A fault diagnosis method and device for a main shaft bearing of a numerical control machine tool

Abstract

The application provides a fault diagnosis method and device for a main shaft bearing of a numerical control machine tool, the method comprising: obtaining a vibration signal of the main shaft bearing of the machine tool, dividing the vibration signal into a fault sample set and labeling the vibration signal; calculating the similarity between samples in the fault sample set according to dynamic time similarity, and building a fault sample graph by using a preset model; building a fault diagnosis model for the main shaft bearing of the numerical control machine tool based on a topological adaptive graph convolution network; training the fault diagnosis model for the main shaft bearing of the numerical control machine tool, and evaluating the diagnosis result of the fault diagnosis model. The method provided by the application can directly learn the characteristics of the vibration signal, does not depend on feature extraction of vibration signal processing, and can ensure the accuracy of the diagnosis result of the fault diagnosis model under the condition of a small number of labeled samples.

Description

Technical Field

[0001] This invention relates to the field of bearing vibration signal identification, and specifically to a fault diagnosis method and device for CNC machine tool spindle bearings. Background Technology

[0002] CNC machine tools are widely used in the machining of parts for various equipment such as ships, vehicles, aircraft, and high-speed trains. As an important production equipment, ensuring its normal operation is of great significance to improving the economic benefits of enterprises. The spindle bearing is one of the most critical components of a CNC machine tool, supporting the machine tool during long-term operation. It faces harsh working environments with varying operating conditions and high loads, and is prone to failure, leading to machine tool damage and seriously threatening the safe production of enterprises and the personal safety of workers.

[0003] Faced with the non-stationary and nonlinear characteristics of vibration signals from CNC machine tool spindle bearings, traditional fault diagnosis algorithms based on feature extraction and pattern recognition rely too heavily on manually designed feature indicators, lacking adaptability and resulting in poor generalization performance. Classic deep learning-based fault diagnosis methods require a huge number of training samples and class labels, leading to numerous parameters to be trained and an overly complex training process, making it difficult to meet the complexity and uncertainty requirements of modern equipment.

[0004] Deep learning-based fault diagnosis models can only achieve ideal fault diagnosis results when there are a sufficient number of labeled fault samples. In real-world scenarios, when a CNC machine tool malfunctions, it may continue to operate for safety reasons. This results in a scarcity of labeled fault samples, leading to an imbalance in the fault sample pool and affecting the accuracy of deep learning-based fault diagnosis models. Summary of the Invention

[0005] In view of this, it is necessary to provide a fault diagnosis method and device for CNC machine tool spindle bearings to solve the problem of low fault diagnosis accuracy of CNC machine tool spindle bearing fault models when there are few bearing data tags.

[0006] To address the aforementioned technical problems, this invention provides a method for diagnosing faults in CNC machine tool spindle bearings, comprising:

[0007] The vibration signal of the machine tool spindle bearing is acquired, and the vibration signal is divided into a fault sample set and labeled according to the health condition of the machine tool spindle bearing;

[0008] The fault sample set is subjected to max-min standardization, and the similarity between samples in the fault sample set is calculated based on dynamic temporal similarity. A fault sample map is constructed using a preset model.

[0009] A fault diagnosis model for CNC machine tool spindle bearings based on topology adaptive graph convolutional networks was constructed.

[0010] The fault sample set is used to train the fault diagnosis model of the CNC machine tool spindle bearing, and the diagnosis results of the fault diagnosis model of the CNC machine tool spindle bearing are evaluated.

[0011] In one possible implementation, acquiring the vibration signal of the machine tool spindle bearing, and dividing the vibration signal into a fault sample set and labeling it according to the health condition of the machine tool spindle bearing, includes:

[0012] Vibration signals of machine tool spindle bearings under different health conditions are obtained using an accelerometer.

[0013] The vibration signals are divided into fault sample sets according to the different health conditions of the machine tool spindle bearing and then labeled.

[0014] The health status of the machine tool spindle bearing is classified into at least two of the following: normal, inner ring failure, outer ring failure, cage failure, and lubrication failure.

[0015] In one possible implementation, the step of performing max-min normalization on the fault sample set, calculating the similarity between two samples based on dynamic temporal similarity, and constructing a fault sample map using a preset model includes:

[0016] The fault sample set is subjected to max-min standardization.

[0017] The similarity between samples in the fault sample set is calculated based on a similarity algorithm;

[0018] Based on the K-nearest neighbor criterion, the first sample with the highest similarity is connected by an edge;

[0019] A fault sample graph is constructed using samples from the fault sample set as nodes and the connections between samples in the fault sample set as edges.

[0020] In one possible implementation, the similarity algorithm is calculated using the following formula:

[0021]

[0022] Where D(a,b) represents the similarity value between samples, and d(a,b)=|h i (a)-h j (b)|,h i (a) represents the a-th sample point of the i-th sample, h j (b) represents the b-th sample point of the j-th sample, and m represents the number of data points in each sample.

[0023] In one possible implementation, the adjacency matrix between the samples is expressed as:

[0024]

[0025] Among them, A ij Let x represent the adjacency matrix between samples. i ∈KNN(x j ) represents x i It is x j The K nearest neighbors with the highest similarity x represents i Not x j The K nearest neighbors with the highest similarity.

[0026] In one possible implementation, the construction of the CNC machine tool spindle bearing fault diagnosis model based on a topology adaptive graph convolutional network includes:

[0027] Calculate the Laplace matrix of the fault sample graph based on the connection relationships between the nodes in the fault sample graph;

[0028] A topological adaptive graph convolutional layer is constructed based on the Laplacian matrix of the fault sample graph;

[0029] Based on the aforementioned topology adaptive graph convolutional layer, a fault diagnosis model for CNC machine tool spindle bearings with two layers of topology adaptive graph convolutional layers is constructed.

[0030] In one possible implementation, the formula for calculating the Laplace matrix of the fault sample graph based on the connection relationships between the nodes in the fault sample graph is as follows:

[0031]

[0032]

[0033] Among them, L sym D represents the Laplace matrix of the fault sample map. ii A represents the degree matrix between samples. ij Let A represent the adjacency matrix between samples, D represent the degree matrix of the fault sample graph nodes, and A represent the adjacency matrix of the fault sample graph nodes.

[0034] In one possible implementation, the activation function of the first layer topological adaptive graph convolution of the CNC machine tool spindle bearing fault diagnosis model is ReLU, and the activation function of the first layer topological adaptive graph convolution and the second layer topological adaptive graph convolution of the CNC machine tool spindle bearing fault diagnosis model are the same.

[0035] In one possible implementation, the fault sample set includes a training dataset and a test dataset. The step of training the CNC machine tool spindle bearing fault diagnosis model using the fault sample set and evaluating the diagnostic results of the CNC machine tool spindle bearing fault diagnosis model includes:

[0036] The CNC machine tool spindle bearing fault diagnosis model is trained using the training dataset, and then the test dataset is input into the trained CNC machine tool spindle bearing fault diagnosis model to obtain the fault diagnosis labels of the samples in the test dataset.

[0037] To address the aforementioned problems, the present invention also provides a fault diagnosis device for CNC machine tool spindle bearings, comprising:

[0038] The signal acquisition module is used to acquire the vibration signal of the machine tool spindle bearing, and divide the vibration signal into a fault sample set and label it according to the health condition of the machine tool spindle bearing;

[0039] The fault sample map building module is used to calculate the similarity between samples in the fault sample set based on dynamic time similarity and to build a fault sample map using a preset model.

[0040] The diagnostic model building module is used to build a fault diagnosis model for CNC machine tool spindle bearings based on topological adaptive graph convolutional networks.

[0041] The evaluation module is used to train the fault diagnosis model of the CNC machine tool spindle bearing and evaluate the diagnosis results of the fault diagnosis model of the CNC machine tool spindle bearing.

[0042] The beneficial effects of the above embodiments are as follows: the present invention constructs a fault sample set from bearing vibration signals collected under various health conditions; constructs a fault sample map by measuring the dynamic time planning similarity between samples; and realizes the diagnosis and identification of fault states by constructing and training a topological adaptive graph convolutional network model. The method proposed in this invention can not only directly learn features from vibration signals without relying on feature extraction from vibration signal processing, but also ensure the accuracy of diagnostic results with a small number of labeled samples. Attached Figure Description

[0043] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0044] Figure 1This is a flowchart illustrating an embodiment of a fault diagnosis method for CNC machine tool spindle bearings provided by the present invention.

[0045] Figure 2 This is a schematic diagram illustrating the construction of a fault sample set in an embodiment of a fault diagnosis method for CNC machine tool spindle bearings provided by the present invention.

[0046] Figure 3 This is a schematic diagram of a scenario in an embodiment of a fault sample diagram in a fault diagnosis method for CNC machine tool spindle bearings provided by the present invention;

[0047] Figure 4 This is a schematic diagram of a topological adaptive graph convolutional network structure, representing one embodiment of a fault diagnosis method for CNC machine tool spindle bearings provided by the present invention.

[0048] Figure 5 A schematic diagram of one embodiment of a fault diagnosis device for CNC machine tool spindle bearings provided by the present invention;

[0049] Figure 6 This is a schematic diagram of another embodiment of a fault diagnosis device for CNC machine tool spindle bearings provided by the present invention. Detailed Implementation

[0050] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0051] Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities can be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor systems and / or microcontroller systems.

[0052] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of the invention. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0053] This invention provides a method and apparatus for diagnosing faults in CNC machine tool spindle bearings, which will be described below.

[0054] Figure 1 This is a flowchart illustrating an embodiment of a fault diagnosis method for CNC machine tool spindle bearings provided by the present invention.

[0055] Reference Figure 1 This invention provides a method for diagnosing faults in CNC machine tool spindle bearings, comprising:

[0056] S101. Obtain the vibration signal of the machine tool spindle bearing, and divide the vibration signal into a fault sample set and label it according to the health status of the machine tool spindle bearing;

[0057] S102. Perform max-min standardization on the fault sample set, calculate the similarity between samples in the fault sample set based on dynamic time similarity, and build a fault sample map using a preset model.

[0058] S103. Construct a fault diagnosis model for CNC machine tool spindle bearings based on topological adaptive graph convolutional networks;

[0059] S104. The fault sample set is used to train the fault diagnosis model of the CNC machine tool spindle bearing, and the diagnosis results of the fault diagnosis model of the CNC machine tool spindle bearing are evaluated.

[0060] This invention constructs a fault sample set from bearing vibration signals collected under various health conditions; it constructs a fault sample map by measuring the dynamic time planning similarity between samples; and it achieves fault state diagnosis and identification by constructing and training a topological adaptive graph convolutional network model. The method proposed in this invention not only directly learns features from vibration signals without relying on feature extraction from vibration signal processing, but also ensures high accuracy of diagnostic results even with a small number of labeled samples.

[0061] Figure 2 This is a schematic diagram illustrating the construction of a fault sample set in an embodiment of a fault diagnosis method for CNC machine tool spindle bearings provided by the present invention.

[0062] Reference Figure 2 In one possible implementation, acquiring the vibration signal of the machine tool spindle bearing, and dividing the vibration signal into a fault sample set and labeling it according to the health condition of the machine tool spindle bearing, includes:

[0063] Vibration signals of machine tool spindle bearings under different health conditions are obtained using an accelerometer.

[0064] The vibration signals are divided into fault sample sets according to the different health conditions of the machine tool spindle bearing and then labeled.

[0065] The health status of the machine tool spindle bearing is classified into at least two of the following: normal, inner ring failure, outer ring failure, cage failure, and lubrication failure.

[0066] For example, the spindle bearing of a standard CNC machine tool was used as the experimental object. The experimental bearing model was NSK-40BNR10H deep groove ball rolling bearing, which supports the spindle and the tool for cutting motion. The vibration signal of the tested bearing was collected by a Dytran3263A2 three-dimensional vibration sensor, and the vibration signal of the bearing during operation was collected by an NI9234 data acquisition card at a sampling frequency of 25.6kHz. The spindle bearing has five health conditions, including normal (NOR), inner ring failure (IRF), outer ring failure (ORF), cage failure (RBF), and lubrication failure (LSF).

[0067] For each bearing health condition, the spindle bearing vibration signal was continuously acquired for 150 seconds using an NI9234 data acquisition card at a sampling frequency of 25.6kHz. Each sample consisted of 2048 data points, and 300 samples were collected for each health condition, for a total of 1500 samples for the five bearing health conditions. These 1500 samples were used as the fault sample set.

[0068] Figure 3 This invention provides a schematic diagram of a scenario in an embodiment of a fault sample diagram in a fault diagnosis method for CNC machine tool spindle bearings.

[0069] Reference Figure 3 In one embodiment, the step of performing max-min normalization on the fault sample set, calculating the similarity between two samples based on dynamic temporal similarity, and constructing a fault sample map using a preset model includes:

[0070] The fault sample set is subjected to max-min standardization.

[0071] The similarity between samples in the fault sample set is calculated based on a similarity algorithm;

[0072] Based on the K-nearest neighbor criterion, the first sample with the highest similarity is connected by an edge;

[0073] A fault sample graph is constructed using samples from the fault sample set as nodes and the connections between samples in the fault sample set as edges.

[0074] In one embodiment, the similarity algorithm is calculated using the following formula:

[0075]

[0076] Where D(a,b) represents the similarity value between samples, and d(a,b)=|h i (a)-hj (b)|,h i (a) represents the a-th sample point of the i-th sample, h j (b) represents the b-th sample point of the j-th sample, and m represents the number of data points in each sample. The smaller the value of D(a, b), the more similar the two samples are.

[0077] In one embodiment, the adjacency matrix between the samples is expressed as:

[0078]

[0079] Among them, A ij Let x represent the adjacency matrix between samples. i ∈KNN(x j ) represents x i It is x j The K nearest neighbors with the highest similarity x represents i Not x j The K nearest neighbors with the highest similarity.

[0080] Using sample x from the fault sample set i Let V be a node, and let the connection relationships between the samples in the fault sample set be edges E, construct the fault sample graph G(V, E).

[0081] Figure 4 This is a schematic diagram of a topological adaptive graph convolutional network structure, representing one embodiment of a fault diagnosis method for CNC machine tool spindle bearings provided by the present invention.

[0082] Reference Figure 4 In one embodiment, the construction of the CNC machine tool spindle bearing fault diagnosis model based on a topological adaptive graph convolutional network includes:

[0083] Calculate the Laplace matrix L of the fault sample graph G(V, E) based on the connection relationship between the nodes of the fault sample graph;

[0084] A topological adaptive graph convolutional layer is constructed based on the Laplacian matrix of the fault sample graph;

[0085] Based on the aforementioned topology adaptive graph convolutional layer, a fault diagnosis model for CNC machine tool spindle bearings with two layers of topology adaptive graph convolutional layers is constructed.

[0086] In one embodiment, the formula for calculating the Laplacian matrix of the fault sample graph based on the connection relationships between the nodes of the fault sample graph is as follows:

[0087]

[0088]

[0089] Among them, L sym D represents the Laplace matrix of the fault sample map. ii A represents the degree matrix between samples. ij Let A represent the adjacency matrix between samples, D represent the degree matrix of the fault sample graph nodes, and A represent the adjacency matrix of the fault sample graph nodes.

[0090] In one embodiment, the formula for constructing the topology adaptive graph convolutional layer is as follows:

[0091] H (l+1) =σ(G (l) H (l) +bI)

[0092] Among them, H (l) Let bI represent the input feature matrix of the node, bI represent the bias term, and H represent the input feature matrix of the node. (l+1) G represents the output node feature matrix obtained after topological adaptive graph convolution operation. (l) σ is a multi-topological graph convolution kernel of graph convolution layer l, where σ represents the activation function.

[0093] Furthermore, the expression for constructing a CNC machine tool spindle bearing fault classification model with two layers of topological adaptive graph convolutional layers is as follows:

[0094] Y = softmax(FCL(ReLU(G)) 1 ReLU(G 0 H 0 +b 0 I)+b 1 I)))

[0095] The output Y is the label of the sample.

[0096] In one embodiment, the activation function of the first layer of the topological adaptive graph convolution of the CNC machine tool spindle bearing fault diagnosis model is ReLU, the activation function of the first layer of the topological adaptive graph convolution of the CNC machine tool spindle bearing fault diagnosis model is the same as the activation function of the second layer of the topological adaptive graph convolution, and the last layer is a fully connected layer.

[0097] In one embodiment, the fault sample set includes a training dataset and a test dataset. The step of training the CNC machine tool spindle bearing fault diagnosis model using the fault sample set and evaluating the diagnostic results of the CNC machine tool spindle bearing fault diagnosis model includes:

[0098] The CNC machine tool spindle bearing fault diagnosis model is trained using the training dataset, and then the test dataset is input into the trained CNC machine tool spindle bearing fault diagnosis model to obtain the fault diagnosis labels of the samples in the test dataset.

[0099] Furthermore, the nodes of the fault sample map in the fault sample set are divided into training dataset and test dataset according to different proportions, and the training sample set is input into the CNC machine tool spindle bearing fault diagnosis model for model training.

[0100] For example, after obtaining the fault sample images, the fault sample set is divided into training and test datasets according to the label ratio given in the table below. The training dataset is input into a multi-topology graph convolutional network model for model training, selecting the target fault diagnosis model for CNC machine tool spindle bearings. The number of hidden nodes is set to 64, the learning rate to 0.001, the number of iterations to 500, and the convolution parameter of the topology adaptive graph convolutional layer is set to 3. Finally, the test dataset is input into the trained CNC machine tool spindle bearing fault diagnosis model to obtain the fault diagnosis labels for the test data samples.

[0101] Accuracy of multi-topology graph convolutional networks under different label ratios

[0102]

[0103] The accuracy of the test model under different label ratios is calculated using the following formula:

[0104]

[0105] Furthermore, the loss function is the cross-entropy function. Where Y is the actual label of the training set sample. Predict labels for the model. Obtain the target fault diagnosis model for CNC machine tool spindle bearings through multiple iterations of the backpropagation algorithm.

[0106] In one embodiment, the Adam algorithm is selected as the optimization algorithm for the fault diagnosis model of CNC machine tool spindle bearing to optimize the fault diagnosis model of CNC machine tool spindle bearing.

[0107] Figure 5 This is a schematic diagram of one embodiment of a fault diagnosis device for CNC machine tool spindle bearings provided by the present invention.

[0108] Reference Figure 5 The present invention also provides a fault diagnosis device for CNC machine tool spindle bearings, comprising:

[0109] The signal acquisition module 501 is used to acquire the vibration signal of the machine tool spindle bearing, and divide the vibration signal into a fault sample set and label it according to the health status of the machine tool spindle bearing;

[0110] The fault sample map building module 502 is used to calculate the similarity between samples in the fault sample set based on dynamic time similarity and to build a fault sample map using a preset model.

[0111] The diagnostic model building module 503 is used to build a fault diagnosis model for CNC machine tool spindle bearings based on a topology adaptive graph convolutional network.

[0112] Evaluation module 504 is used to train the fault diagnosis model of the CNC machine tool spindle bearing and evaluate the diagnosis results of the fault diagnosis model of the CNC machine tool spindle bearing.

[0113] The beneficial effects of the above embodiments are as follows: the signal acquisition module 501 is used to acquire the vibration signal of the machine tool spindle bearing, divide the vibration signal into a fault sample set and label it according to the health status of the machine tool spindle bearing; the fault sample map building module 502 is used to calculate the similarity between samples in the fault sample set according to dynamic time similarity, and build a fault sample map using a preset model; the diagnostic model building module 503 is used to build a CNC machine tool spindle bearing fault diagnosis model based on a topological adaptive graph convolutional network; the evaluation module 504 is used to train the CNC machine tool spindle bearing fault diagnosis model and evaluate the diagnostic results of the CNC machine tool spindle bearing fault diagnosis model. The method proposed in this invention can not only directly learn the features of the vibration signal without relying on the feature extraction of vibration signal processing, but also ensure the accuracy of the diagnostic results with a small number of labeled samples.

[0114] Figure 6 This is a schematic diagram of another embodiment of a fault diagnosis device for CNC machine tool spindle bearings provided by the present invention.

[0115] Reference Figure 6 The present invention also provides a fault diagnosis device 600 for CNC machine tool spindle bearings. The fault diagnosis device 600 for CNC machine tool spindle bearings includes a processor 601, a memory 602, and a display 603. Figure 6 Only some components of the fault diagnosis device 600 for CNC machine tool spindle bearings are shown. However, it should be understood that it is not required to implement all the components shown, and more or fewer components may be implemented instead.

[0116] In some embodiments, memory 602 may be an internal storage unit of the fault diagnosis device 600 for CNC machine tool spindle bearings, such as a hard disk or memory of the fault diagnosis device 600 for CNC machine tool spindle bearings. In other embodiments, memory 602 may also be an external storage device of the fault diagnosis device 600 for CNC machine tool spindle bearings, such as a pluggable hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc., equipped on the fault diagnosis device 600 for CNC machine tool spindle bearings.

[0117] Furthermore, the memory 602 may include both internal storage units of the fault diagnosis device 600 for CNC machine tool spindle bearings and external storage devices. The memory 602 is used to store the application software and various types of data of the fault diagnosis device 600 for CNC machine tool spindle bearings.

[0118] In some embodiments, processor 601 may be a central processing unit (CPU), microprocessor, or other data processing chip, used to run program code stored in memory 602 or process data, such as the fault diagnosis method for machine tool spindle bearings in this invention.

[0119] In some embodiments, display 603 may be an LED display, a liquid crystal display, a touch-screen liquid crystal display, or an OLED (Organic Light-Emitting Diode) touchscreen. Display 603 is used to display information from the fault diagnosis device 600 for CNC machine tool spindle bearings and to display a visual user interface. Components 601-603 of the fault diagnosis device 600 for CNC machine tool spindle bearings communicate with each other via a system bus.

[0120] In some embodiments of the present invention, when the processor 601 executes the fault diagnosis method for the CNC machine tool spindle bearing in the memory 602, the following steps can be implemented:

[0121] The vibration signal of the machine tool spindle bearing is acquired, and the vibration signal is divided into a fault sample set and labeled according to the health condition of the machine tool spindle bearing;

[0122] The fault sample set is subjected to max-min standardization, and the similarity between samples in the fault sample set is calculated based on dynamic temporal similarity. A fault sample map is constructed using a preset model.

[0123] A fault diagnosis model for CNC machine tool spindle bearings based on topology adaptive graph convolutional networks was constructed.

[0124] The fault sample set is used to train the fault diagnosis model of the CNC machine tool spindle bearing, and the diagnosis results of the fault diagnosis model of the CNC machine tool spindle bearing are evaluated.

[0125] It should be understood that when the processor 601 executes the calculation program for fault diagnosis of the CNC machine tool spindle bearing in the memory 602, in addition to the functions mentioned above, it can also perform other functions, as can be found in the description of the corresponding method embodiments above.

[0126] Accordingly, this application also provides a computer-readable storage medium for storing computer-readable programs or instructions. When the programs or instructions are executed by a processor, they can implement the steps or functions of the fault diagnosis method for CNC machine tool spindle bearings provided in the above-described method embodiments.

[0127] The above embodiments provide a fault diagnosis method for CNC machine tool spindle bearings, which can realize the technical solutions described in the embodiments of the fault diagnosis device for CNC machine tool spindle bearings. The specific implementation principles of each module or unit can be found in the corresponding content of the embodiments of the fault diagnosis device for CNC machine tool spindle bearings, and will not be repeated here.

[0128] The fault diagnosis device for CNC machine tool spindle bearings provided by the present invention has been described in detail above. Specific examples have been used to illustrate the principle and implementation of the present invention. The description of the above embodiments is only for the purpose of helping to understand the method and core idea of ​​the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation and application scope based on the idea of ​​the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A method for fault diagnosis of a main spindle bearing of a numerically controlled machine tool, characterized in that, The method comprises the following steps: acquiring vibration signals of a machine tool spindle bearing, and dividing the vibration signals into a fault sample set and labeling according to the health status of the machine tool spindle bearing; performing maximum and minimum standardization processing on the fault sample set, calculating the similarity between samples in the fault sample set according to dynamic time similarity, and building a fault sample graph by using a preset model, comprising: performing maximum and minimum standardization processing on the fault sample set; calculating the similarity between samples in the fault sample set based on a similarity algorithm; connecting the first sample with the highest similarity based on the K-neighborhood criterion; taking the samples of the fault sample set as nodes and the connection relationship between the samples of the fault sample set as edges to construct a fault sample graph; building a numerical control machine tool spindle bearing fault diagnosis model based on a topological adaptive graph convolution network, comprising: calculating the Laplacian matrix of the fault sample graph according to the connection relationship between the nodes of the fault sample graph; constructing a topological adaptive graph convolution layer based on the Laplacian matrix of the fault sample graph; and constructing a numerical control machine tool spindle bearing fault diagnosis model with two layers of topological adaptive graph convolution layers based on the topological adaptive graph convolution layer; training the numerical control machine tool spindle bearing fault diagnosis model by using the fault sample set, and evaluating the diagnosis result of the numerical control machine tool spindle bearing fault diagnosis model.

2. The method for diagnosing a failure of a main shaft bearing of a numerical control machine tool according to claim 1, characterized in that, The method comprises the following steps: acquiring vibration signals of a machine tool spindle bearing, and dividing the vibration signals into a fault sample set and labeling according to the health status of the machine tool spindle bearing; acquiring vibration signals of a machine tool spindle bearing in different health statuses based on an acceleration sensor; dividing the vibration signals into a fault sample set and labeling according to the different health statuses of the machine tool spindle bearing; 3. The method of fault diagnosis of a CNC machine tool main spindle bearing according to claim 1, characterized in that, wherein the health status of the machine tool spindle bearing is divided into at least two of normal, inner ring fault, outer ring fault, cage fault and lubrication fault. ； in, This represents the similarity value between samples. , Indicates the first The first sample One sample point, Indicates the first The first sample There are 100 sample points, where m represents the number of data points in each sample.

4. The method of fault diagnosis of a CNC machine tool main spindle bearing according to claim 1, characterized in that, The calculation formula of the similarity algorithm is: ； wherein, denotes an adjacency matrix between samples, denotes is the K nearest neighbors with the highest similarity, denotes is not the K nearest neighbors with the highest similarity.

5. The method of fault diagnosis of a CNC machine tool spindle bearing according to claim 1, characterized in that, The expression of the adjacency matrix between the samples is: wherein, L represents a Laplacian matrix of a failure sample graph, D represents a degree matrix of nodes of the failure sample graph, and A represents an adjacency matrix of nodes of the failure sample graph ， D represents a degree matrix of nodes of the failure sample graph, and A represents an adjacency matrix of nodes of the failure sample graph.

6. The method of fault diagnosis of a CNC machine tool spindle bearing according to claim 1, characterized in that, The calculation formula for calculating the Laplacian matrix of the fault sample graph according to the connection relationship between the nodes of the fault sample graph is:

7. The method of fault diagnosis of a CNC machine tool spindle bearing according to claim 1, characterized in that, The activation function of the first layer of topological adaptive graph convolution of the numerical control machine tool spindle bearing fault diagnosis model is Relu, and the activation functions of the first layer of topological adaptive graph convolution and the second layer of topological adaptive graph convolution of the numerical control machine tool spindle bearing fault diagnosis model are the same. The fault sample set comprises a training data set and a test data set, and the training of the numerical control machine tool spindle bearing fault diagnosis model by using the fault sample set and the evaluation of the diagnosis result of the numerical control machine tool spindle bearing fault diagnosis model comprise:

8. A fault diagnosis device for a main spindle bearing of a numerically controlled machine tool, characterized by, training the numerical control machine tool spindle bearing fault diagnosis model by using the training data set, and then inputting the test data set into the trained numerical control machine tool spindle bearing fault diagnosis model to obtain the fault diagnosis label of the samples in the test data set. The method comprises the following steps: a signal acquisition module is configured to acquire vibration signals of a machine tool spindle bearing, and divide the vibration signals into a fault sample set and label according to the health status of the machine tool spindle bearing; The fault sample graph establishment module is configured to calculate the similarity between samples in the fault sample set based on dynamic time similarity, and to establish a fault sample graph using a preset model, including: performing maximum-minimum standardization processing on the fault sample set; calculating the similarity between samples in the fault sample set based on a similarity algorithm; connecting the first sample with the highest similarity based on a K-nearest neighbor criterion; and constructing a fault sample graph by taking the samples of the fault sample set as nodes and the connection relationship between the samples of the fault sample set as edges. The diagnostic model establishment module is configured to establish a numerical control machine tool spindle bearing fault diagnosis model based on a topology adaptive graph convolution network, including: calculating a Laplacian matrix of the fault sample graph based on the connection relationship between the nodes of the fault sample graph; constructing a topology adaptive graph convolution layer based on the Laplacian matrix of the fault sample graph; and constructing a numerical control machine tool spindle bearing fault diagnosis model with two layers of topology adaptive graph convolution layers based on the topology adaptive graph convolution layer. The evaluation module is configured to train the numerical control machine tool spindle bearing fault diagnosis model and evaluate the diagnostic results of the numerical control machine tool spindle bearing fault diagnosis model.

Images