Hydraulic plunger pump fault diagnosis method and device under small sample

Abstract

The invention provides a hydraulic plunger pump fault diagnosis method and device under a small sample, and belongs to the technical field of fault diagnosis based on a neural network. The method comprises the steps that vibration signals of the hydraulic plunger pump in all health states are obtained; randomly generating a sample set based on the vibration signal, generating a query set and a support set based on the sample set, training a relation network, and generating a hydraulic plunger pump fault diagnosis model; wherein the hydraulic plunger pump fault diagnosis model is used for completing fault diagnosis of a hydraulic plunger pump based on a vibration signal, and the relation network at least comprises a feature embedding module, a type traversing module and a two-stage relation measurement module in the direction from the input end to the output end; and real-time vibration signals of the hydraulic plunger pump are collected in real time, and faults of the hydraulic plunger pump are recognized based on the real-time vibration signals and the hydraulic plunger pump fault diagnosis model. According to the hydraulic plunger pump fault diagnosis method and device under the small sample, the hydraulic plunger pump fault can be accurately and effectively recognized under the small sample.

Description

Technical Field

[0001] The present application relates to the technical field of neural network-based fault diagnosis, and in particular to a method and device for diagnosing hydraulic plunger pump faults under a small sample. Background Art

[0002] As a high-pressure rotating mechanical equipment, hydraulic piston pump plays a key role in many industrial fields. Its operating status directly affects the stability and reliability of the entire system. Once a failure occurs, it may cause the upstream and downstream equipment to shut down, resulting in huge economic losses. In the actual operation process, the piston pump is prone to various failures, among which the wear of the three major friction pairs (the contact surface between the plunger ball head and the sliding shoe, the contact surface between the sliding shoe and the swash plate, and the contact surface between the plunger and the cylinder hole) is more common, which will cause the performance of the piston pump to decline. In addition, main shaft cracks, bearing wear, plunger cracks and other failures also occur from time to time, which are extremely harmful to the normal operation of the piston pump.

[0003] Currently, the commonly used fault diagnosis methods for hydraulic piston pumps include traditional fault detection methods (such as vibration analysis methods) and machine learning-based methods. Among them, traditional fault detection methods require a large number of samples and have limited diagnostic capabilities for complex faults, while machine learning-based methods also usually require a large number of training samples to achieve better performance. However, the number of fault samples accumulated in actual operation of piston pumps is very small, which makes it difficult to meet the sample size requirements of these models. This makes the model prone to overfitting during the training process, that is, it performs well on the training set, but has poor generalization capabilities for new samples in actual applications and cannot accurately diagnose unknown fault conditions. Summary of the invention

[0004] In view of this, the present application provides a hydraulic plunger pump fault diagnosis method and device under a small sample, which can achieve accurate and effective identification of hydraulic plunger pump faults under a small sample.

[0005] Specifically, the present application is implemented through the following technical solutions:

[0006] The first aspect of the present application provides a method for diagnosing a hydraulic plunger pump fault under a small sample, the method comprising:

[0007] Obtain vibration signals of hydraulic piston pumps under various health conditions;

[0008] Randomly generate a sample set based on the vibration signal, generate a query set and a support set based on the sample set, train a relationship network, and generate a hydraulic plunger pump fault diagnosis model;

[0009] Wherein, the hydraulic piston pump fault diagnosis model is used to complete the fault diagnosis of the hydraulic piston pump based on the vibration signal, and the relationship network includes at least a feature embedding module, a category traversal module and a two-level relationship measurement module in the direction from the input end to the output end. The feature embedding module extracts the feature information of the vibration signal input into the relationship network under each category, and generates the first feature graphs corresponding to the query set and the support set respectively. The category traversal module filters the first feature graph, extracts the second feature graphs corresponding to the query set and the support set respectively, and the second feature graph includes the common feature information within the class and the unique feature information between the classes. The two-level relationship measurement module identifies the image similarity and class similarity between the second feature graph of the query set and the second feature graph of the support set, and identifies the health status of the pump corresponding to the vibration signal input into the relationship network by combining the image similarity and class similarity;

[0010] A real-time vibration signal of the hydraulic piston pump is collected in real time, and a fault of the hydraulic piston pump is identified based on the real-time vibration signal and the hydraulic piston pump fault diagnosis model.

[0011] A second aspect of the present application provides a hydraulic plunger pump fault diagnosis device under a small sample, the device comprising an acquisition module, a generation module and an identification module;

[0012] The acquisition module is used to acquire the vibration signal of the hydraulic plunger pump under each health state;

[0013] The generation module is used to randomly generate a sample set based on the vibration signal, generate a query set and a support set based on the sample set, train a relationship network, and generate a hydraulic plunger pump fault diagnosis model;

[0014] Wherein, the hydraulic piston pump fault diagnosis model is used to complete the fault diagnosis of the hydraulic piston pump based on the vibration signal, and the relationship network includes at least a feature embedding module, a category traversal module and a two-level relationship measurement module in the direction from the input end to the output end. The feature embedding module extracts the feature information of the vibration signal input into the relationship network under each category, and generates the first feature graphs corresponding to the query set and the support set respectively. The category traversal module filters the first feature graph, extracts the second feature graphs corresponding to the query set and the support set respectively, and the second feature graph includes the common feature information within the class and the unique feature information between the classes. The two-level relationship measurement module identifies the image similarity and class similarity between the second feature graph of the query set and the second feature graph of the support set, and identifies the health status of the pump corresponding to the vibration signal input into the relationship network by combining the image similarity and class similarity;

[0015] The identification module is used to collect the real-time vibration signal of the hydraulic piston pump in real time, and identify the fault of the hydraulic piston pump based on the real-time vibration signal and the hydraulic piston pump fault diagnosis model.

[0016] The hydraulic plunger pump fault diagnosis method and device under small samples provided by the present application introduces a category traversal module and a two-level similarity measurement module into the relationship network, so that the model can perform effective fault diagnosis under limited sample data. Specifically, the addition of the category traversal module can filter the feature graph output by the feature embedding module. After observing all fault categories, it can concentrate on extracting features that reflect the commonality within the class and the uniqueness between classes. This means that in the case of small samples, the essential differences between different fault types can be grasped more accurately, thereby providing a more powerful basis for accurate fault diagnosis. The introduction of the two-level similarity measurement module improves the inadequacy of image-level metrics in expressing category distribution. It combines image and image metrics as well as image and class metrics, and comprehensively considers multiple factors to calculate the similarity score, so as to more accurately determine the fault category to which the sample belongs. In practical applications, it can more finely distinguish different fault categories, reduce misjudgments, improve the accuracy of diagnosis, and ensure accurate identification of plunger pump faults. The method provided in this application starts with building a relationship network, clarifies the basic framework of fault diagnosis, including key links such as feature extraction and relationship measurement, and then optimizes each link by adding specific modules. The final fault prediction category considers the entire task as a whole, comprehensively considers the relationship between categories, and improves the classification and recognition capabilities in fault diagnosis. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 A flow chart of a first embodiment of a method for diagnosing a hydraulic plunger pump fault under a small sample provided by the present application;

[0018] Figure 2 A schematic diagram of the axial signal spectrum shown in this application;

[0019] Figure 3 A schematic diagram of a radial signal spectrum shown in this application;

[0020] Figure 4 A schematic diagram of the sample generation, training and testing process of the hydraulic piston pump fault diagnosis model shown in this application;

[0021] Figure 5 A schematic diagram of a conventional relationship network shown in this application;

[0022] Figure 6 A schematic diagram of the improved relationship network after training shown in this application;

[0023] Figure 7 This is a schematic diagram of the structure of the category traversal module shown in this application;

[0024] Figure 8 This is a structural schematic diagram of Example 2 of the hydraulic plunger pump fault diagnosis device under the small sample provided in this application. DETAILED DESCRIPTION

[0025] Here, exemplary embodiments are described in detail, and examples thereof are shown in the accompanying drawings. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The implementations described in the following exemplary embodiments do not represent all implementations consistent with the present application.

[0026] The terms used in this application are only for the purpose of describing specific embodiments and are not intended to limit this application. The singular forms of "a", "said" and "the" used in this application are also intended to include plural forms, unless the context clearly indicates other meanings. It should also be understood that the term "and / or" used in this article refers to and includes any or all possible combinations of one or more associated listed items.

[0027] It should be understood that although the terms first, second, third, etc. may be used in the present application to describe various information, these information should not be limited to these terms. These terms are only used to distinguish the same type of information from each other. For example, without departing from the scope of the present application, the first information may also be referred to as the second information, and similarly, the second information may also be referred to as the first information. Depending on the context, the word "if" as used herein may be interpreted as "at the time of" or "when" or "in response to determining".

[0028] Specific embodiments are given below to introduce the technical solution of the present application in detail.

[0029] Figure 1 This is a flow chart of the first embodiment of the hydraulic plunger pump fault diagnosis method under the small sample provided by this application. Please refer to Figure 1 , the method provided in this embodiment may include:

[0030] S101, obtaining vibration signals of the hydraulic piston pump in various health states.

[0031] Specifically, the vibration signals of the hydraulic piston pump under various healthy states include the axial and radial vibration signals of the piston pump housing, among which the axial and radial vibration signals are a direct reflection of the operating state of the hydraulic piston pump. Under normal operating conditions, the vibration signal has a certain characteristic mode, and its vibration spectrum may show a low and stable amplitude in certain specific frequency bands, reflecting the normal wear and operation state of various friction pairs, bearings and other components. When a fault occurs, the characteristics of the vibration signal will change. For example, when the three major friction pairs of the piston pump (the contact surface between the piston ball head and the sliding shoe, the contact surface between the sliding shoe and the inclined plate, and the contact surface between the piston and the cylinder hole) are worn, or when faults such as main shaft cracks, bearing wear, and piston cracks occur, the mechanical relationship inside the piston pump changes, and this change will be directly reflected in the vibration signal. For example, friction pair wear may cause an increase in vibration amplitude, which is manifested as an amplitude change of a specific frequency component in the spectrum; main shaft cracks may cause abnormal fluctuations in vibration frequency, etc. Figure 2 This is a schematic diagram of the axial signal spectrum shown in this application. Figure 3 For a schematic diagram of the radial signal spectrum shown in this application, please refer to Figure 2 and Figure 3 By acquiring these signals, the most original data basis can be provided for subsequent fault diagnosis. Specifically, for example, the axial and radial vibration signals of the housing surface can be monitored and data samples can be acquired through tests of the plunger pump under normal and fault conditions (plunger pump slipper groove fault, slipper wear fault, bearing outer ring fault).

[0032] S102, randomly generating a sample set based on the vibration signal, generating a query set and a support set based on the sample set, training a relationship network, and generating a hydraulic plunger pump fault diagnosis model.

[0033] Among them, the hydraulic piston pump fault diagnosis model is used to complete the fault diagnosis of the hydraulic piston pump based on the vibration signal. The relationship network includes at least a feature embedding module, a category traversal module and a two-level relationship measurement module in the direction from the input end to the output end. The feature embedding module extracts the feature information of the vibration signal input into the relationship network under each category, and generates the first feature graph corresponding to the query set and the support set respectively. The category traversal module filters the first feature graph and extracts the second feature graph corresponding to the query set and the support set respectively. The second feature graph includes common feature information within the class and unique feature information between classes. The two-level relationship measurement module identifies the image similarity and class similarity between the second feature graph of the query set and the second feature graph of the support set, and comprehensively identifies the health status of the pump corresponding to the vibration signal input into the relationship network by combining the image similarity and class similarity.

[0034] Figure 4 This is a schematic diagram of the sample generation, training and testing process of the hydraulic piston pump fault diagnosis model shown in this application. Please refer to Figure 4Specifically, before randomly generating a sample set based on the vibration signal, it includes: Fourier transforming the axial and radial vibration signals respectively to obtain two corresponding spectra; filtering the two spectra based on the frequency band; and reconstructing the N*N dimensional feature graph by row based on the filtered axial vibration signal spectrum amplitude and radial vibration signal spectrum amplitude. Among them, Fourier transform is a mathematical tool that converts time domain signals into frequency domain signals. In the fault diagnosis of hydraulic piston pumps, time domain vibration signals are often more complex and it is difficult to directly find fault features from them. After obtaining the spectrum through Fourier transform, the signal can be converted from the time dimension to the frequency dimension, clearly showing the amplitude of different frequency components in the signal.

[0035] In addition, different fault types usually show different characteristics in the spectrum. For example, a sudden increase in the amplitude of a specific frequency component may correspond to the resonance of a component, indicating that the component may be faulty. By analyzing the spectrum, characteristic information related to the fault can be more effectively extracted, providing a basis for accurate fault diagnosis.

[0036] It should also be noted that merging the spectral amplitudes of the axial and radial vibration signals into a feature map can integrate the vibration information in the two directions and provide a richer representation of fault features. In a hydraulic piston pump, the occurrence of a fault may show characteristic changes in both axial and radial vibrations. In this way, information from both directions can be comprehensively considered to improve the accuracy of fault diagnosis. In addition, reconstructing it into an N*N dimensional feature map facilitates subsequent input into models such as relational networks for processing.

[0037] Specifically, the axial and radial vibration signals are Fourier transformed respectively to obtain two spectra, and the low-frequency bands of the two spectra are selected, with amplitudes of A=[a1, a2, ..., am] and B=[b1, b2, ..., bm], respectively. A and B are merged into C=[a1, a2, ..., am, b1, b2, ..., bm], and C is reconstructed row by row into a signal D of N*N dimension. If the length of C does not meet the N*N size requirement, 0 is added to the position where the signal D is insufficient. Among them, the low-frequency bands of the two spectra are selected because the low-frequency bands often contain characteristic information related to the main faults of the plunger pump.

[0038] In specific implementation, for example, the sampling rate of the vibration signal is set to 50kHz, the frequency domain of the vibration signal of each second is transformed to obtain the frequency amplitude of 0 to 25kHz, the low-frequency spectrum of the frequency band 1 to 3200Hz is taken, that is, the axial spectrum is taken as A=[a1, a2, ..., a3200], and the radial spectrum is taken as B=[b1, b2, ..., b3200], and the two signals are reconstructed into a feature map of 80*80 dimensions by row, as shown below:

[0039] .

[0040] Specifically, randomly generating a sample set based on the vibration signal includes:

[0041] (1) Generate a sample set based on a random generator and the vibration signal.

[0042] It should be noted that the random generator is a tool or algorithmic mechanism used to expand the number of samples and increase the diversity of samples. Specifically, the random generator plays an important role in the process of generating sample sets. Since there are very few fault samples in the actual operation of hydraulic piston pumps, directly using limited original samples for model training can easily lead to overfitting and poor model generalization ability. By introducing randomness, the random generator can expand the number of samples and increase the diversity of samples to a certain extent. For example, it can randomly transform the original vibration signal, such as randomly changing the amplitude, phase and other parameters of the signal within a certain range, thereby generating new samples that are similar to but different from the original samples.

[0043] (2) In the training task of N state categories and K sample points, select K samples as the support set for each state category, and select the query set from the remaining training samples. The state categories include at least normal state and fault state.

[0044] It should be noted that the training task setting method of N state categories and K sample points is to effectively train the network classifier under limited small sample conditions. The fault types are divided into N state categories, and K sample points are selected as support sets for each category. These support set samples can be regarded as typical representatives of each type of fault. For example, in the fault diagnosis of hydraulic piston pumps, if there are four types of faults (i.e., N=4), namely, normal, slipper groove fault, slipper wear fault, and bearing outer ring fault, 5 samples (i.e., K=5) are selected as support sets for each fault, then these 20 support set samples contain the key feature information of each type of fault. In addition, several of the remaining training samples are selected as query sets. The query set samples are used to compare and learn with the support set samples. By allowing the model to continuously match and learn features between the support set and the query set, the model can better understand the differences between different fault categories, thereby improving the classification ability.

[0045] It should also be noted that when selecting samples, the diversity and representativeness of the samples should be ensured as much as possible. For the support set, different manifestations of various types of faults should be covered, such as different degrees of wear or cracks, so that the model can fully grasp the characteristic range of each type of fault during the learning process. For the query set, its selection should be able to effectively test the model's learning effect on the support set features and prompt the model to continuously optimize the classification boundaries. If the sample selection is unreasonable, for example, the support set is too single or the query set cannot effectively test the model, it may cause the model to learn incomplete features or inaccurate classification capabilities, affecting the final fault diagnosis effect.

[0046] (3) Based on the support set and the query set, a network classifier is trained to learn feature extraction patterns of samples in different state categories.

[0047] It should be noted that the purpose of training the network classifier is to enable the model to extract the key features of various faults and accurately distinguish them through interactive learning between the support set and the query set. Specifically, the support set provides the basic feature pattern of each category, while the query set promotes the model to find clear classification boundaries in the feature space by comparing and matching with the support set.

[0048] It should also be noted that the training process also includes a testing process. Specifically, the setting of the support set and the query set in the testing process is kept consistent with that in the training process to ensure the accuracy and comparability of the evaluation results. In the testing phase, the sample distribution and task form faced by the model are the same as those in the training phase, which can truly reflect the performance of the model in practical applications. For example, if the samples are divided into 4 state categories and 5 sample points during the training process, then this method is also used in the testing process, so that the classification ability of the model can be accurately evaluated when facing new samples with the same structure. Specifically, the plunger pump data set (training set and test set) exemplified in this application is shown in Table 1:

[0049] Table 1 Piston pump data set

[0050]

[0051] Please refer to Table 1 to clearly see the number of samples.

[0052] Figure 5 A schematic diagram of a conventional relationship network shown in this application, Figure 6 The improved relationship network diagram after training shown in this application is shown in Figure 1. Further, based on the sample set, a query set and a support set are generated to train the relationship network, specifically, including:

[0053] (1) Extracting first feature graphs of support set samples and query set samples based on the feature embedding module.

[0054] It should be noted that the feature embedding module includes four sequentially connected convolution blocks, wherein each convolution block includes a two-dimensional convolution layer, a batch normalization layer and a ReLU activation function layer. In addition, the first two convolution blocks also include a maximum pooling layer. Specifically, each convolution block in the feature embedding module includes a two-dimensional convolution layer, which slides on the sample feature map through the convolution kernel to extract local features at different positions. For example, for the spectrum feature map of the vibration signal of the hydraulic piston pump, it can capture the feature patterns corresponding to different frequency bands. The batch normalization layer helps to stabilize the input data distribution of each layer, accelerate the convergence of the model, and improve the training efficiency. The ReLU activation function layer introduces nonlinear factors to enhance the model's ability to express complex feature relationships, because there is often a nonlinear relationship between the fault features and normal features of the hydraulic piston pump. The maximum pooling layer in the first two convolution blocks can reduce the resolution of the feature map, reduce the amount of calculation, retain key feature information, and highlight the trend of changes in the main features. For example, when processing high-frequency vibration signal features, pooling can focus on the main frequency feature changes, remove some detail noise, and make the features more prominent.

[0055] When the support set samples and query set samples are input into the feature embedding module, they are processed layer by layer by these convolution blocks, and finally the corresponding first feature maps are obtained. These first feature maps contain the feature information of the samples in different dimensions, laying the foundation for further processing and analysis. For example, for a normal sample and a fault sample (such as a slipper wear fault sample) of a hydraulic piston pump, the first feature map obtained after processing by the feature embedding module will show different feature patterns in terms of texture, amplitude distribution, etc. These differences will be used to distinguish different fault categories in subsequent steps.

[0056] (2) Extracting second feature maps of support set samples and query set samples based on the category traversal module, and concatenating the extracted second feature maps in the channel dimension.

[0057] The main function of the category traversal module is to filter and optimize the first feature map output by the feature embedding module. In the fault diagnosis of hydraulic piston pumps, the first feature map extracted by the feature embedding module may contain redundant information or information that is not critical to fault classification. Through a specific internal structure and algorithm, the category traversal module can filter out features that reflect the commonality within a class and the uniqueness between classes after observing all fault categories. For example, for different degrees of slipper wear faults, there may be some common vibration feature patterns, and these common features will be highlighted after being processed by the category traversal module; at the same time, the unique difference features between the slipper wear fault and other faults (such as plunger crack fault, bearing outer ring fault, etc.) will also be extracted.

[0058] In addition, since the hydraulic piston pump has very few fault samples accumulated in actual operation, traditional feature extraction methods may not be able to fully utilize the limited samples to learn effective fault features. The introduction of the category traversal module can more effectively mine the relationship between fault categories in the case of small samples, thereby improving the effectiveness of feature extraction. It avoids over-reliance on or misjudgment of individual features when the sample size is limited by comprehensively considering all fault categories, making the extracted features more representative and discriminative.

[0059] Figure 7 For the structural diagram of the type traversal module shown in this application, please refer to Figure 7 , the category traversal module filters the first feature graph and extracts the second feature graphs corresponding to the query set and the support set respectively, including: reconstructing the first feature graphs of the query set and the support set to obtain the inter-class unique feature information of the query set and the support set respectively; calculating the inter-class related feature information of the support set based on the cascade module; averaging and projecting the inter-class related feature information to obtain prompt information; and generating the intra-class common feature information of the query set and the support set respectively based on the prompt information and the reconstructed information.

[0060] In specific operations, the category traversal module first reconstructs the first feature graphs of the support set and the query set to obtain the inter-class unique feature information of the query set and the support set, respectively. Then, the inter-class related feature information of the support set is calculated based on the cascade module, and then the inter-class related feature information is averaged and projected to obtain the prompt information. Finally, based on the prompt information and the reconstructed information, the intra-class common feature information of the query set and the support set is generated respectively, thereby obtaining the second feature graph. The extracted second feature graphs of the support set and the query set are spliced ​​in the channel dimension, so that the network can simultaneously consider the feature information of the support set and the query set after being optimized by the category traversal module. For example, if the support set sample is , the query set sample is The second feature graphs of the extracted support set and query set are and , then concatenated in the channel dimension to obtain .

[0061] (3) The concatenated feature graph is passed through a two-level relationship measurement module to output the similarity score of each support set sample corresponding to the fault category, which is used to characterize the similarity between the query sample under each type of state and the support set sample corresponding to each type of state.

[0062] It should be noted that the two-level relationship measurement module also includes two measurement convolution blocks and two fully connected layers. The measurement convolution blocks are alternately connected to the fully connected layers. Each measurement convolution block contains a convolution layer, a batch normalization layer, a ReLU activation layer, and a maximum pooling layer. Among them, the convolution layer in each convolution block continues to extract features from the spliced ​​feature map, and further explores the deep feature relationship in the feature map. The role of the batch normalization layer, the ReLU activation layer, and the maximum pooling layer is similar to that in the feature embedding module. They work together on the feature map to make its feature expression more effective. The feature map processed by the convolution block enters the fully connected layer. The fully connected layer integrates the local features extracted by the convolution layer to form a global representation of the entire sample feature, so that the similarity between the query sample and each type of sample in the support set can be measured.

[0063] Specifically, the concatenated feature graphs are fed into the two-level relational metric module, which calculates the similarity score between the query sample and each type of sample in the support set. This score represents the similarity between the query sample and each type of fault sample in the support set. If the query sample is very similar to a certain type of fault sample, its similarity score will be close to 1; conversely, if the feature difference is large, the score will be close to 0.

[0064] In a specific implementation, the two-level relationship measurement module identifies the image similarity and class similarity between the second feature graph of the query set and the second feature graph of the support set, including:

[0065] (31) Generate a support set class prototype based on the query set and the second feature graph of the support set.

[0066] It should be noted that the support set prototype is a representative summary of each fault category in the feature space. In the fault diagnosis of hydraulic piston pumps, the support set prototype is generated by processing the second feature map of the support set. The second feature map is optimized by the category traversal module and contains more representative and distinguishing fault feature information. The support set prototype can be understood as an "average" or "typical" feature pattern of each fault category based on these feature information.

[0067] (32) Calculate a similarity measure between images based on the support set class prototype and the second feature map of the query set.

[0068] It should be noted that the image-to-image similarity measure mainly focuses on the similarity between the query set sample and the support set class prototype at the feature map image level. In principle, it measures the similarity between two feature maps in terms of visual features, such as comparing their similarity in terms of texture, amplitude distribution, etc. of the spectral feature map. For hydraulic piston pump fault diagnosis, this similarity reflects the closeness of the query sample and the fault category represented by the support set in terms of vibration signal feature performance. For example, if the second feature map of a query sample is very similar to the support set class prototype of the slipper wear fault in terms of texture and amplitude distribution, then their values ​​in the image-to-image similarity measure will be high. The calculation method of the similarity measure can be based on a variety of distance metrics or similarity functions, such as Euclidean distance, cosine similarity, etc.

[0069] (33) Generate a feature map of the local descriptor connection space based on the second feature map of the query set and the support set.

[0070] It should be noted that generating a feature map of the local descriptor connection space based on the query set and the second feature map of the support set includes:

[0071] (i) constructing a local description subspace of the support set and the query set based on the second feature graphs of the query set and the support set.

[0072] Assume that the second feature graphs of the support set and query set are and , whose size is , can be regarded as a space D composed of d×d c-dimensional local descriptors. Among them, c is the number of feature map channels, and d is the length and width of the feature map. Specifically, the local description subspace D is as follows:

[0073] ;

[0074] in, represents the i-th local descriptor, Indicates the number of local descriptors.

[0075] (ii) Construct the local descriptor connection space based on the dot product result of the support set local descriptor subspace and the query set local descriptor subspace.

[0076] If the local description subspaces of the support set and the query set are , , and the dot product of the two forms the local descriptor connection space L. Among them, the dot product operation associates the local descriptors of the support set and the query set in space, so that the relationship between them can be considered as a whole. This association method can capture the similarity patterns of different samples at the local feature level.

[0077] (iii) extracting a feature map of the local descriptor connection space under the local descriptor connection space.

[0078] The purpose of extracting feature graphs in the local descriptor connection space is to further integrate and refine the information in the connection space so that it can be used more effectively to calculate the similarity measurement of images and classes. Although the connection space already contains the association information of the support set and the query set at the local descriptor level, this information is relatively scattered and primitive. By extracting feature graphs, this information can be reorganized and represented to highlight the key features related to fault classification. In addition, the specific method of extracting feature graphs can be designed according to the specific algorithm and model requirements.

[0079] (34) Calculate a similarity measure between the image and the class based on the feature map of the local descriptor connection space.

[0080] It should be noted that image similarity mainly focuses on the similarity between the query sample and the support set sample at the feature map image level, such as comparing their similarity in terms of texture, amplitude distribution, etc. of the spectral feature map. Class similarity, on the other hand, considers the relationship between the query sample and the overall feature pattern (class level) of each type of fault from a more macroscopic perspective. For example, for hydraulic piston pump faults, the degree of conformity of the query sample's features with the overall feature pattern of the slipper wear fault class will be considered, rather than just the similarity with a single support set sample. The metric network analyzes and calculates the local descriptor connection space L to obtain these two similarity metrics. These similarity metrics are quantitative representations of the relationship between the query sample and the support set sample, and can reflect the degree of possibility that the query sample belongs to different fault categories.

[0081] Furthermore, the health status of the pump corresponding to the vibration signal input into the relationship network can be identified by combining the image similarity and the class similarity. Specifically, the method includes: setting a first weight vector to be trained for the image similarity, and setting a second weight vector to be trained for the class similarity; the sum of the first weight vector and the second weight vector is 1; training the first weight vector and the second weight vector with the mean square error as the objective function; and identifying the health status of the pump corresponding to the vibration signal input into the relationship network based on the trained first weight vector and the second weight vector in combination with the image similarity and the class similarity.

[0082] Set the weight vector a to be trained for the image similarity S1 1 , set the weight vector a to be trained for class similarity S2 2 , where a 1 [i]+a 2[i]=1,i=1,2,...,N (N is the N health states of the pump). The objective function is the mean square error. Two weight vectors are trained to obtain the trained weight vectors. When the online vibration signal is input, the network can output the weighted similarity measurement result and then output the identified pump state.

[0083] (4) Select the fault category corresponding to the support set sample with the highest similarity score as the fault category of the query sample.

[0084] It should be noted that after the two-level relationship measurement module calculates the similarity scores with the query sample for each of the N fault categories in the support set, the fault category with the highest score is selected as the predicted fault category of the query sample. For example, in a task where the support set has four types of faults (such as normal, slipper groove fault, slipper wear fault, and bearing outer ring fault), the two-level relationship measurement module will calculate the similarity scores with the four types of fault samples for the query sample. Assuming that the similarity score with the normal state sample is 0.3, the similarity score with the slipper groove fault sample is 0.7, the similarity score with the slipper wear fault sample is 0.5, and the similarity score with the bearing outer ring fault sample is 0.4, then the network will determine the query sample as a slipper groove fault.

[0085] There may be many different types of faults in hydraulic piston pumps. By comparing the similarity scores between the query sample and various fault samples, all possible fault categories can be comprehensively considered, improving the accuracy and reliability of fault diagnosis, avoiding the misjudgment that may be caused by simple binary classification or single judgment methods, and being able to more carefully identify different fault types, timely discover and handle hydraulic piston pump faults, and ensure the normal operation of the equipment.

[0086] It should also be noted that in the process of building the relationship network, based on the number of support set samples, the number of query set samples, the support set sample labels and the query set sample labels, the mean square error is used as the loss function of the network training. Among them, the mean square error is the average of the square of the difference between the predicted value of the calculation model and the true value. In the relationship network of hydraulic piston pump fault diagnosis, this error reflects the inaccuracy of the model's judgment on the fault category to which the query sample belongs.

[0087] Specifically, the optimization goal of the mean square error loss function (when the query set and support set sample categories are consistent, the similarity score is 1, and when the categories are inconsistent, the score is 0) guides the relational network to learn the characteristic patterns of samples of different fault categories. In the fault diagnosis of hydraulic piston pumps, the network gradually understands the differences and commonalities of features such as the vibration signal spectrum feature map under different fault states (such as piston pump slipper groove fault, slipper wear fault, bearing outer ring fault, etc.) by learning a large number of support set samples and query set samples. In addition, under small sample conditions (hydraulic piston pump fault samples are usually small), the model is prone to overfitting, that is, over-learning the detailed features of the training samples and losing the generalization ability of new samples. The mean square error loss function can help prevent overfitting to a certain extent.

[0088] Specifically, the mean square error shown in this application may be as follows:

[0089] ;

[0090] Among them, m is the number of support set samples, n is the number of query set samples, is the support set sample, is a query set sample, and They are , The corresponding label, Score the relationship.

[0091] S103 . Collecting a real-time vibration signal of the hydraulic piston pump in real time, and identifying a fault of the hydraulic piston pump based on the real-time vibration signal and the hydraulic piston pump fault diagnosis model.

[0092] It should be noted that during the operation of the equipment, faults may occur at any time. By continuously collecting vibration signals, the operating status of the pump can be monitored in real time. It should be noted that the previously trained hydraulic piston pump fault diagnosis model is used to analyze the real-time collected vibration signals. The model is built based on a relational network. After training with a large number of samples (including samples expanded by a random generator), the characteristic patterns of different fault categories on the vibration signal feature map have been learned. When the real-time signal is input into the model, it will first be processed according to the previous process, that is, after Fourier transform, frequency band screening, feature map reconstruction and other steps, the real-time signal is converted into a feature map form that can be processed by the model. Then, the feature map is extracted and similarity calculated through the feature embedding module, the category traversal module and the two-level relationship measurement module. Finally, according to the similarity score calculated by the two-level relationship measurement module, the model selects the fault category with the highest similarity score as the fault prediction category corresponding to the real-time collected signal. If the model determines that the real-time signal is most similar to the characteristic pattern of a certain fault category (such as the slipper wear fault), that is, the similarity score is the highest, then it is determined that the hydraulic piston pump may currently have a slipper wear fault.

[0093] The method provided in this embodiment starts from the data acquisition stage, by reasonably collecting the axial and radial vibration signals of the shell under normal and fault conditions, providing basic data for subsequent analysis. In the sample preparation stage, the signal is subjected to Fourier transformation, frequency band selection and feature map reconstruction, and the data is processed into a format suitable for small sample learning (such as N*N dimensional feature map). Then, when dividing the training and test samples, the N-category K-sample method is adopted to make full use of limited samples for model training and evaluation. The category traversal module added after the feature embedding module filters the output feature map. After observing all fault categories, it can concentrate on extracting features that reflect the commonality within the class and the uniqueness between classes. In the fault diagnosis of hydraulic piston pumps, for different types of faults such as slipper groove faults, slipper wear faults, and bearing outer ring faults, this module can highlight their respective unique vibration signal feature patterns and reduce the interference of irrelevant or redundant features. The two-level similarity measurement module is introduced after the relationship measurement module, which improves the inadequacy of image-level measurement in expressing category distribution. By constructing local description subspaces and connection spaces, and combining image and image measurement as well as image and class measurement, the similarity score is calculated by comprehensively considering multiple factors. When judging the fault category of the hydraulic piston pump, not only the similarity of the samples at the image feature level is considered, but also the relationship with the overall characteristic pattern of each type of fault is considered, so as to more comprehensively and accurately determine the fault category and improve the diagnostic accuracy. In addition, the method provided in this application starts with the construction of a relationship network, clarifies the basic framework of fault diagnosis, including key links such as feature extraction and relationship measurement, and then optimizes each link by adding specific modules. Finally, the final fault prediction category is obtained based on the improved relationship network, and the entire task is considered as a whole, and the relationship between categories is comprehensively considered, which improves the classification and recognition ability of the entire task.

[0094] This application also provides an analysis of the test results. Specifically, the test samples are divided into a support set and a query set in the same way as the training set. The method of this application (RN) is compared with other small sample classic models (Single Network SN, Prototype Network PN, Relational Network RN). The comparison results are shown in Table 2:

[0095] Table 2 Comparison results

[0096] Model 4way 5shot / % SN 85.98 PN 83.04 RN 89.41 RNThis article 92.94

[0097] Please refer to Table 2. Compared with other small sample network models, the method of this application has an improvement of about 3.5% to 7% in accuracy.

[0098] In addition, an ablation test analysis was conducted to confirm the effectiveness of the results by adding a category traversal module and a two-level relationship measurement module to the model. Specifically, the comparison results of the ablation test are shown in Table 3:

[0099] Table 3 Ablation test

[0100] method first / % first / % first / % first / % first / % Average accuracy / % RN1 90.98 87.35 89.90 91.76 87.06 89.41±2.12 RN2 93.97 90.29 90.98 92.99 91.08 91.80±1.55 RN3 87.30 91.62 90.54 90.78 89.02 89.85±1.71 RNThis article 94.26 90.93 94.66 92.16 92.70 92.94±1.53

[0101] Please refer to Table 3, RN1 represents the relationship network, RN2 represents the relationship network with the category traversal module added, and RN3 represents the network with the two-level relationship measurement module introduced. Taking the average of multiple test results, it can be seen that the introduction of the category traversal module and the two-level relationship measurement module alone improves the accuracy by about 2.39% and 0.44%, respectively.

[0102] Corresponding to the aforementioned embodiment of a method for diagnosing a hydraulic plunger pump fault under a small sample, the present application also provides an embodiment of a device for diagnosing a hydraulic plunger pump fault under a small sample.

[0103] Figure 8 This is a structural diagram of the second embodiment of the hydraulic plunger pump fault diagnosis device under the small sample provided by this application. Please refer to Figure 8 , the device provided in this embodiment includes an acquisition module 810, a generation module 820 and an identification module 830;

[0104] The acquisition module 810 is used to acquire the vibration signal of the hydraulic plunger pump in each health state;

[0105] The generating module 820 is used to randomly generate a sample set based on the vibration signal, generate a query set and a support set based on the sample set, train a relationship network, and generate a hydraulic plunger pump fault diagnosis model;

[0106] Wherein, the hydraulic piston pump fault diagnosis model is used to complete the fault diagnosis of the hydraulic piston pump based on the vibration signal, and the relationship network includes at least a feature embedding module, a category traversal module and a two-level relationship measurement module in the direction from the input end to the output end. The feature embedding module extracts the feature information of the vibration signal input into the relationship network under each category, and generates the first feature graphs corresponding to the query set and the support set respectively. The category traversal module filters the first feature graph, extracts the second feature graphs corresponding to the query set and the support set respectively, and the second feature graph includes the common feature information within the class and the unique feature information between the classes. The two-level relationship measurement module identifies the image similarity and class similarity between the second feature graph of the query set and the second feature graph of the support set, and identifies the health status of the pump corresponding to the vibration signal input into the relationship network by combining the image similarity and class similarity;

[0107] The identification module 830 is used to collect the real-time vibration signal of the hydraulic piston pump in real time, and identify the fault of the hydraulic piston pump based on the real-time vibration signal and the hydraulic piston pump fault diagnosis model.

[0108] The device of this embodiment can be used to perform Figure 1The steps, specific implementation principles and implementation processes of the method embodiment shown are similar and will not be repeated here.

[0109] The implementation process of the functions and effects of each unit in the above-mentioned device is specifically described in the implementation process of the corresponding steps in the above-mentioned method, and will not be repeated here.

[0110] For the device embodiment, since it basically corresponds to the method embodiment, the relevant parts can refer to the partial description of the method embodiment. The device embodiment described above is only schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the present application scheme. A person of ordinary skill in the art can understand and implement it without paying any creative work.

[0111] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application shall be included in the scope of protection of the present application.

Claims

1. A method for diagnosing hydraulic piston pump faults under a small sample, characterized in that: The method comprises: Obtain vibration signals of hydraulic piston pumps under various health conditions; Randomly generate a sample set based on the vibration signal, generate a query set and a support set based on the sample set, train a relationship network, and generate a hydraulic plunger pump fault diagnosis model; Wherein, the hydraulic piston pump fault diagnosis model is used to complete the fault diagnosis of the hydraulic piston pump based on the vibration signal, and the relationship network includes at least a feature embedding module, a category traversal module and a two-level relationship measurement module in the direction from the input end to the output end. The feature embedding module extracts the feature information of the vibration signal input into the relationship network under each category, and generates the first feature graphs corresponding to the query set and the support set respectively. The category traversal module filters the first feature graph, extracts the second feature graphs corresponding to the query set and the support set respectively, and the second feature graph includes the common feature information within the class and the unique feature information between the classes. The two-level relationship measurement module identifies the image similarity and class similarity between the second feature graph of the query set and the second feature graph of the support set, and identifies the health status of the pump corresponding to the vibration signal input into the relationship network by combining the image similarity and class similarity; A real-time vibration signal of the hydraulic piston pump is collected in real time, and a fault of the hydraulic piston pump is identified based on the real-time vibration signal and the hydraulic piston pump fault diagnosis model.

2. The method according to claim 1, characterized in that: The category traversal module filters the first feature graph and extracts the second feature graphs corresponding to the query set and the support set, respectively, including: Reconstructing the first feature graphs of the query set and the support set to obtain inter-class unique feature information of the query set and the support set respectively; Calculate inter-class correlation feature information of the support set based on the cascade module; Average and project the inter-class related feature information to obtain prompt information; Based on the prompt information and the reconstructed information, the intra-class common feature information of the query set and the support set is generated respectively.

3. The method according to claim 1, characterized in that The two-level relationship measurement module identifies the image similarity and class similarity between the second feature graph of the query set and the second feature graph of the support set, including: generating a support set class prototype based on the query set and the second feature graph of the support set; Calculating a similarity measure between images based on the support set class prototype and the second feature map of the query set; Generate a feature map of a local descriptor connection space based on the query set and the second feature map of the support set; A similarity measure between the image and the class is calculated based on the feature map of the local descriptor connection space.

4. The method according to claim 3, characterized in that The generating a feature graph of a local descriptor connection space based on the query set and the second feature graph of the support set comprises: Constructing a local description subspace of the support set and the query set based on the second feature graph of the query set and the support set; Construct the local descriptor connection space according to the dot product result of the local descriptor subspace of the support set and the local descriptor subspace of the query set; A feature map of the local descriptor connection space is extracted under the local descriptor connection space.

5. The method according to claim 1, characterized in that Before randomly generating a sample set based on the vibration signal, the method includes: Perform Fourier transform on the axial and radial vibration signals respectively to obtain two corresponding frequency spectra; filtering the two spectra based on frequency bins; Based on the filtered axial vibration signal spectrum amplitude and radial vibration signal spectrum amplitude, the spectrum amplitude is reconstructed row by row into an N*N dimensional feature map.

6. The method according to claim 1, characterized in that Randomly generating a sample set based on the vibration signal, comprising: generating a sample set based on a random generator and the vibration signal; In the training task of N state categories and K sample points, K samples are selected as the support set for each state category, and the query set is selected from the remaining training samples, and the state categories include at least normal state and fault state; Based on the support set and the query set, a network classifier is trained to learn feature extraction patterns of samples in different state categories.

7. The method according to claim 1, characterized in that The generating of a query set and a support set based on the sample set and training a relational network includes: Extracting a first feature graph of a support set sample and a query set sample based on the feature embedding module; Extracting second feature maps of support set samples and query set samples based on the category traversal module, and splicing the extracted second feature maps in the channel dimension; The spliced ​​feature graph is passed through a two-level relationship measurement module to output the similarity score of each support set sample corresponding to the fault category, which is used to characterize the similarity between the query sample under each state and the support set sample corresponding to each state; The fault category corresponding to the support set sample with the highest similarity score is selected as the fault category of the query sample.

8. The method according to claim 1, characterized in that The feature embedding module includes four sequentially connected convolution blocks, wherein each convolution block includes a two-dimensional convolution layer, a batch normalization layer and a ReLU activation function layer; The two-level relationship measurement module includes two measurement convolution blocks and two fully connected layers. The measurement convolution blocks are alternately connected to the fully connected layers. Each measurement convolution block includes a convolution layer, a batch normalization layer, a ReLU activation layer and a maximum pooling layer.

9. The method according to claim 1, characterized in that: Combining the image similarity and the class similarity to identify the health status of the pump corresponding to the vibration signal input into the relationship network, includes: Setting a first weight vector to be trained for the image similarity and setting a second weight vector to be trained for the class similarity; the sum of the first weight vector and the second weight vector is 1; Training the first weight vector and the second weight vector using mean square error as an objective function; Based on the trained first weight vector and the second weight vector combined with the image similarity and the class similarity, the health status of the pump corresponding to the vibration signal input into the relationship network is identified.

10. A hydraulic plunger pump fault diagnosis device under small sample conditions, characterized in that: The device comprises an acquisition module, a generation module and an identification module; The acquisition module is used to acquire the vibration signal of the hydraulic plunger pump under each health state; The generation module is used to randomly generate a sample set based on the vibration signal, generate a query set and a support set based on the sample set, train a relationship network, and generate a hydraulic plunger pump fault diagnosis model; Wherein, the hydraulic piston pump fault diagnosis model is used to complete the fault diagnosis of the hydraulic piston pump based on the vibration signal, and the relationship network includes at least a feature embedding module, a category traversal module and a two-level relationship measurement module in the direction from the input end to the output end. The feature embedding module extracts the feature information of the vibration signal input into the relationship network under each category, and generates the first feature graphs corresponding to the query set and the support set respectively. The category traversal module filters the first feature graph, extracts the second feature graphs corresponding to the query set and the support set respectively, and the second feature graph includes the common feature information within the class and the unique feature information between the classes. The two-level relationship measurement module identifies the image similarity and class similarity between the second feature graph of the query set and the second feature graph of the support set, and identifies the health status of the pump corresponding to the vibration signal input into the relationship network by combining the image similarity and class similarity; The identification module is used to collect the real-time vibration signal of the hydraulic piston pump in real time, and identify the fault of the hydraulic piston pump based on the real-time vibration signal and the hydraulic piston pump fault diagnosis model.

Images