Motor vibration signal reconstruction method, device, storage medium and electronic device

Abstract

The present specification discloses a motor vibration signal reconstruction method, device, storage medium and electronic device. The motor vibration signal reconstruction method comprises: obtaining each basic vibration signal of the target motor, inputting each basic vibration signal into a preset reconstruction model, so that the reconstruction model predicts the vibration signal of the signal collection point where the sensor is not installed on the target motor according to each basic vibration signal, as the predicted vibration signal, and reconstructing the vibration signal of the target motor according to each basic vibration signal and the predicted vibration signal, and performing a vibration test on the target motor based on the reconstructed vibration signal of the target motor, so as to reduce the cost of the vibration test.

Description

Technical Field

[0001] The present invention relates to the field of computer vision technology, and in particular to a method, device, storage medium and electronic device for reconstructing a motor vibration signal. Background Art

[0002] The vibration test of high-power electric drive system plays an extremely important role in product development and product quality monitoring. Through the vibration test of high-power electric drive system, it can be tested whether the electric drive system can resist the strong vibration caused by high-speed rotation, load changes and other factors under high-power operation state, and then the structural strength and durability of the electric drive system can be determined, and possible faults in the electric drive system can be eliminated.

[0003] Generally speaking, vibration testing of high-power electric drive systems requires experienced professional testers to use a large number of high-quality sensor equipment and high-quality testing equipment to test the electric drive system, which makes the cost of vibration testing of high-power electric drive systems high.

[0004] Therefore, how to reduce the cost of vibration testing of high-power electric drive systems is an urgent problem to be solved. Summary of the invention

[0005] This specification provides a motor vibration signal reconstruction method, device, storage medium and electronic device to partially solve the above-mentioned problems existing in the prior art.

[0006] This manual adopts the following technical solutions:

[0007] This specification provides a method for reconstructing a motor vibration signal, including:

[0008] Acquire each basic vibration signal of the target motor, wherein each basic vibration signal is collected by a sensor installed at at least part of a signal collection location on the target motor, and the basic vibration signal is data used to characterize the mechanical vibration generated by the target motor under different test conditions;

[0009] Inputting each basic vibration signal into a preset reconstruction model, so that the reconstruction model determines node features of at least part of the nodes included in a pre-constructed node relationship graph according to each basic vibration signal as each basic node feature, and determines the association relationship between each signal collection point according to the edge features between each node included in the node relationship graph, and predicts the vibration signal of the signal collection point where the sensor is not installed on the target motor according to the association relationship between each basic node feature and each signal collection point, as a predicted vibration signal;

[0010] A task of reconstructing the vibration signal of the target motor is performed according to the basic vibration signals and the predicted vibration signal.

[0011] Optionally, node features of at least some nodes included in a pre-constructed node relationship diagram are determined according to the basic vibration signals as basic node features, specifically including:

[0012] For each basic vibration signal, determine a node corresponding to a sensor used to collect the basic vibration signal in a pre-constructed node relationship graph as the node corresponding to the basic vibration signal;

[0013] Feature extraction is performed on the basic vibration signal to determine a vibration signal feature corresponding to the basic vibration signal as a basic node feature of the node corresponding to the basic vibration signal.

[0014] Optionally, construct a node relationship graph, specifically including:

[0015] For each signal collection point on the target motor, the signal collection point is used as a node in a node relationship graph, and the association relationship between the signal collection point and each other signal collection point is used as an edge in the node relationship graph to construct the node relationship graph.

[0016] Optionally, predicting the vibration signal of the signal collection point on the target motor where no sensor is installed according to the correlation between the basic node features and the signal collection points specifically includes:

[0017] For each signal collection location on the target motor where no sensor is installed, determine the node corresponding to the signal collection location in the node relationship diagram as the target node, and determine the node corresponding to the signal collection location where the sensor for collecting each basic vibration signal is located in the node relationship diagram as each basic node;

[0018] The basic node features of each basic node and the edge features used to characterize the association relationship between each signal collection point are aggregated to the target node to obtain the predicted node features corresponding to the target node, and based on the predicted node features, the vibration signal of the signal collection point on the target motor where no sensor is installed is predicted.

[0019] Optionally, the basic node features of each basic node and the edge features used to characterize the association relationship between each signal collection location are aggregated to the target node to obtain the predicted node features corresponding to the target node, specifically including:

[0020] Perform multiple rounds of aggregation on each node included in the node relationship graph to aggregate the basic node features of each basic node and the edge features used to characterize the association relationship between each signal collection point to the target node; wherein,

[0021] For each round of aggregation, the initial node features of each node included in the node relationship graph in the round of aggregation are determined, and for each node included in the node relationship graph, the initial node features of the node, the initial node features of other nodes, and the edge features between the node and each other node are aggregated to obtain the corresponding aggregated node features of the node in the round of aggregation, and the corresponding aggregated node features of the node in the round of aggregation are used as the initial node features of the node in the next round of aggregation, wherein the initial node features are obtained by iterating the basic node features to the previous round of aggregation;

[0022] When it is determined that the preset termination condition is met, the aggregated node feature corresponding to the target node in this round of aggregation is used as the predicted node feature corresponding to the target node.

[0023] Optionally, for each node included in the node relationship graph, the initial node feature of the node, the initial node features of other nodes, and the edge features between the node and each other node are aggregated to obtain the corresponding aggregated node feature of the node in this round of aggregation, specifically including:

[0024] For each node included in the node relationship graph, the node weight of each other node with respect to the node is determined, and based on the node weight of each other node with respect to the node, the initial node features of each other node and the edge features between the node and each other node are fused with the initial node features of the node to obtain the corresponding aggregated node features of the node in this round of aggregation.

[0025] Optionally, the basic vibration signal includes: an acceleration signal of the target motor vibration, and an inertia signal of the target motor vibration, and the basic node feature includes: a first basic node feature determined according to the acceleration signal of the target motor vibration and a second basic node feature determined according to the inertia signal of the target motor vibration;

[0026] Aggregating the basic node features of each basic node and the edge features used to characterize the association relationship between each signal collection point to the target node to obtain the predicted node features corresponding to the target node, specifically including:

[0027] Aggregating the first basic node features of each basic node and the edge features used to characterize the association relationship between each signal collection location to the target node to obtain a first sub-prediction node feature corresponding to the target node, and aggregating the second basic node features of each basic node and the edge features used to characterize the association relationship between each signal collection location to the target node to obtain a second sub-prediction node feature corresponding to the target node;

[0028] The first sub-prediction node feature and the second sub-prediction node feature are input into a preset point convolutional network layer, so that the point convolutional network layer fuses the first sub-prediction node feature and the second sub-prediction node feature to obtain the corresponding prediction node feature of the target node.

[0029] This specification provides a motor vibration signal reconstruction device, including:

[0030] an acquisition module, configured to acquire each basic vibration signal of the target motor, wherein each basic vibration signal is acquired by a sensor installed at at least a portion of a signal acquisition location on the target motor, and the basic vibration signal is data for characterizing the mechanical vibration generated by the target motor under different test conditions;

[0031] A prediction module, configured to input each basic vibration signal into a preset reconstruction model, so that the reconstruction model determines node features of at least part of the nodes included in a pre-constructed node relationship graph according to each basic vibration signal as each basic node feature, and determines the association relationship between each signal collection location according to the edge features between each node included in the node relationship graph, and predicts the vibration signal of the signal collection location on which the sensor is not installed on the target motor according to the association relationship between each basic node feature and each signal collection location as a predicted vibration signal;

[0032] An execution module is used to perform a task of reconstructing the vibration signal of the target motor according to the basic vibration signals and the predicted vibration signal.

[0033] At least one of the above technical solutions adopted in this specification can achieve the following beneficial effects:

[0034] In the motor vibration signal reconstruction method provided in the present specification, each basic vibration signal of the target motor is obtained, wherein each basic vibration signal is collected by a sensor installed at at least part of the signal collection point on the target motor, and the basic vibration signal is data used to characterize the mechanical vibration generated by the target motor under different test conditions. Each basic vibration signal is input into a preset reconstruction model, so that the reconstruction model determines the node features of at least part of the nodes contained in a pre-constructed node relationship graph according to each basic vibration signal, as each basic node feature, and determines the association relationship between each signal collection point according to the edge features between each node contained in the node relationship graph, and predicts the vibration signal of the signal collection point on the target motor where no sensor is installed according to each basic node feature as a predicted vibration signal, and performs the reconstruction task of the vibration signal of the target motor according to each basic vibration signal and the predicted vibration signal.

[0035] It can be seen from the above method that the graph neural network can be used to predict the vibration signals at other signal collection points based on the vibration signals collected by the collectors set at some signal collection points, and then the target motor can be tested based on the collected vibration signals and the predicted vibration signals, thereby reducing the number of sensors required to test the target motor, thereby effectively reducing the testing cost. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] The drawings described herein are used to provide a further understanding of this specification and constitute a part of this specification. The illustrative embodiments and descriptions of this specification are used to explain this specification and do not constitute an improper limitation on this specification. In the drawings:

[0037] Figure 1 A schematic diagram of a flow chart of a motor vibration signal reconstruction method provided in this specification;

[0038] Figure 2 It is a schematic diagram of each signal collection point provided in this manual;

[0039] Figure 3 A schematic diagram of the sensor distribution strategy provided in this specification;

[0040] Figure 4 A schematic diagram of a motor vibration signal reconstruction device provided in this specification;

[0041] Figure 5 A method corresponding to the Figure 1 Schematic diagram of electronic equipment. DETAILED DESCRIPTION

[0042] In order to make the purpose, technical solutions and advantages of this specification more clear, the technical solutions of this specification will be clearly and completely described below in combination with the specific embodiments of this specification and the corresponding drawings. Obviously, the described embodiments are only part of the embodiments of this specification, not all of them. Based on the embodiments in this specification, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this specification.

[0043] The technical solutions provided by the embodiments of this specification are described in detail below in conjunction with the accompanying drawings.

[0044] Figure 1 A flow chart of a motor vibration signal reconstruction method provided in this specification includes the following steps:

[0045] S101: Acquire basic vibration signals of a target motor, wherein the basic vibration signals are collected by sensors installed at at least part of a signal collection location on the target motor, and the basic vibration signals are data for characterizing mechanical vibrations generated by the target motor under different test conditions.

[0046] In this specification, the user can select at least some of the signal collection points from the various signal collection points set on the target motor as target signal collection points, and then install sensors at the target signal collection points to collect vibration signals of the target motor under different test conditions as various basic vibration signals.

[0047] The vibration signals under the above-mentioned different test conditions can be determined according to actual needs, for example: when the target motor is controlled to operate according to the specified control parameters, the vibration signal of the target motor. Another example: when the target motor is not in operation (that is, the target motor is in a stationary state), a force is applied to a hammer excitation point set on the target motor (which can be set on the tooth surface of the target motor stator, such as: selecting ten evenly distributed points on the tooth surface of the target motor stator as hammer excitation points), and the vibration signal of the target motor under the force is collected.

[0048] In the above content, each basic vibration signal can be data used to characterize the mechanical vibration generated by the target motor under different test conditions, including: an acceleration signal of the target motor vibration, and an inertia signal of the target motor vibration, wherein the above-mentioned acceleration signal of the target motor vibration can be directly determined by the acceleration value of the target motor vibration collected by the sensor, and the above-mentioned inertia signal of the target motor vibration can be calculated based on the acceleration value and displacement value of the target motor vibration collected by the sensor and the magnitude of the force applied to the hammer excitation point set on the target motor when the target motor is in a stationary state.

[0049] The above-mentioned signal collection points can be set on the outer surface of the frame of the target motor stator, and the number of each signal collection point can be set according to actual needs. Figure 2 shown.

[0050] Figure 2 This is a schematic diagram of each signal collection point provided in this manual.

[0051] Combination Figure 2 It can be seen that 20 signal collection points can be evenly arranged on the outer surface of the frame of the target motor stator. In other words, each signal collection point can divide the circumference corresponding to the outer surface of the frame of the target motor stator into m equal parts, where m is the number of each signal collection point.

[0052] Furthermore, the business platform can obtain the basic vibration signals of the target motor, and predict the vibration signals at other signal collection locations other than the target signal collection location (i.e., signal collection locations where sensors are not installed) set on the target motor based on the basic vibration signals. In other words, the business platform can predict the vibration signals of the target motor that can be collected by the sensors at signal collection locations other than the target signal collection location if sensors are installed at the signal collection locations other than the target signal collection location based on the basic vibration signals. Then, the business platform can reconstruct the complete vibration signal of the target motor based on the basic vibration signals and the predicted vibration signals as the reconstructed vibration signal of the target motor. Then, the target motor can be tested and analyzed based on the reconstructed vibration signal of the target motor to determine the test result of the target motor.

[0053] The above-mentioned test analysis can be based on the reconstructed vibration signal to analyze the vibration characteristics of the target motor under actual operating conditions, including vibration amplitude, frequency response, structural resonance, vibration transmission path, dynamic load and fatigue durability, etc., and based on the reconstructed vibration signal, analyze the structural characteristics, load distribution, component assembly clearance, stiffness and static balance of the target motor when it is not in operation.

[0054] In this specification, the execution entity used to implement the motor vibration signal reconstruction method can refer to a designated device such as a server set on a business platform, or it can refer to terminal devices such as mobile phones, laptops, desktop computers, etc. For the sake of ease of description, the motor vibration signal reconstruction method provided in this specification is explained below using the server as an example of the execution entity.

[0055] S102: Input each basic vibration signal into a preset reconstruction model, so that the reconstruction model determines the node features of at least part of the nodes contained in the pre-constructed node relationship diagram according to each basic vibration signal, as each basic node feature, and determines the correlation relationship between each signal collection point according to the edge features between each node contained in the node relationship diagram, and predicts the vibration signal of the signal collection point where the sensor is not installed on the target motor according to the correlation relationship between each basic node feature and each signal collection point, as a predicted vibration signal.

[0056] In the present specification, after obtaining each basic vibration signal of the target motor, the server can input each basic vibration signal into a preset reconstruction model, so that the reconstruction model can determine, for each basic vibration signal, the node corresponding to the sensor used to collect the basic vibration signal in the pre-constructed node relationship diagram as the node corresponding to the basic vibration signal, and then perform feature extraction on the basic vibration signal to determine the vibration signal feature corresponding to the basic vibration signal as the basic node feature of the node corresponding to the basic vibration signal, and according to the edge features between the nodes contained in the node relationship diagram, the association relationship between the signal collection points can be determined, and then according to the association relationship between the basic node features and the signal collection points, the vibration signal of the signal collection point where the sensor is not installed on the target motor can be predicted as a predicted vibration signal.

[0057] Specifically, the server can determine, for each signal collection point on the target motor where no sensor is installed, the corresponding node in the node relationship graph of the signal collection point as the target node, and determine the corresponding node in the node relationship graph of the signal collection point where the sensor that collects each basic vibration signal is located as each basic node. Then, the basic node features of each basic node and the edge features used to characterize the association relationship between each signal collection point can be aggregated to the target node to obtain the predicted node features corresponding to the target node, and based on the predicted node features, the vibration signal of the signal collection point on the target motor where no sensor is installed is predicted.

[0058] It is worth noting that the above-mentioned basic vibration signals are time domain signals (i.e., signals that change with time), and in the time domain, the signals may appear as complex waveforms, and their internal structures and patterns are not easy to identify, resulting in low accuracy of the predicted vibration signals. Therefore, the server can also perform signal preprocessing (i.e., perform Fourier transform) on the collected basic vibration signals before inputting them into the preset reconstruction model, so as to convert the basic vibration signals into frequency domain signals (i.e., the signal form used to characterize the relative intensity or energy distribution of each frequency component of the signal) to obtain the converted basic vibration signals, and then the converted basic vibration signals can be input into the preset reconstruction model.

[0059] It should be noted that the above-mentioned node relationship graph can be constructed by the server before training the reconstruction model, wherein the method for the server to construct the above-mentioned node relationship graph can be, for each signal collection point on the target motor, the signal collection point is used as a node in the node relationship graph, and the association relationship between the signal collection point and each other signal collection point is used as an edge in the node relationship graph (it can be understood that there is an edge between any two nodes, and this edge is used to characterize the association relationship between the signal collection points corresponding to the nodes connected at both ends of the edge) to construct the node relationship graph.

[0060] In actual application scenarios, for each signal collection point, the vibration signal at the signal collection point is not only directly affected by the exciting force, but also affected by the exciting force transmitted from the signal collection point adjacent to the signal collection point, as well as the exciting force transmitted from other signal collection points to the signal collection points adjacent to the signal collection point, and then transmitted from the signal collection point adjacent to the signal collection point to the signal collection point. For ease of understanding, the factors affecting the vibration signal of each signal collection point are explained below using inertia as an example.

[0061] Specifically, for each signal collection point, the factors that affect the inertial signal of the target motor vibration collected at the signal collection point include, in addition to the exciting force of the self-inherited inertia at the signal collection point, the exciting force transmitted by the mutual inertia. That is to say, the inertial signal of the target motor vibration collected at the signal collection point is the sum of the self-inherited inertia at the signal collection point and the transmitted mutual inertia.

[0062] Based on this, the server can also perform multiple rounds of aggregation on each node included in the node relationship graph to aggregate the basic node features of each basic node and the edge features used to characterize the association relationship between each signal collection point to the target node.

[0063] Among them, the server can determine the initial node features of each node included in the node relationship graph in each round of aggregation, and for each node included in the node relationship graph, aggregate the initial node features of the node, the initial node features of other nodes, and the edge features between the node and each other node to obtain the corresponding aggregated node features of the node in this round of aggregation, and use the corresponding aggregated node features of the node in this round of aggregation as the initial node features of the node in the next round of aggregation. The initial node features here are obtained by iterating each basic node feature as the initial node feature in the first round of aggregation to the previous round of aggregation.

[0064] Furthermore, when it is determined that the preset termination condition is met, the server can use the post-aggregation node features corresponding to the target node in this round of aggregation as the predicted node features corresponding to the target node, wherein the above-mentioned preset termination condition can be determined according to actual needs, for example: when it is determined that the number of aggregation rounds reaches the preset round number threshold, it can be considered that the termination condition is met. For example: when it is determined that the post-aggregation node features corresponding to the target node in this round of aggregation are higher than the preset similarity threshold between the post-aggregation node features corresponding to the target node in the previous round of aggregation, it can be considered that the termination condition is met.

[0065] It can be seen from the above content that the server can determine the different levels of influence relationships between each basic vibration signal and each predicted vibration signal through multiple rounds of aggregation, thereby improving the accuracy of the determined predicted vibration signal.

[0066] It should be noted that the above-mentioned basic vibration signals can be saved in the reconstruction model in the form of a feature matrix, wherein each row of the feature matrix is ​​a feature vector for characterizing the node features corresponding to the vibration signal of each signal collection point, and each column is the dimension value of each dimension contained in each feature vector. The feature vectors corresponding to the signal collection points other than the signal collection points corresponding to the basic node features in the above-mentioned feature matrix can be empty vectors filled with 0 or default vectors filled with default values. The above-mentioned node relationship graph can be saved in the reconstruction model in the form of an adjacency matrix, so that the reconstruction model can aggregate the basic node features of each basic node and the edge features used to characterize the association relationship between the signal collection points to the nodes corresponding to the signal collection points other than the signal collection points corresponding to the basic node features through the matrix calculation between the above-mentioned feature matrix and the adjacency matrix, to obtain the predicted node features, and replace the above-mentioned empty vectors or default vectors with the predicted node features.

[0067] In actual application scenarios, the server can also reconstruct the model to make targeted predictions on the acceleration signal and inertia signal at the signal collection point on the target motor where no sensor is installed, so as to improve the accuracy of the predicted vibration signal at the signal collection point on the target motor where no sensor is installed.

[0068] Specifically, the server can input the acceleration signal of the target motor vibration and the inertia signal of the target motor vibration into the reconstruction model, so that the reconstruction model can determine, for each acceleration signal, the node corresponding to the sensor corresponding to the acceleration signal in the pre-constructed first node relationship diagram as the node corresponding to the acceleration signal, and then perform feature extraction on the acceleration signal to determine the vibration signal feature corresponding to the acceleration signal as the first basic node feature of the node corresponding to the acceleration signal.

[0069] Similarly, the reconstruction model can determine, for each inertial signal, the node corresponding to the sensor corresponding to the inertial signal in the pre-constructed second node relationship graph as the node corresponding to the inertial signal, and then perform feature extraction on the inertial signal to determine the vibration signal feature corresponding to the inertial signal as the second basic node feature of the node corresponding to the inertial signal, wherein the above-mentioned first node relationship graph and second node relationship graph are heterogeneous graphs.

[0070] Furthermore, the reconstruction model can aggregate the first basic node features of each basic node and the edge features used to characterize the association relationship between each signal collection point to the target node to obtain the first sub-prediction node features corresponding to the target node, and aggregate the second basic node features of each basic node and the edge features used to characterize the association relationship between each signal collection point to the target node to obtain the second sub-prediction node features corresponding to the target node, and then the first sub-prediction node features and the second sub-prediction node features can be input into a preset point convolutional network layer so that the point convolutional network layer fuses the first sub-prediction node features and the second sub-prediction node features to obtain the corresponding prediction node features of the target node.

[0071] In the above content, the point convolution network layer can be a network layer included in the reconstruction model, or it can be a separate convolution model. The method for the point convolution network layer to fuse the first sub-prediction node feature and the second sub-prediction node feature can be to use the feature matrix corresponding to the first node relationship graph and the feature matrix corresponding to the second node relationship graph as two feature channels through the point convolution network layer, and then perform point convolution on the feature matrix corresponding to the first node relationship graph and the feature matrix corresponding to the second node relationship graph, so as to fuse the first basic node feature or the first sub-prediction node feature of each node in the first node relationship graph with the first basic node feature or the first sub-prediction node feature of the node corresponding to the node in the second node relationship graph to obtain the prediction node feature corresponding to the node.

[0072] The above point convolution refers to performing a linear transformation on the features of all feature channels of each node, and then generating a specified number of output channels. Each output channel is calculated by applying a convolution kernel to the data of the feature channel.

[0073] It should be noted that in conventional convolution operations, the convolution kernel will slide on the entire input feature map and perform calculations to generate a new feature map, while in point convolution, the predicted node features of each target node are obtained by summing up the first sub-node features and the second sub-node features corresponding to the input feature channel, which are independently multiplied by the convolution kernel. That is, the point convolution operation is performed at a single node level and does not involve the integration of spatial neighborhood information, which can effectively reduce the number of parameters and complexity of the reconstruction model.

[0074] In addition, the above-mentioned reconstruction model needs to be trained in advance before it can be deployed to the server, wherein the method for the server to train the above-mentioned reconstruction model can be to obtain a set of sample vibration signals, where each set of sample vibration signals can be composed of historical vibration signals collected by sensors at each signal collection point set on the target motor or other motors that are the same as the target motor.

[0075] Furthermore, the server can select at least part of the sample vibration signal sets from each sample vibration signal set as each test set, and use each sample vibration signal set other than each test set as each training set, and then for each training set, according to different sensor distribution strategies, screen out historical vibration signals collected by sensors installed at least part of the signal collection points from the training set as each basic historical vibration signal, and then input each basic historical vibration signal into the reconstruction model, so that the reconstruction model determines the node features of at least part of the nodes included in the pre-constructed node relationship graph according to each basic historical vibration signal, as each basic historical node feature, and determines the association relationship between each signal collection point according to the initial edge features (which can be randomly generated) between each node included in the node relationship graph, and predicts other historical vibration signals in the training set except for each historical basic vibration signal according to each basic historical node feature and the association relationship between each signal collection point, as a sample predicted vibration signal.

[0076] Furthermore, the server can take the deviation between each sample predicted vibration signal and the historical vibration signal corresponding to each sample predicted vibration signal as the optimization target, and adjust the initial edge features between the nodes contained in the above-mentioned node relationship graph to train the reconstruction model and obtain the trained reconstruction model.

[0077] The different sensor distribution strategies mentioned above can be Figure 3 shown.

[0078] Figure 3 A schematic diagram of the sensor distribution strategy provided in this specification.

[0079] Combination Figure 3 It can be seen that the above different sensor distribution strategies can be solutions for selecting sensors according to different interval rules, for example: Figure 3 In the scheme 1 and scheme 2, a specified number of closely connected sensors can be selected from each sensor. For example: Figure 3 In scheme 3 and scheme 4, each sensor can be selected from each sensor according to a specified interval rule.

[0080] S103: Reconstructing the vibration signal of the target motor according to the basic vibration signals and the predicted vibration signal.

[0081] Furthermore, the server can determine the vibration signals corresponding to the signal collection points of the target motor based on the basic vibration signals and the predicted vibration signals, and then perform a vibration test on the target motor based on the vibration signals corresponding to the signal collection points of the target motor to obtain the test results corresponding to the target motor.

[0082] From the above content, it can be seen that the server can predict the vibration signals at other signal collection points based on the vibration signals collected by the collectors set at some signal collection points through the graph neural network, and then test the target motor according to the collected vibration signals and the predicted vibration signals, thereby reducing the number of sensors required to test the target motor, thereby effectively reducing the testing cost.

[0083] In addition, reconstructing each vibration signal through a graph neural network to perform vibration testing on the target motor can also avoid the occurrence of erroneous vibration test results of the target motor due to unreasonable sensor position setting or failure of some sensors resulting in missing sensor data.

[0084] The above are one or more implementations of the motor vibration signal reconstruction method of this specification. Based on the same idea, this specification also provides a corresponding motor vibration signal reconstruction device, such as Figure 4 shown.

[0085] Figure 4 A schematic diagram of a motor vibration signal reconstruction device provided in this specification includes:

[0086] An acquisition module 401 is used to acquire each basic vibration signal of a target motor, wherein each basic vibration signal is acquired by a sensor installed at at least a portion of a signal acquisition location on the target motor, and the basic vibration signal is data used to characterize the mechanical vibration generated by the target motor under different test conditions;

[0087] A prediction module 402 is used to input the basic vibration signals into a preset reconstruction model, so that the reconstruction model determines node features of at least part of the nodes included in the pre-constructed node relationship diagram according to the basic vibration signals as basic node features, and determines the association relationship between the signal collection locations according to the edge features between the nodes included in the node relationship diagram, and predicts the vibration signal of the signal collection location where the sensor is not installed on the target motor according to the association relationship between the basic node features and the signal collection locations as a predicted vibration signal;

[0088] The execution module 403 is used to perform a task of reconstructing the vibration signal of the target motor according to the basic vibration signals and the predicted vibration signal.

[0089] Optionally, the prediction module 402 is specifically used to, for each basic vibration signal, determine a node corresponding to a sensor used to collect the basic vibration signal in a pre-constructed node relationship diagram as the node corresponding to the basic vibration signal; and perform feature extraction on the basic vibration signal to determine a vibration signal feature corresponding to the basic vibration signal as a basic node feature of the node corresponding to the basic vibration signal.

[0090] Optionally, the device further includes: a construction module 404;

[0091] The construction module 404 is specifically used to, for each signal collection point on the target motor, use the signal collection point as a node in the node relationship graph, and use the association relationship between the signal collection point and each other signal collection point as an edge in the node relationship graph, so as to construct the node relationship graph.

[0092] Optionally, the prediction module 402 is specifically used to, for each signal collection point on the target motor where no sensor is installed, determine the node corresponding to the signal collection point in the node relationship diagram as the target node, and determine the node corresponding to the signal collection point where the sensor for collecting each basic vibration signal is located in the node relationship diagram as each basic node; aggregate the basic node features of each basic node and the edge features used to characterize the association relationship between each signal collection point to the target node to obtain the predicted node features corresponding to the target node, and predict the vibration signal of the signal collection point on the target motor where no sensor is installed based on the predicted node features.

[0093] Optionally, the prediction module 402 is specifically used to perform multiple rounds of aggregation on each node included in the node relationship graph, so as to aggregate the basic node features of each basic node and the edge features used to characterize the association relationship between each signal collection point to the target node; wherein, for each round of aggregation, the initial node features of each node included in the node relationship graph in the round of aggregation are determined, and for each node included in the node relationship graph, the initial node features of the node, the initial node features of other nodes, and the edge features between the node and each other node are aggregated to obtain the corresponding aggregated node features of the node in the round of aggregation, and the corresponding aggregated node features of the node in the round of aggregation are used as the initial node features of the node in the next round of aggregation, and the initial node features are obtained by iterating the basic node features to the previous round of aggregation; when it is determined that the preset termination condition is met, the corresponding aggregated node features of the target node in the round of aggregation are used as the predicted node features corresponding to the target node.

[0094] Optionally, the prediction module 402 is specifically used to determine, for each node included in the node relationship graph, the node weight of each other node with respect to the node, and based on the node weight of each other node with respect to the node, fuse the initial node features of each other node and the edge features between the node and each other node with the initial node features of the node to obtain the corresponding aggregated node features of the node in this round of aggregation.

[0095] Optionally, the basic vibration signal includes: an acceleration signal of the target motor vibration, and an inertia signal of the target motor vibration, and the basic node feature includes: a first basic node feature determined according to the acceleration signal of the target motor vibration and a second basic node feature determined according to the inertia signal of the target motor vibration;

[0096] The prediction module 402 is specifically used to aggregate the first basic node features of each basic node and the edge features used to characterize the association relationship between each signal collection point to the target node to obtain the first sub-prediction node feature corresponding to the target node, and to aggregate the second basic node features of each basic node and the edge features used to characterize the association relationship between each signal collection point to the target node to obtain the second sub-prediction node feature corresponding to the target node; input the first sub-prediction node feature and the second sub-prediction node feature into a preset point convolutional network layer so that the point convolutional network layer fuses the first sub-prediction node feature and the second sub-prediction node feature to obtain the corresponding prediction node feature of the target node.

[0097] This specification also provides a computer-readable storage medium, which stores a computer program, which can be used to execute the above Figure 1 A motor vibration signal reconstruction method is provided.

[0098] This manual also provides Figure 5 The one shown corresponds to Figure 1 A schematic diagram of the electronic device. Figure 5 As mentioned above, at the hardware level, the electronic device includes a processor, an internal bus, a network interface, a memory, and a non-volatile memory, and may also include other hardware required for the business. The processor reads the corresponding computer program from the non-volatile memory into the memory and then runs it to achieve the above Figure 1 The motor vibration signal reconstruction method. Of course, in addition to the software implementation, this specification does not exclude other implementations, such as logic devices or a combination of software and hardware, etc., that is, the execution subject of the following processing flow is not limited to each logic unit, but can also be hardware or logic devices.

[0099] For the improvement of a technology, it can be clearly distinguished whether it is a hardware improvement (for example, improvement of the circuit structure of diodes, transistors, switches, etc.) or a software improvement (improvement of the method flow). However, with the development of technology, many improvements of the method flow today can be regarded as direct improvements of the hardware circuit structure. Designers almost always obtain the corresponding hardware circuit structure by programming the improved method flow into the hardware circuit. Therefore, it cannot be said that the improvement of a method flow cannot be implemented with a hardware entity module. For example, a programmable logic device (PLD) (such as a field programmable gate array (FPGA)) is such an integrated circuit whose logical function is determined by the user's programming of the device. Designers can "integrate" a digital system on a PLD by programming themselves, without having to ask chip manufacturers to design and make dedicated integrated circuit chips. Moreover, nowadays, instead of manually making integrated circuit chips, this kind of programming is mostly implemented by "logic compiler" software, which is similar to the software compiler used when developing and writing programs, and the original code before compilation must also be written in a specific programming language, which is called hardware description language (HDL). There is not only one HDL, but many kinds, such as ABEL (Advanced Boolean Expression Language), AHDL (Altera Hardware Description Language), Confluence, CUPL (Cornell University Programming Language), HDCal, JHDL (Java Hardware Description Language), Lava, Lola, MyHDL, PALASM, RHDL (Ruby Hardware Description Language), etc. The most commonly used ones are VHDL (Very-High-Speed ​​Integrated Circuit Hardware Description Language) and Verilog. Those skilled in the art should also know that it is only necessary to program the method flow slightly in the above-mentioned hardware description languages ​​and program it into the integrated circuit, and then it is easy to obtain the hardware circuit that implements the logic method flow.

[0100] The controller can be implemented in any appropriate manner, for example, the controller can take the form of a microprocessor or processor and a computer-readable medium storing a computer-readable program code (such as software or firmware) that can be executed by the (micro)processor, a logic gate, a switch, an application-specific integrated circuit (ASIC), a programmable logic controller, and an embedded microcontroller. Examples of controllers include, but are not limited to, the following microcontrollers: ARC 625D, Atmel AT91SAM, Microchip PIC18F26K20, and Silicone Labs C8051F320. The memory controller can also be implemented as part of the control logic of the memory. Those skilled in the art also know that in addition to implementing the controller in a purely computer-readable program code manner, the controller can be implemented in the form of a logic gate, a switch, an application-specific integrated circuit, a programmable logic controller, and an embedded microcontroller by logically programming the method steps. Therefore, this controller can be considered as a hardware component, and the devices included therein for implementing various functions can also be regarded as structures within the hardware component. Or even, the devices for implementing various functions can be regarded as both software modules for implementing the method and structures within the hardware component.

[0101] The systems, devices, modules or units described in the above embodiments may be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, the computer may be, for example, a personal computer, a laptop computer, a cellular phone, a camera phone, a smart phone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or a combination of any of these devices.

[0102] For the convenience of description, the above device is described in various units according to their functions. Of course, when implementing this specification, the functions of each unit can be implemented in the same or multiple software and / or hardware.

[0103] Those skilled in the art will appreciate that the embodiments of this specification may be provided as methods, systems, or computer program products. Therefore, this specification may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, this specification may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0104] This specification is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of this specification. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0105] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0106] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0107] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0108] The memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0109] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0110] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0111] Those skilled in the art will appreciate that the embodiments of this specification may be provided as methods, systems or computer program products. Therefore, this specification may take the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware. Moreover, this specification may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0112] This specification may be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. This specification may also be practiced in distributed computing environments where tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules may be located in local and remote computer storage media, including storage devices.

[0113] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0114] The above description is only an embodiment of the present specification and is not intended to limit the present specification. For those skilled in the art, the present specification may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present specification shall be included in the scope of the claims of the present specification.

Claims

1. A method for reconstructing a motor vibration signal, characterized in that: include: Acquire each basic vibration signal of the target motor, wherein each basic vibration signal is collected by a sensor installed at at least part of a signal collection location on the target motor, and the basic vibration signal is data used to characterize the mechanical vibration generated by the target motor under different test conditions; Inputting each basic vibration signal into a preset reconstruction model, so that the reconstruction model determines node features of at least part of the nodes included in a pre-constructed node relationship graph according to each basic vibration signal as each basic node feature, and determines the association relationship between each signal collection point according to the edge features between each node included in the node relationship graph, and predicts the vibration signal of the signal collection point where the sensor is not installed on the target motor according to the association relationship between each basic node feature and each signal collection point, as a predicted vibration signal; A task of reconstructing the vibration signal of the target motor is performed according to the basic vibration signals and the predicted vibration signal.

2. The method according to claim 1, characterized in that According to the basic vibration signals, node features of at least some nodes included in the pre-constructed node relationship diagram are determined as basic node features, specifically including: For each basic vibration signal, determine a node corresponding to a sensor used to collect the basic vibration signal in a pre-constructed node relationship graph as the node corresponding to the basic vibration signal; Feature extraction is performed on the basic vibration signal to determine a vibration signal feature corresponding to the basic vibration signal as a basic node feature of the node corresponding to the basic vibration signal.

3. The method according to claim 1, characterized in that Construct a node relationship graph, including: For each signal collection point on the target motor, the signal collection point is used as a node in a node relationship graph, and the association relationship between the signal collection point and each other signal collection point is used as an edge in the node relationship graph to construct the node relationship graph.

4. The method according to claim 1, characterized in that Predicting the vibration signal of the signal collection point on the target motor where no sensor is installed according to the correlation between the basic node features and the signal collection points specifically includes: For each signal collection location on the target motor where no sensor is installed, determine the node corresponding to the signal collection location in the node relationship diagram as the target node, and determine the node corresponding to the signal collection location where the sensor for collecting each basic vibration signal is located in the node relationship diagram as each basic node; The basic node features of each basic node and the edge features used to characterize the association relationship between each signal collection point are aggregated to the target node to obtain the predicted node features corresponding to the target node, and based on the predicted node features, the vibration signal of the signal collection point on the target motor where no sensor is installed is predicted.

5. The method according to claim 4, characterized in that Aggregating the basic node features of each basic node and the edge features used to characterize the association relationship between each signal collection point to the target node to obtain the predicted node features corresponding to the target node, specifically including: Perform multiple rounds of aggregation on each node included in the node relationship graph to aggregate the basic node features of each basic node and the edge features used to characterize the association relationship between each signal collection point to the target node; wherein, For each round of aggregation, the initial node features of each node included in the node relationship graph in the round of aggregation are determined, and for each node included in the node relationship graph, the initial node features of the node, the initial node features of other nodes, and the edge features between the node and each other node are aggregated to obtain the corresponding aggregated node features of the node in the round of aggregation, and the corresponding aggregated node features of the node in the round of aggregation are used as the initial node features of the node in the next round of aggregation, wherein the initial node features are obtained by iterating the basic node features to the previous round of aggregation; When it is determined that the preset termination condition is met, the aggregated node feature corresponding to the target node in this round of aggregation is used as the predicted node feature corresponding to the target node.

6. The method according to claim 5, characterized in that For each node included in the node relationship graph, the initial node feature of the node, the initial node features of other nodes, and the edge features between the node and each other node are aggregated to obtain the corresponding aggregated node features of the node in this round of aggregation, specifically including: For each node included in the node relationship graph, the node weight of each other node with respect to the node is determined, and based on the node weight of each other node with respect to the node, the initial node features of each other node and the edge features between the node and each other node are fused with the initial node features of the node to obtain the corresponding aggregated node features of the node in this round of aggregation.

7. The method according to claim 4, characterized in that The basic vibration signal includes: an acceleration signal of the target motor vibration, and an inertia signal of the target motor vibration; the basic node feature includes: a first basic node feature determined according to the acceleration signal of the target motor vibration and a second basic node feature determined according to the inertia signal of the target motor vibration; Aggregating the basic node features of each basic node and the edge features used to characterize the association relationship between each signal collection point to the target node to obtain the predicted node features corresponding to the target node, specifically including: Aggregating the first basic node features of each basic node and the edge features used to characterize the association relationship between each signal collection location to the target node to obtain a first sub-prediction node feature corresponding to the target node, and aggregating the second basic node features of each basic node and the edge features used to characterize the association relationship between each signal collection location to the target node to obtain a second sub-prediction node feature corresponding to the target node; The first sub-prediction node feature and the second sub-prediction node feature are input into a preset point convolutional network layer, so that the point convolutional network layer fuses the first sub-prediction node feature and the second sub-prediction node feature to obtain the corresponding prediction node feature of the target node.

8. A motor vibration signal reconstruction device, characterized in that: include: an acquisition module, configured to acquire each basic vibration signal of the target motor, wherein each basic vibration signal is acquired by a sensor installed at at least a portion of a signal acquisition location on the target motor, and the basic vibration signal is data for characterizing the mechanical vibration generated by the target motor under different test conditions; A prediction module, configured to input each basic vibration signal into a preset reconstruction model, so that the reconstruction model determines node features of at least part of the nodes included in a pre-constructed node relationship graph according to each basic vibration signal as each basic node feature, and determines the association relationship between each signal collection location according to the edge features between each node included in the node relationship graph, and predicts the vibration signal of the signal collection location on which the sensor is not installed on the target motor according to the association relationship between each basic node feature and each signal collection location as a predicted vibration signal; An execution module is used to perform a task of reconstructing the vibration signal of the target motor according to the basic vibration signals and the predicted vibration signal.

9. A computer-readable storage medium, characterized in that: The storage medium stores a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the method according to any one of claims 1 to 7 is implemented.

Images