[0071] Step 120, inputting the motor signal data into a trained multi-branch neural network to obtain a motor fault detection result;

[0072] In an embodiment of the present application, the signal data about the motor collected by the sensor can be input into a trained multi-branch neural network, and the trained multi-branch neural network can be used to predict whether the motor is faulty, thereby obtaining a motor fault detection result that records whether the motor is faulty.

[0073] Wherein, the multi-branch neural network includes a local branch framework, a global branch framework and a time series branch framework, and the local branch framework, the global branch framework and the time series branch framework have a parallel relationship, and the trained multi-branch neural network is obtained by the following steps:

[0074] Step 130, obtaining a multidimensional data set from a motor, and preprocessing the multidimensional data set to obtain a motor training set;

[0075] In some embodiments, the multidimensional data set includes several sensor data subsets, each of the sensor data subsets corresponds to a different vibration sensor, and the preprocessing of the multidimensional data set to obtain the motor training set includes:

[0076] A1. Performing weight calculation on the sensor data subsets according to a diffusion entropy calculation function to obtain a sensor weight corresponding to each of the sensor data subsets;

[0077] A2. According to the sensor weights, data fusion is performed on all the sensor data subsets to obtain a fused data set;

[0078] A3. Adding noise to the fused data set to obtain the motor training set.

[0079] In the embodiment of the present application, the multidimensional data set of the motor can be obtained by deploying multiple high-precision vibration sensors at different positions of the motor, such as bearings, bases or transmission parts. Each vibration sensor converts mechanical vibration into electrical signals through its built-in sensitive elements, and then these electrical signals are processed by preliminary amplification, filtering, etc. to obtain a multidimensional data set.

[0080] It can be understood that the diffusion entropy calculation function is used to describe the degree of irregularity of the pattern sequence. Specifically, for the electrical signal sampling points corresponding to a certain vibration sensor, its time series X = {x1, x2, ..., x j ,…,x N}, x j is the time subsequence of the jth electrical signal sampling point, and N is the number of sampling points. Use the normal distribution function to map the time series to a new sequence Y, Y = {y j},j=1,2,…,N,y j ∈(0,1);

[0081]

[0082] Where μ is the expected value, σ is the standard deviation, and y j For x j The corresponding new time subseries; t is the integral variable, and exp is the natural exponential function.

[0083] Then for y j Perform a linear transformation so that y j Assigned to an integer class, the equivalent formula for linear transformation can be expressed as:

[0084]

[0085] in, for y j The corresponding integer class ranges from [1,2,…,c]; int is the rounding function and c is the category number.

[0086] Calculate the embedding vector corresponding to each integer class Each embedding vector with embedding dimension m and time delay d is created according to the following formula

[0087]

[0088] Each embedding vector Mapping to scatter mode in d is the time delay, m is the embedding dimension, c m represents the total number of potential scatter modes when the embedding dimension is m.

[0089] For each scatter mode, its relative frequency is calculated as:

[0090]

[0091] Among them, Num is the summation function, which is used to solve the scatter pattern Assigned to the embedding vector the number of is the relative frequency.

[0092] The equivalent expression of the diffusion entropy calculation function can be:

[0093]

[0094] Where DE(X) is the diffusion entropy calculation function, and ln represents the natural logarithm.

[0095] It can be understood that the larger the DE value is, the greater the complexity of the sequence is, and the higher the irregularity of the sequence is. Specifically, step A1 may first calculate the diffusion entropy (DE) value of the sensor data subset corresponding to each vibration sensor. For a certain acquisition signal in the sensor data subset, the equivalent expression of its diffusion entropy value is:

[0096] B i =DE(X i ),i=1~n

[0097] Among them, B i is the diffusion entropy value X corresponding to the i-th collected vibration signal i ; DE() is the diffusion entropy calculation function, and n is the total number of vibration sensors.

[0098] After obtaining the diffuse entropy values ​​of all collected signals of each sensor, the redundancy of the corresponding vibration sensor can be calculated according to the set of diffuse entropy values ​​corresponding to each sensor data subset. The equivalent expression of its redundancy is:

[0099] D i =|1-B i |,i=1~n

[0100] Among them, D i is the redundancy of the diffuse entropy value of the i-th vibration sensor.

[0101] According to the redundancy corresponding to each vibration sensor, a weight is assigned to the sensor data subset corresponding to the vibration sensor. The equivalent expression of the sensor weight of a sensor data subset is:

[0102]

[0103] Among them, w i is the sensor weight of the i-th sensor data subset.

[0104] It should be noted that the data fusion in step A2 can be based on the sensor weights corresponding to each sensor data subset, and the sensor data subsets are weightedly fused to obtain a fused data set corresponding to multiple sensor data subsets. Step A3 can first obtain a preset signal-to-noise ratio (SNR) value, and then based on the preset signal-to-noise ratio value, use a random noise generation algorithm to generate a noise signal of corresponding intensity, and superimpose the generated noise signal on the fused data set to obtain a motor training set.

[0105] Step 140: input the motor training set into the local branch framework to obtain a first branch feature set, where the first branch feature set is used to characterize a set of local feature information corresponding to the motor training set;

[0106] In an embodiment of the present application, a local branch framework is used to extract deep feature representations in a motor training set and capture local features and detail information in the motor training set. It may include multiple residual neural networks (ResNet), and the multiple residual neural networks are connected in series. The output of the previous residual neural network is used as the input of the next residual neural network, and the output of the last residual neural network is used as the first branch feature set.

[0107] Step 150: input the motor training set into a global branch framework to obtain a second branch feature set, where the second branch feature set is used to characterize a set of global context information corresponding to the motor training set;

[0108] In an embodiment of the present application, the global branch framework includes an input layer, a convolution (Conv) layer, multiple global context shrinkage networks, a batch normalization layer, a ReLU activation layer, a global average pooling (GAP) layer and a fully connected (FC) layer in sequence, and the motor training set is adaptively denoised and signal feature extracted through the Conv layer and multiple global context shrinkage networks.

[0109] In some embodiments, the global branch framework in step 150 includes a plurality of global context shrinking networks, the global context shrinking networks are in series relationship, and the global context shrinking networks are used to perform the following steps:

[0110] B1. Obtain an input data set, and perform nonlinear transformation on the input data set to obtain a linear transformation data set;

[0111] In an embodiment of the present application, for the first global context contraction network in a series relationship, its input data set can be a motor training set; and for the second and subsequent global context contraction networks in a series relationship, their input data can be the output data set of the previous global context contraction network.

[0112] It can be understood that step B1 can be to input the input data set into two series-connected Conv layers in the global context shrinkage network for nonlinear transformation, and use the features output by the second Conv layer as the linear transformation data set.

[0113] B2. Extracting global context features from the linear transformation data set to obtain a global data set;

[0114] Furthermore, in step B2, global context feature extraction is performed on the linear transformation data set to obtain a global data set:

[0115] B21, performing feature combination on the linear transformation data set to obtain a combined data set;

[0116] B22, performing a first bottleneck transformation on the combined data set to obtain a first bottleneck data set, and performing a second bottleneck transformation on the combined data set to obtain a second bottleneck data set;

[0117] B23, performing soft thresholding processing on the second bottleneck data set to obtain a first soft threshold data set;

[0118] B24. Perform element-wise multiplication on the first bottleneck data set according to the first soft threshold data set to obtain the global data set.

[0119] In an embodiment of the present application, step B21 is used to combine all features in the linear transformation data set into a global context feature. Specifically, the feature combination in step B21 can be to perform a nonlinear transformation on the linear transformation data set based on a 1×1 convolution layer, and then use the Softmax activation function to obtain the corresponding attention weight, and finally obtain a combined data set representing the global context feature information through attention pooling.

[0120] It can be understood that the first bottleneck transformation and the second bottleneck transformation in step B22 are used to capture the interdependence between channels, realize dimensional compression of global context features, and reduce the burden of multi-branch neural networks. Specifically, the first bottleneck transformation in step B22 can be inputting the combined data set into a 1×1 convolution layer, a layer norm, a 1×1 convolution layer, a ReLU layer, an absolute value function layer, and a global bottleneck pooling (GAP) layer in sequence to obtain a first bottleneck data set; the second bottleneck transformation in step B22 can be inputting the combined data set into another bottleneck transformation branch, which is composed of a 1×1 convolution layer, a layer norm, a 1×1 convolution layer, and a ReLU layer in sequence to obtain a second bottleneck data set.

[0121] The soft thresholding process of step B23 may be to sequentially input the second bottleneck data set into an absolute value function layer, a global bottleneck pooling (GAP) layer, and two fully connected layers, and then use a Sigmoid function to scale the output of the second fully connected layer to a range of (0, 1), thereby obtaining a first soft threshold data set. Step B24 may perform an element-by-element multiplication operation on each feature element in the first soft threshold data set and each feature element in the first bottleneck data set, thereby obtaining a global data set.

[0122] B3. According to the linear transformation data set, perform noise reduction extraction on the global data set to obtain an output data set corresponding to the global context shrinkage network.

[0123] Furthermore, the step B3, performing denoising extraction on the global data set according to the linear transformation data set to obtain an output data set corresponding to the global context shrinkage network, includes:

[0124] B31. Performing soft threshold extraction on the linear transformation data set according to the global data set to obtain a second soft threshold data set;

[0125] B32. According to the linear transformation data set, perform data splicing on the second soft threshold data set to obtain the output data set.

[0126] In an embodiment of the present application, the global data set in step B31 is used to provide dynamic soft thresholds of different channels to the soft threshold module in the global branch framework, and the soft threshold module can set the features within the threshold interval to zero to achieve the effect of suppressing noise. Specifically, the soft threshold extraction in step B31 can be to input the linear transformation data set into the updated soft threshold module, and the threshold interval in the updated soft threshold module is determined based on the latest received global data set, so as to obtain the second soft threshold data set. Step B32 can be based on a shortcut identity connection, and each feature element in the linear transformation data set and each feature element in the second soft threshold data set are added element by element to obtain the output data set.

[0127] It should be noted that, for the last global context shrinkage network in the series relationship, its output data set can be sequentially input into the batch normalization layer, the ReLU activation layer, the global average pooling (GAP) layer, and the fully connected (FC) layer to obtain the second branch feature set output by the global branch framework.

[0128] Step 160, inputting the motor training set into the time series branch framework to obtain a third branch feature set, wherein the third branch feature set is used to characterize a set of time series information corresponding to the motor training set;

[0129] In an embodiment of the present application, a time series branch framework can be constructed based on a Grey Wolf Algorithm-optimized gated recurrent unit neural network, by applying the global search capability of the Grey Wolf Algorithm (GWO) to the parameter optimization of the gated recurrent unit (GRU), so as to find the optimal gating parameters of the gated recurrent unit; then, based on the update gate and reset gate in the gated recurrent unit at the optimal gating parameters, the long-term dependency information contained in the sequence data in the input motor training set is captured, so as to obtain a set having the time series information of the motor training set (i.e., the third branch feature set).

[0130] Step 170: Based on the first branch feature set, perform attention fusion on the second branch feature set and the third branch feature set to obtain a fused data set;

[0131] In some embodiments, the step 170, performing attention fusion on the second branch feature set and the third branch feature set according to the first branch feature set to obtain a fused data set, includes:

[0132] C1. Based on the third branch feature set, perform feature splicing on the second branch feature set to obtain a spliced ​​feature set;

[0133] C2. According to the concatenated feature set, the first branch feature set is subjected to attention weighted fusion to obtain the fused data set.

[0134] In an embodiment of the present application, the first branch feature set can be used as a key and value, the second branch feature set can be used as a query representation, and the third branch feature set can be used as a dynamic time series weight in the weighted fusion process of the attention mechanism. Specifically, step C1 can be to concatenate the third branch feature set and the second branch feature set to obtain a concatenated feature set; step C2 can be to use the concatenated feature set as the first input of the attention mechanism module, and the first branch feature set as the second input of the attention mechanism module, and obtain a fused data set by calculating the attention weight and weighted fusion.

[0135] Step 180: Update the parameters of the initialized multi-branch neural network according to the fused data set to obtain the trained multi-branch neural network.

[0136] It is understandable that for machine learning models, the accuracy of the model prediction results can be measured by a loss function, which is defined on a single training data and is used to measure the prediction error of a training data. Specifically, the loss value of the training data is determined by the label of a single training data and the prediction result of the model for the training data. In actual training, a training data set has a lot of training data, so a cost function is generally used to measure the overall error of the training data set. The cost function is defined on the entire training data set and is used to calculate the average value of the prediction error of all training data, which can better measure the prediction effect of the model. For general machine learning models, based on the aforementioned cost function, plus a regularization term that measures the complexity of the model, it can be used as the objective function of the training, and the loss value of the entire training data set can be calculated based on the objective function. There are many types of commonly used loss functions, such as 0-1 loss function, square loss function, absolute loss function, logarithmic loss function, cross entropy loss function, etc., which can all be used as loss functions of machine learning models, which will not be described one by one here. In the embodiment of the present application, one loss function can be selected to determine the loss value of the training, such as the cross entropy loss function. Based on the training loss value, the back propagation algorithm is used to update the parameters of the model. After several iterations, a trained multi-branch neural network can be obtained. The specific number of iterations can be set in advance, or the training is considered complete when the test set reaches the accuracy requirement.

[0137] A motor fault detection system based on a multi-branch neural network proposed according to an embodiment of the present application is described in detail below with reference to the accompanying drawings.

[0138] Reference Figure 3 , a motor fault detection system based on a multi-branch neural network proposed in an embodiment of the present application includes:

[0139] An acquisition module 101 is used to acquire motor signal data;

[0140] A processing module 102, used for inputting the motor signal data into a trained multi-branch neural network to obtain a motor fault detection result;

[0141] Wherein, the multi-branch neural network includes a local branch framework, a global branch framework and a time series branch framework, and the local branch framework, the global branch framework and the time series branch framework have a parallel relationship, and the trained multi-branch neural network is obtained by the following steps:

[0142] Acquire a multidimensional data set from a motor, and preprocess the multidimensional data set to obtain a motor training set;

[0143] Inputting the motor training set into the local branch framework to obtain a first branch feature set, wherein the first branch feature set is used to characterize a set of local feature information corresponding to the motor training set;

[0144] Inputting the motor training set into a global branch framework to obtain a second branch feature set, wherein the second branch feature set is used to characterize a set of global context information corresponding to the motor training set;

[0145] Inputting the motor training set into the time series branch framework to obtain a third branch feature set, wherein the third branch feature set is used to characterize a set of time series information corresponding to the motor training set;

[0146] According to the first branch feature set, performing attention fusion on the second branch feature set and the third branch feature set to obtain a fused data set;

[0147] According to the fused data set, the parameters of the initialized multi-branch neural network are updated to obtain the trained multi-branch neural network.

[0148] In some embodiments, the global branch framework includes a plurality of global context shrinking networks, the global context shrinking networks are in series relationship, and the global context shrinking networks are used to perform the following steps:

[0149] Acquire an input data set, and perform nonlinear transformation on the input data set to obtain a linear transformation data set;

[0150] Performing global context feature extraction on the linear transformation data set to obtain a global data set;

[0151] According to the linear transformation data set, the global data set is subjected to denoising extraction to obtain an output data set corresponding to the global context shrinkage network.

[0152] It can be understood that the contents of the above method embodiments are all applicable to the present system embodiments, the functions specifically implemented by the present system embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0153] Reference Figure 4 , the embodiment of the present application further provides an electronic device, including:

[0154] at least one processor 201;

[0155] At least one memory 202, used to store at least one program;

[0156] When the at least one program is executed by the at least one processor 201, the at least one processor 201 implements the above method embodiment.

[0157] Similarly, it can be understood that the contents of the above method embodiments are all applicable to the present device embodiments, the functions specifically implemented by the present device embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0158] The embodiment of the present application further provides a computer-readable storage medium, in which a program executable by the processor 201 is stored. The program executable by the processor 201 is used to implement the above-mentioned method embodiment when executed by the processor 201.

[0159] Similarly, the contents of the above method embodiments are all applicable to the computer-readable storage medium embodiments. The functions specifically implemented by the computer-readable storage medium embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0160] In some selectable embodiments, the function / operation mentioned in the block diagram may not occur in the order mentioned in the operation diagram. For example, depending on the function / operation involved, the two boxes shown in succession can actually be executed substantially simultaneously or the boxes can sometimes be executed in reverse order. In addition, the embodiment presented and described in the flow chart of the application is provided by way of example, for the purpose of providing a more comprehensive understanding of technology. The disclosed method is not limited to the operation and logic flow presented herein. Selectable embodiments are expected, wherein the order of various operations is changed and the sub-operation of a part described as a larger operation is performed independently.

[0161] In addition, although the present application is described in the context of functional modules, it should be understood that, unless otherwise specified, one or more of the functions and / or features can be integrated into a single physical device and / or software module, or one or more functions and / or features can be implemented in a separate physical device or software module. It is also understood that a detailed discussion of the actual implementation of each module is unnecessary for understanding the present application. More specifically, in view of the properties, functions, and internal relationships of the various functional modules in the device disclosed herein, the actual implementation of the module will be understood within the conventional techniques of the engineer. Therefore, those skilled in the art can implement the present application set forth in the claims without excessive experimentation using ordinary techniques. It is also understood that the specific concepts disclosed are merely illustrative and are not intended to limit the scope of the present application, which is determined by the full scope of the attached claims and their equivalents.

[0162] If the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the method of the embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0163] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0164] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), a fiber optic device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering or, if necessary, processing in another suitable manner, and then stored in a computer memory.

[0165] It should be understood that the various parts of the present application can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0166] In the above description of this specification, the description with reference to the terms "one embodiment / example", "another embodiment / example" or "certain embodiments / examples" 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 present application. 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 may be combined in any one or more embodiments or examples in a suitable manner.

[0167] Although the embodiments of the present application have been shown and described, those skilled in the art will appreciate that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present application, and that the scope of the present application is defined by the claims and their equivalents.

[0168] The above is a specific description of the preferred implementation of the present application, but the present application is not limited to the embodiments. Technical personnel familiar with the field can make various equivalent modifications or substitutions without violating the spirit of the present application. These equivalent modifications or substitutions are all included in the scope defined by the claims of the present application.

Claims

1. A motor fault detection method based on a multi-branch neural network, characterized in that: include: Get motor signal data; Inputting the motor signal data into a trained multi-branch neural network to obtain a motor fault detection result; Wherein, the multi-branch neural network includes a local branch framework, a global branch framework and a time series branch framework, and the local branch framework, the global branch framework and the time series branch framework have a parallel relationship, and the trained multi-branch neural network is obtained by the following steps: Acquire a multidimensional data set from a motor, and preprocess the multidimensional data set to obtain a motor training set; Inputting the motor training set into the local branch framework to obtain a first branch feature set, wherein the first branch feature set is used to characterize a set of local feature information corresponding to the motor training set; Inputting the motor training set into a global branch framework to obtain a second branch feature set, wherein the second branch feature set is used to characterize a set of global context information corresponding to the motor training set; Inputting the motor training set into the time series branch framework to obtain a third branch feature set, wherein the third branch feature set is used to characterize a set of time series information corresponding to the motor training set; According to the first branch feature set, performing attention fusion on the second branch feature set and the third branch feature set to obtain a fused data set; According to the fused data set, the parameters of the initialized multi-branch neural network are updated to obtain the trained multi-branch neural network.

2. The motor fault detection method according to claim 1, characterized in that: The multidimensional data set includes a plurality of sensor data subsets, each of which corresponds to a different vibration sensor, and the multidimensional data set is preprocessed to obtain a motor training set, including: According to the diffusion entropy calculation function, weight calculation is performed on the sensor data subsets to obtain a sensor weight corresponding to each of the sensor data subsets; According to the sensor weights, data fusion is performed on all the sensor data subsets to obtain a fused data set; Noise is added to the fused data set to obtain the motor training set.

3. The motor fault detection method according to claim 1, characterized in that: The global branch framework includes a plurality of global context shrinking networks, the global context shrinking networks are in series connection, and the global context shrinking networks are used to perform the following steps: Acquire an input data set, and perform nonlinear transformation on the input data set to obtain a linear transformation data set; Performing global context feature extraction on the linear transformation data set to obtain a global data set; According to the linear transformation data set, the global data set is subjected to denoising extraction to obtain an output data set corresponding to the global context shrinkage network.

4. The motor fault detection method according to claim 3, characterized in that: The global context feature extraction is performed on the linear transformation data set to obtain a global data set: Performing feature combination on the linear transformation data set to obtain a combined data set; Performing a first bottleneck transformation on the combined data set to obtain a first bottleneck data set, and performing a second bottleneck transformation on the combined data set to obtain a second bottleneck data set; Performing soft thresholding processing on the second bottleneck data set to obtain a first soft threshold data set; The first bottleneck data set is element-wise multiplied according to the first soft threshold data set to obtain the global data set.

5. The motor fault detection method according to claim 3, characterized in that: The step of performing denoising extraction on the global data set according to the linear transformation data set to obtain an output data set corresponding to the global context shrinkage network includes: According to the global data set, performing soft threshold extraction on the linear transformation data set to obtain a second soft threshold data set; According to the linear transformation data set, data splicing is performed on the second soft threshold data set to obtain the output data set.

6. The motor fault detection method according to claim 1, characterized in that: The step of performing attention fusion on the second branch feature set and the third branch feature set according to the first branch feature set to obtain a fused data set includes: According to the third branch feature set, feature splicing is performed on the second branch feature set to obtain a spliced ​​feature set; According to the concatenated feature set, the first branch feature set is subjected to attention weighted fusion to obtain the fused data set.

7. A motor fault detection system based on a multi-branch neural network, characterized in that: include: An acquisition module, used for acquiring motor signal data; A processing module, used for inputting the motor signal data into a trained multi-branch neural network to obtain a motor fault detection result; Wherein, the multi-branch neural network includes a local branch framework, a global branch framework and a time series branch framework, and the local branch framework, the global branch framework and the time series branch framework have a parallel relationship, and the trained multi-branch neural network is obtained by the following steps: Acquire a multidimensional data set from a motor, and preprocess the multidimensional data set to obtain a motor training set; Inputting the motor training set into the local branch framework to obtain a first branch feature set, wherein the first branch feature set is used to characterize a set of local feature information corresponding to the motor training set; Inputting the motor training set into a global branch framework to obtain a second branch feature set, wherein the second branch feature set is used to characterize a set of global context information corresponding to the motor training set; Inputting the motor training set into the time series branch framework to obtain a third branch feature set, wherein the third branch feature set is used to characterize a set of time series information corresponding to the motor training set; According to the first branch feature set, performing attention fusion on the second branch feature set and the third branch feature set to obtain a fused data set; According to the fused data set, the parameters of the initialized multi-branch neural network are updated to obtain the trained multi-branch neural network.

8. The motor fault detection system according to claim 7, characterized in that: The global branch framework includes a plurality of global context shrinking networks, the global context shrinking networks are in series connection, and the global context shrinking networks are used to perform the following steps: Acquire an input data set, and perform nonlinear transformation on the input data set to obtain a linear transformation data set; Performing global context feature extraction on the linear transformation data set to obtain a global data set; According to the linear transformation data set, the global data set is subjected to denoising extraction to obtain an output data set corresponding to the global context shrinkage network.

9. An electronic device, characterized in that: include: at least one processor; at least one memory for storing at least one program; When the at least one program is executed by the at least one processor, the at least one processor implements the method according to any one of claims 1 to 6.

10. A computer-readable storage medium storing a program executable by a processor, characterized in that: The program executable by the processor is used to implement the method according to any one of claims 1 to 6 when executed by the processor.

Images