A smart spindle service state diagnosis method, system, device and medium

Abstract

The application provides a kind of intelligent spindle service state diagnostic method, system, equipment and medium, including building data acquisition platform to realize the data acquisition of intelligent spindle, the data enhancement of original signal is carried out by the method of overlapping sampling, and the original signal is converted into two-dimensional time-frequency image by CWT, and the original signal after data enhancement is retained as the input of one-dimensional channel;The original signal after data enhancement and time-frequency pattern sample are respectively divided into training set, test set and verification set according to 7:2:1 interval;The training set is input into the improved double-channel DenseNet model for training;The verification set is input into the improved model to carry out hyperparameter optimization by bayesian optimization;The test set is input into the model that has been trained, and the final state diagnosis result of intelligent spindle is obtained;The application collects the data of intelligent spindle, analyzes the collected signal, extracts more detailed features, and provides a basis for subsequent spindle performance evaluation.

Description

Technical Field

[0001] This invention belongs to the field of intelligent manufacturing technology, specifically relating to a method, system, equipment, and medium for diagnosing the service status of an intelligent spindle. Background Technology

[0002] Intelligent manufacturing, as an important new market in the manufacturing industry, relies on the core competitiveness of enterprises in this field, including the digitization and networking of equipment, the analysis of equipment processing data, and the construction of intelligent workshops. CNC machine tools are crucial equipment in the workshop, and the spindle, as a core component of CNC machine tools, is particularly important for achieving intelligent operation. To meet the processing requirements of modern manufacturing and realize the intelligent operation of the spindle system, the primary task is to solve the spindle sensing problem. To improve the accuracy of system sensing, considering the randomness and singularity of a single physical field, multi-sensor information fusion technology can be used to achieve mutual supplementation of data, increase the diversity and complexity of information, effectively reduce the uncertainty brought by single-sensor information, and improve the stability and reliability of data. By adding sensors such as acceleration, displacement, and temperature to the spindle system, and collecting multi-physical field information such as vibration, displacement, and temperature fields, this information can be used to determine the processing status and health of the intelligent spindle, and to diagnose faults in case of malfunctions. This is of great significance for improving the processing efficiency of intelligent spindles.

[0003] In actual machining processes, factors such as spindle vibration, temperature rise, and localized cracks can affect machining accuracy and efficiency. Therefore, improving the performance of the spindle system is particularly important. Many modern machining processes are not performed independently but are interdependent. In this complex machining environment, if the spindle system fails to meet the actual machining requirements, it will directly affect the next step of production, ultimately leading to a decrease in product production efficiency and machining quality. Existing technologies, such as expert experience, time-frequency domain analysis, and wavelet analysis, are increasingly difficult to achieve autonomous spindle sensing. Traditional machine learning methods, such as BP neural networks, SVM, and Hidden Markov Models, are limited by their shallow structure and cannot extract feature information that can characterize the spindle's operating status from the massive and complex measured signal data. Summary of the Invention

[0004] To address the problems existing in the prior art, this invention provides an intelligent spindle service status diagnosis method, system, device, and medium, which can extract more complete detailed features of the spindle status, providing an effective basis for subsequent fault mode identification and classification.

[0005] This invention is achieved through the following technical solution:

[0006] A method for diagnosing the service status of an intelligent spindle, characterized by comprising the following steps:

[0007] a. Build a data acquisition platform to acquire data from the intelligent spindle, enhance the original signal by overlay sampling, convert the original signal into a two-dimensional time-frequency image by CWT, and retain the enhanced original signal as the input of the one-dimensional channel.

[0008] b. Divide the data-enhanced original signal and the time-frequency graph samples into training set, test set and validation set according to the interval ratio of 7:2:1;

[0009] c. Input the training set into the improved dual-channel DenseNet model for training;

[0010] d. Input the validation set into the improved model and perform hyperparameter optimization using Bayesian optimization;

[0011] e. Input the test set into the already trained model to obtain the final state diagnosis result of the intelligent spindle.

[0012] Further, step c involves inputting the training set into the improved dual-channel DenseNet model for training; the improved dual-channel DenseNet model based on feature fusion consists of 2 input layers, 2 convolutional layers and max pooling layers, 2 three-level dense connection blocks, 2 dense connection blocks with ECA, 2 three transition layers, 2 BN layers + Conv layers + Maxpooling layers, 2 LSTM layers, 2 flattening layers, one concatenation channel merging layer, one fully connected layer, and an output layer.

[0013] Furthermore, step d involves inputting the validation set into the improved model and performing hyperparameter optimization through Bayesian optimization. The hyperparameters are optimized within a given range, and the main hyperparameters to be optimized include the learning rate, batch size, number of training epochs, number and size of convolutional kernels, and number of neurons in the fully connected layer.

[0014] Furthermore, step e inputs the test set into the already trained model to obtain the final state diagnosis result of the intelligent spindle;

[0015]

[0016] Where TP represents true positive samples; TN represents true negative samples; FP represents false positive samples; and FN represents false negative samples.

[0017] Furthermore, the improvement process of the dual-channel DenseNet model is as follows:

[0018] The extraction of local and global features of the original signal is accomplished by combining the improved DenseNet network with the LSTM network.

[0019] The network parameters are tuned and overfitting is suppressed by using a fully connected layer and a dropout layer;

[0020] The output features are normalized using the Softmax function, a normalization exponential function, converting all output values ​​into probabilities. The sum of all probability values ​​is 1. The formula for the Softmax function is as follows:

[0021]

[0022] Where j = 1, ..., K, and K refers to the number of specific categories.

[0023] Furthermore, an ECA attention mechanism is added to the last densely connected block of the 2-3 level densely connected blocks; ECA, based on the SE module, changes the method of learning channel attention information using fully connected layers (FC) in SE to learning channel attention information using 1*1 convolutions. Specific steps include:

[0024] S1: First, input the feature map, whose dimensions are H*W*C;

[0025] S2: Perform spatial feature compression on the input feature map. In the spatial dimension, use global average pooling (GAP) to obtain a 1*1*C feature map.

[0026] S3: Perform channel feature learning on the compressed feature map to achieve the following: learn the importance between different channels through 1*1 convolution, and the output dimension is still 1*1*C.

[0027] S4: Channel attention integration, combining the features of channel attention Figure 1 The *1*C and the original input feature map H*W*C are multiplied channel by channel to output a feature map with channel attention.

[0028] Furthermore, the Adam optimizer is selected to optimize the model. The Adam optimizer optimizes the model as follows:

[0029]

[0030] Where M is the number of categories; y ic The sign function is 0 or 1; it takes 1 if the true class of sample i is equal to C, and 0 otherwise. ic Let be the predicted probability that observed sample i belongs to category c.

[0031] A smart spindle service condition diagnostic system, comprising:

[0032] The acquisition module is used to build a data acquisition platform to acquire data from the intelligent spindle. It enhances the original signal by using the overlap sampling method, converts the original signal into a two-dimensional time-frequency image by CWT, and retains the enhanced original signal as the input of the one-dimensional channel.

[0033] The partitioning module is used to divide the data-augmented original signal and the time-frequency plot samples into training set, test set and validation set according to the interval ratio of 7:2:1, respectively.

[0034] The training module is used to input the training set into the improved dual-channel DenseNet model for training.

[0035] The optimization module is used to input the validation set into the improved model and perform hyperparameter optimization through Bayesian optimization.

[0036] The diagnostic module is used to input the test set into the trained model to obtain the final state diagnostic results of the intelligent spindle.

[0037] A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of an intelligent spindle service condition diagnosis method.

[0038] A computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of a method for diagnosing the service status of an intelligent spindle.

[0039] Compared with the prior art, the present invention has the following beneficial technical effects:

[0040] This invention provides a method, system, device, and medium for diagnosing the service status of an intelligent spindle. The method includes building a data acquisition platform to collect data from the intelligent spindle; performing data augmentation on the original signal using overlapping sampling; converting the original signal into a two-dimensional time-frequency image using CWT; retaining the augmented original signal as input to a one-dimensional channel; dividing the augmented original signal and the time-frequency image samples into training, testing, and validation sets according to a 7:2:1 interval; inputting the training set into an improved dual-channel DenseNet model for training; inputting the validation set into the improved model for hyperparameter optimization using Bayesian optimization; and inputting the testing set into the already trained model to obtain the final diagnostic result of the intelligent spindle's status. This invention, through the acquisition and analysis of intelligent spindle data, and by combining an improved dense connection network with a long short-term memory network, extracts more detailed features based on a dual-channel fusion approach, providing a basis for subsequent spindle performance evaluation. Attached Figure Description

[0041] Figure 1 This is a flowchart of a smart spindle service status diagnosis method according to the present invention;

[0042] Figure 2 This is a flowchart of the spindle data acquisition and control process of the present invention;

[0043] Figure 3 This is a diagram of the densely connected network structure of the present invention;

[0044] Figure 4 This is a structural diagram of the improved 2D-DenseNet model of the present invention;

[0045] Figure 5 This is a structural diagram of the dual-channel model of the present invention;

[0046] Figure 6 This is a structural diagram of the ECA attention mechanism of the present invention. Detailed Implementation

[0047] The present invention will be further described in detail below with reference to specific embodiments. These descriptions are for explanation purposes only and are not intended to limit the scope of the invention.

[0048] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0049] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0050] This invention provides a method for diagnosing the service status of an intelligent spindle, such as... Figure 1 As shown, it includes the following steps:

[0051] a. Build a data acquisition platform to acquire data from the intelligent spindle, enhance the original signal by overlay sampling, convert the original signal into a two-dimensional time-frequency image by CWT, and retain the enhanced original signal as the input of the one-dimensional channel.

[0052] b. Divide the data-enhanced original signal and the time-frequency graph samples into training set, test set and validation set according to the interval ratio of 7:2:1;

[0053] c. Input the training set into the improved dual-channel DenseNet model for training;

[0054] d. Input the validation set into the improved model and perform hyperparameter optimization through Bayesian optimization. The hyperparameters of the model are optimized through global optimization, which plays a role in fine-tuning the model.

[0055] e. Input the test set into the already trained model to obtain the final state diagnosis result of the intelligent spindle.

[0056] It should be noted that in step a, the vibration and displacement signals of the spindle are acquired using accelerometers and displacement sensors; filters and signal amplifiers are used to perform A / D conversion on the acquired signals; the processed data is stored and transmitted using a data acquisition card; the received raw signal is augmented using an overlap sampling method to preserve the original signal, and time-frequency domain analysis is performed using CWT; in step b, the augmented raw signal and the time-frequency graph samples are divided into training set, validation set, and test set according to a 7:2:1 interval; the augmented signal is analyzed in the time domain, frequency domain, and time-frequency domain, where the time-frequency graph of the original signal obtained through continuous wavelet transform (CWT) can clearly and accurately represent the time-frequency distribution of the vibration.

[0057] It should be further explained that this application first uses a densely connected network to adaptively extract feature information and reduce dimensionality, then constructs a batch normalization layer, a convolutional layer and a max pooling layer to further extract deep features as input to the LSTM layer, and finally uses a Long Short-Term Memory (LSTM) network to extract global features. In this invention, an improved DenseNet network and an LSTM network are used to extract local and global features, which can avoid the phenomenon that the training parameters increase rapidly while the model is deepened, resulting in a significant slowdown in training speed. Fewer LSTM units are used to increase modeling ability, and the extraction of intelligent spindle fault features is more comprehensive.

[0058] Preferred, such as Figure 5As shown, step c involves inputting the training set into the improved dual-channel DenseNet model for training. The improved dual-channel DenseNet model based on feature fusion consists of 2 input layers, 2 convolutional layers and max pooling layers, 2 three-level dense connection blocks, 2 dense connection blocks with ECA, 2 three transition layers, 2 BN layers + Conv layers + Maxpooling layers, 2 LSTM layers, 2 flattening layers, one concatenation channel merging layer, one fully connected layer, and an output layer.

[0059] like Figure 3 The diagram shows the dense connection network structure of this invention. M1: A three-level dense connection method is adopted, and the dense connection mainly consists of two parts: dense block + transition layer; M2: Improved dense layer: BN, ReLU and 1x1Conv, BN, ReLU and 3x3Conv; M3: Improved transition layer: BN, ReLU, 1x1Conv, average pool; M4: An ECA attention module is added after the last level dense block.

[0060] Preferably, step d involves inputting the validation set into the improved model and performing hyperparameter optimization using Bayesian optimization. The hyperparameters are optimized within a given range, and the main hyperparameters to be optimized include learning rate, batch size, number of training epochs, number and size of convolutional kernels, number of neurons in fully connected layers, etc. Bayesian optimization algorithm has many advantages, such as fewer iterations, faster convergence speed, and strong robustness, especially for non-convex problems.

[0061] Preferably, step e inputs the test set into the already trained model to obtain the final state diagnosis result of the intelligent spindle;

[0062]

[0063] Where TP represents true positive samples; TN represents true negative samples; FP represents false positive samples; and FN represents false negative samples.

[0064] like Figure 4 As shown, the improvement process of the dual-channel DenseNet model is as follows:

[0065] The extraction of local and global features of the original signal is accomplished by combining the improved DenseNet network with the LSTM network.

[0066] The network uses a fully connected layer and a dropout layer to adjust parameters and suppress overfitting. The dropout layer is typically added to the fully connected layer to prevent overfitting and improve the model's generalization ability. Dropout is only used during model training and not during model evaluation. The output features are normalized using the Softmax function, a normalization exponential function, converting all output values ​​into probabilities (between 0 and 1). All probability values ​​are summed to 1. The formula for the Softmax function is:

[0067]

[0068] Where j = 1, ..., K, and K refers to the number of specific categories.

[0069] Dimensionality reduction is achieved by flattening multidimensional data into one-dimensional data using the Flatten() function in the flattening layer. A fully connected layer and a dropout layer are used to adjust network parameters and suppress overfitting. The dropout layer is typically added to the fully connected layer to prevent overfitting and improve the model's generalization ability. Dropout is only used during model training; it is not needed during model evaluation. During model evaluation, the dropout layer allows all activation units to pass through.

[0070] like Figure 6 As shown, an ECA attention mechanism is added to the last densely connected block of the 2-3 level densely connected blocks; ECA, based on the SE module, changes the use of fully connected layers (FC) to learn channel attention information in SE to 1*1 convolutions to learn channel attention information. The specific steps include:

[0071] S1: First, input the feature map, whose dimensions are H*W*C;

[0072] S2: Perform spatial feature compression on the input feature map. In the spatial dimension, use global average pooling (GAP) to obtain a 1*1*C feature map.

[0073] S3: Perform channel feature learning on the compressed feature map to achieve the following: learn the importance between different channels through 1*1 convolution, and the output dimension is still 1*1*C.

[0074] S4: Channel attention integration, combining the features of channel attention Figure 1 The *1*C and the original input feature map H*W*C are multiplied channel by channel to output a feature map with channel attention.

[0075] Preferably, the Adam optimizer is selected to optimize the model. The Adam optimizer optimizes the model as follows:

[0076]

[0077] Where M is the number of categories; y ic The sign function is 0 or 1; it takes 1 if the true class of sample i is equal to C, and 0 otherwise. ic Let be the predicted probability that observed sample i belongs to category c.

[0078] This invention provides an intelligent spindle service condition diagnostic system, comprising:

[0079] The acquisition module is used to build a data acquisition platform to acquire data from the intelligent spindle. It enhances the original signal by using the overlap sampling method, converts the original signal into a two-dimensional time-frequency image by CWT, and retains the enhanced original signal as the input of the one-dimensional channel.

[0080] The partitioning module is used to divide the data-augmented original signal and the time-frequency plot samples into training set, test set and validation set according to the interval ratio of 7:2:1, respectively.

[0081] The training module is used to input the training set into the improved dual-channel DenseNet model for training.

[0082] The optimization module is used to input the validation set into the improved model and perform hyperparameter optimization through Bayesian optimization.

[0083] The diagnostic module is used to input the test set into the trained model to obtain the final state diagnostic results of the intelligent spindle.

[0084] In another embodiment of the present invention, such as Figure 2 As shown, a hardware platform for a smart spindle service status diagnosis method includes: a triaxial accelerometer, a thermocouple temperature sensor, a displacement sensor, a data acquisition card, a signal conditioning and storage device, a control module, and a piezoelectric actuator. This application first acquires the raw signals transmitted from the accelerometer, thermocouple temperature sensor, and displacement sensor; performs signal preprocessing such as noise reduction and trend removal through filters and amplifiers; then enhances the received raw signals using overlapping sampling to preserve the original signals, and performs time-frequency domain analysis using CWT to extract the time-frequency features of the signals, which can better characterize the spindle's operating status; the model extracts signal features through feature fusion; finally, the softmax function is used to identify and classify fault modes.

[0085] In another embodiment of the present invention, a computer device is provided, comprising a processor and a memory. The memory stores a computer program, which includes program instructions. The processor executes the program instructions stored in the computer storage medium. The processor may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. It is the computing and control core of the terminal, suitable for implementing one or more instructions, specifically suitable for loading and executing one or more instructions in the computer storage medium to achieve a corresponding method flow or corresponding function. The processor described in this embodiment of the present invention can be used in the operation of an intelligent spindle service status diagnosis method.

[0086] In another embodiment of the present invention, a storage medium is provided, specifically a computer-readable storage medium (Memory), which is a memory device in a computer device used to store programs and data. It is understood that the computer-readable storage medium here can include both the built-in storage medium in the computer device and extended storage media supported by the computer device. The computer-readable storage medium provides storage space that stores the terminal's operating system. Furthermore, the storage space also stores one or more instructions suitable for loading and execution by a processor. These instructions can be one or more computer programs (including program code). It should be noted that the computer-readable storage medium here can be a high-speed RAM memory or a non-volatile memory, such as at least one disk storage device. The processor can load and execute one or more instructions stored in the computer-readable storage medium to implement the corresponding steps of the intelligent spindle service status diagnosis method in the above embodiments.

[0087] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied 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.

[0088] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0089] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0090] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0091] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for diagnosing the service status of an intelligent spindle, characterized in that, Includes the following steps: a. Build a data acquisition platform to acquire data from the intelligent spindle, enhance the original signal by overlay sampling, convert the original signal into a two-dimensional time-frequency image by CWT, and retain the enhanced original signal as the input of the one-dimensional channel. b. Divide the data-enhanced original signal and the time-frequency graph samples into training set, test set and validation set according to the interval ratio of 7:2:1; c. Input the training set into the improved dual-channel DenseNet model for training; d. Input the validation set into the improved model and perform hyperparameter optimization using Bayesian optimization; e. Input the test set into the trained model to obtain the final state diagnosis result of the intelligent spindle; Step c involves inputting the training set into the improved dual-channel DenseNet model for training. The improved dual-channel DenseNet model based on feature fusion consists of 2 input layers, 2 convolutional layers and max pooling layers, 2 three-level dense connected blocks, 2 dense connected blocks with ECA, 2 three transition layers, 2 BN layers + Conv layers + Maxpooling layers, 2 LSTM layers, 2 flattening layers, one concatenation channel merging layer, one fully connected layer, and an output layer. Step d involves inputting the validation set into the improved model and performing hyperparameter optimization using Bayesian optimization. The hyperparameters are optimized within a given range, and the main hyperparameters to be optimized include the learning rate, batch size, number of training epochs, number and size of convolutional kernels, and number of neurons in the fully connected layer. An ECA attention mechanism is added to the last densely connected block of the 2-3 level densely connected blocks. ECA, based on the SE module, replaces the use of fully connected layers (FC) to learn channel attention information in SE with 1*1 convolutions to learn channel attention information. Specific steps include: S1: First, input the feature map, whose dimensions are H*W*C; S2: Perform spatial feature compression on the input feature map. In the spatial dimension, use global average pooling (GAP) to obtain a 1*1*C feature map. S3: Perform channel feature learning on the compressed feature map to achieve the following: learn the importance between different channels through 1*1 convolution, and the output dimension is still 1*1*C. S4: Channel attention combination, multiply the channel attention feature map 1*1*C and the original input feature map H*W*C channel by channel to output a feature map with channel attention; The Adam optimizer is selected to optimize the model. The Adam optimizer optimizes the model as follows: ； Where M is the number of categories; y ic The sign function is 0 or 1; it takes 1 if the true class of sample i is equal to C, otherwise it takes 0. pic is the predicted probability that observed sample i belongs to category c.

2. The intelligent spindle service status diagnosis method according to claim 1, characterized in that, Step e inputs the test set into the already trained model to obtain the final state diagnosis result of the intelligent spindle. ； Where TP represents true positive samples; TN represents true negative samples; FP represents false positive samples; and FN represents false negative samples.

3. The intelligent spindle service status diagnosis method according to claim 1, characterized in that, The improvement process of the dual-channel DenseNet model is as follows: The extraction of local and global features of the original signal is accomplished by combining the improved DenseNet network with the LSTM network. The network parameters are tuned and overfitting is suppressed by using a fully connected layer and a dropout layer; The output features are normalized using the Softmax function, a normalization exponential function, converting all output values ​​into probabilities. The sum of all probability values ​​is 1. The formula for the Softmax function is as follows: ； Where j=1,......,K, and K refers to the number of categories in a specific classification.

4. A smart spindle service condition diagnosis system that implements the smart spindle service condition diagnosis method according to any one of claims 1-3, characterized in that, include: The acquisition module is used to build a data acquisition platform to acquire data from the intelligent spindle. It enhances the original signal by using the overlap sampling method, converts the original signal into a two-dimensional time-frequency image by CWT, and retains the enhanced original signal as the input of the one-dimensional channel. The partitioning module is used to divide the data-augmented original signal and the time-frequency plot samples into training set, test set and validation set according to the interval ratio of 7:2:1, respectively. The training module is used to input the training set into the improved dual-channel DenseNet model for training. The optimization module is used to input the validation set into the improved model and perform hyperparameter optimization through Bayesian optimization. The diagnostic module is used to input the test set into the trained model to obtain the final state diagnostic results of the intelligent spindle.

5. A computer 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 computer program, it implements the steps of the intelligent spindle service status diagnosis method as described in any one of claims 1-3.

6. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the intelligent spindle service status diagnosis method as described in any one of claims 1-3.

Images