Five-axis machine tool spindle state analysis method and system based on multi-dimensional data

Abstract

The invention belongs to the technical field of machine tool safety detection, and particularly relates to a five-axis machine tool spindle state analysis method and system based on multi-dimensional data. The method comprises the following steps: collecting state data of a main shaft in real time through various sensors; performing feature extraction on the preprocessed state data to obtain multi-dimensional feature data; determining the health state of the main shaft through a machine learning model based on the multi-dimensional feature data; influence data generated by the spindle health state on workpiece forming is predicted according to the empirical data, and the defect type generated in the workpiece forming process is determined through a regression model on the basis of the influence data; generating an adjustment strategy according to the defect type; and generating an adjusting instruction according to the adjusting strategy, and processing the workpiece according to the adjusting instruction. The spindle health state can be comprehensively evaluated, workpiece forming defects caused by spindle faults are avoided, the workpiece machining quality is improved, and the downtime and the production loss are reduced.

Description

Technical Field

[0001] The present application belongs to the technical field of machine tool safety detection, and in particular relates to a five-axis machine tool spindle status analysis method and system based on multi-dimensional data. Background Art

[0002] With the development of technology, the demand for ultra-precision machining of parts with complex shapes is increasing, and increasing the number of machine tool axes has become a development trend of ultra-precision machine tools today. If ultra-precision cutting of complex parts such as off-axis free-form surfaces and non-rotationally symmetrical surfaces is required, a multi-axis ultra-precision machine tool is required. A five-axis machine tool is a high-precision machining equipment with five motion axes that can simultaneously control the movement of the tool on the linear and rotary axes, thereby achieving high-precision machining of complex workpieces. As the core component of a five-axis machine tool, the state of the spindle directly affects the machining efficiency and quality. A high-performance spindle can provide stable speed, power and torque to ensure the stability and accuracy of the machining process. If the spindle fails or its performance deteriorates, it will lead to reduced machining efficiency, reduced machining quality, and even damage to the workpiece or tool.

[0003] There are usually two existing spindle status monitoring methods. One is to collect spindle operation data through sensors and set thresholds or rules to judge the spindle status. Although this method is simple to implement, it cannot effectively capture the influence of the coupling effect between multiple factors on the spindle status. The other is to achieve intelligent monitoring of the spindle status through training models. This method can handle more complex multi-dimensional feature relationships, but it has a strong dependence on high-quality labeled data. In addition, the above two spindle status monitoring methods cannot accurately predict the influence of the spindle status on the subsequent workpiece forming. This limitation makes it difficult for existing technologies to provide forward-looking guidance for process optimization and preventive maintenance, which restricts the improvement of the level of intelligent manufacturing.

[0004] Therefore, it is desired to have a technical solution to overcome or alleviate at least one of the above-mentioned defects of the prior art. Summary of the invention

[0005] The purpose of the present application is to provide a five-axis machine tool spindle state analysis method and system based on multi-dimensional data to solve at least one problem existing in the prior art.

[0006] The technical solution of this application is: A first aspect of the present application provides a five-axis machine tool spindle state analysis method based on multidimensional data, comprising: S100, collects spindle status data in real time through multiple sensors; S200, extracting features from the preprocessed state data to obtain multi-dimensional feature data; S300, determining the health status of the spindle through a machine learning model based on the multi-dimensional feature data; S400, predicting the influence data of the spindle health status on the workpiece forming according to the empirical data, and determining the defect type occurring in the workpiece forming process through a regression model based on the influence data; S500, generating an adjustment strategy according to the defect type; S600: Generate an adjustment instruction according to the adjustment strategy, and process the workpiece according to the adjustment instruction.

[0007] In at least one embodiment of the present application, the state data includes vibration data, temperature data, torque data, and rotation speed data.

[0008] In at least one embodiment of the present application, in S200, feature extraction is performed on the preprocessed state data to obtain multi-dimensional feature data, including: S210, sorting the state data according to timestamps, and synchronizing the data by interpolation; S220, preprocessing the state data, wherein the preprocessing method at least includes cleaning, denoising, and filtering; S230, extracting multidimensional feature data related to the health status of the spindle from the status data; S240: Fusing the multi-dimensional feature data by using a principal component analysis method.

[0009] In at least one embodiment of the present application, in S230, extracting multidimensional feature data related to the spindle health status from the status data includes: Extracting vibration characteristic data related to the health status of the spindle from the vibration data, including vibration amplitude and vibration frequency; Extracting temperature characteristic data related to the health status of the spindle from the temperature data, including a temperature change rate and a temperature change trend; Extracting torque characteristic data related to the health status of the spindle from the torque data, including load fluctuation amplitude and load fluctuation frequency; Speed ​​characteristic data related to the spindle health status is extracted from the speed data, including the speed fluctuation amplitude.

[0010] In at least one embodiment of the present application, in S240, the multi-dimensional feature data is fused by a principal component analysis method, including: S241, performing standardization processing on the multi-dimensional feature data, including: Calculate the mean of multidimensional feature data ; Calculate the standard deviation of multidimensional feature data ; Standardize multidimensional feature data by mean and standard deviation: ; in, For the data set i Multi-dimensional feature data, is the mean of the multidimensional feature data, is the standard deviation of the multidimensional feature data, For the data set i Data points obtained by standardization of multi-dimensional feature data; S242, construct a covariance matrix, solve the eigenvalues ​​and eigenvectors, including: Construct the covariance matrix: ; in, Z For data points The multidimensional feature matrix composed of n is the number of samples, for Z The transposed matrix of is the covariance matrix; Construct the characteristic polynomial from the covariance matrix: ; Where I is the identity matrix, is the characteristic polynomial; By solving the characteristic polynomial , get all the eigenvalues ​​of the covariance matrix , m is the total number of eigenvalues; For each eigenvalue , by solving the system of equations , and obtain the corresponding feature vector , k =1, 2, 3, ..., m ; S243, sorting the eigenvalues ​​from large to small, and selecting a plurality of principal components preset previously; S244: Project the multidimensional feature data onto a plurality of selected principal components to obtain multidimensional feature data after dimensionality reduction.

[0011] In at least one embodiment of the present application, in S300, determining the spindle health state through a machine learning model based on the multi-dimensional feature data includes: S310, dividing the multidimensional feature data into a training set, a validation set, and a test set; S320, using the deep learning framework Keras to build a fully connected neural network model; S330, training the fully connected neural network model through the training set, performing model selection and hyperparameter tuning on the fully connected neural network model through the validation set, and evaluating the fully connected neural network model through the test set; S340. Determine the health status of the spindle through the fully connected neural network model, where the types of the health status of the spindle include healthy, slightly abnormal, and seriously abnormal.

[0012] In at least one embodiment of the present application, in S400, predicting the influence data of the spindle health status on the workpiece forming according to the empirical data, and determining the defect type occurring in the workpiece forming process through a regression model based on the influence data, includes: S410, predicting the influence data of the spindle health status on workpiece forming according to the empirical data, wherein the influence data includes a health status score and a plurality of processing parameters; S420, inputting the health status score and the multiple processing parameters into a regression model to calculate the workpiece quality: ; Where Y is the mass of the workpiece, is the intercept, is the corresponding regression coefficient, is a health status score or a processing parameter; S430, determining the type of defects occurring during the workpiece forming process according to the workpiece quality.

[0013] In at least one embodiment of the present application, the processing parameters include at least cutting speed, feed rate, cutting depth, tool wear, and coolant status.

[0014] In at least one embodiment of the present application, in S500, the adjustment strategy generated according to the defect type includes: Adjust the spindle speed or feed speed; Adjust the spindle load; Optimize the cooling system or replace the lubricating oil; Replace or repair faulty parts.

[0015] A second aspect of the present application provides a five-axis machine tool spindle state analysis system based on multidimensional data, based on the five-axis machine tool spindle state analysis method based on multidimensional data as described above, comprising: Data acquisition module, used to collect the status data of the spindle in real time through a variety of sensors; A data processing module, used for extracting features from the preprocessed state data to obtain multi-dimensional feature data; A state analysis module, used to determine the health state of the spindle through a machine learning model based on the multi-dimensional feature data; A defect prediction module is used to predict the influence data of the spindle health status on the workpiece forming according to the empirical data, and determine the defect type occurring in the workpiece forming process through a regression model based on the influence data; An adjustment module, used for generating an adjustment strategy according to the defect type; The output module is used to generate an adjustment instruction according to the adjustment strategy and process the workpiece according to the adjustment instruction.

[0016] The invention has at least the following beneficial technical effects: The five-axis machine tool spindle status analysis method based on multi-dimensional data of the present application can comprehensively evaluate the spindle health status and discover potential problems in advance by real-time monitoring and analysis of multi-dimensional feature data through a machine learning model; predict the impact of the spindle health status on subsequent workpiece forming based on empirical data, and identify the defect type in combination with a regression model to avoid workpiece forming defects caused by spindle failure; avoid machining errors caused by spindle abnormalities through early adjustment, improve workpiece machining quality, and reduce downtime and production losses; and help reduce the spindle workload, reduce spindle failures, and extend the spindle service life through real-time monitoring and intelligent adjustment. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a flow chart of a five-axis machine tool spindle state analysis method based on multidimensional data according to an embodiment of the present application; Figure 2 It is a schematic diagram of a five-axis machine tool spindle status analysis system based on multidimensional data according to an embodiment of the present application. DETAILED DESCRIPTION

[0018] In order to make the purpose, technical scheme and advantages of the implementation of this application clearer, the technical scheme in the embodiment of this application will be described in more detail below in conjunction with the drawings in the embodiment of this application. In the drawings, the same or similar reference numerals throughout represent the same or similar elements or elements with the same or similar functions. The described embodiments are part of the embodiments of this application, not all of them. The embodiments described below with reference to the drawings are exemplary and are intended to be used to explain this application, and should not be construed as limitations on this application. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application. The embodiments of this application are described in detail below in conjunction with the drawings.

[0019] In the description of the present application, it should be understood that the terms "center", "longitudinal", "lateral", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", etc., indicating orientations or positional relationships, are based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as limiting the scope of protection of the present application.

[0020] The following is combined with Figure 1 to Figure 2 This application is described in further detail.

[0021] The first aspect of the present application provides a five-axis machine tool spindle state analysis method based on multidimensional data, such as Figure 1 As shown, the following steps are included: S100, collects spindle status data in real time through multiple sensors; S200, extracting features from the preprocessed state data to obtain multi-dimensional feature data; S300, based on multi-dimensional feature data, determine the health status of the spindle through a machine learning model; S400, predicting the influence data of the spindle health state on the workpiece forming according to the empirical data, and determining the defect type occurring in the workpiece forming process through a regression model based on the influence data; S500, generating an adjustment strategy according to the defect type; S600 , generating an adjustment instruction according to the adjustment strategy, and processing the workpiece according to the adjustment instruction.

[0022] The five-axis machine tool spindle state analysis method based on multi-dimensional data of the present application, first, in S100, the spindle state data such as vibration data, temperature data, torque data and speed data are collected in real time through a variety of sensors. Among them, the vibration sensor can monitor the vibration amplitude and frequency of the spindle to determine whether there is an abnormal vibration mode; the temperature sensor can monitor the temperature of the spindle to prevent damage caused by overheating; the torque sensor can measure the torque of the spindle to analyze the changes in the spindle load; the speed sensor is used to monitor the speed of the spindle in real time.

[0023] In the five-axis machine tool spindle state analysis method based on multidimensional data of the present application, in S200, the collected state data is preprocessed and feature extracted to extract key features related to the health status of the spindle. Different sensors usually collect different time series data. In order to ensure the time consistency between the multidimensional feature data, these data need to be time synchronized, and the data collected by each sensor is accompanied by a timestamp. In the preferred embodiment of the present application, in S200, feature extraction is performed on the preprocessed state data to obtain multidimensional feature data, including: S210, sorting the status data according to timestamps, and synchronizing the data by interpolation; The status data of all sensors are sorted by timestamp, and all data sequences are aligned at a uniform time step through interpolation methods. During the data synchronization process, if data at certain time points is missing (such as a sensor failure or loss), data filling (such as interpolation methods based on surrounding valid data) is required to repair the missing data to ensure data continuity.

[0024] S220, preprocessing the state data, where the preprocessing method at least includes cleaning, denoising, and filtering; By preprocessing the synchronized status data, noise and unnecessary high-frequency interference are removed, thus improving the accuracy of subsequent analysis.

[0025] S230, extracting multidimensional feature data related to the health status of the spindle from the status data; In this embodiment, the multi-dimensional feature data extraction mainly includes vibration feature extraction, temperature feature extraction, torque feature extraction, and speed feature extraction. Specifically: Vibration feature extraction: Extract vibration feature data related to the health status of the spindle from the vibration data, including vibration amplitude and vibration frequency; Temperature feature extraction: Extract temperature feature data related to the health status of the spindle from the temperature data, including the temperature change rate and temperature change trend; Torque feature extraction: Extract torque feature data related to the health status of the spindle from the torque data, including load fluctuation amplitude and load fluctuation frequency; Speed ​​feature extraction: Extract speed feature data related to the health status of the spindle from the speed data, including the speed fluctuation amplitude.

[0026] Among them, regarding vibration feature extraction, the extracted vibration signal features can not only extract vibration amplitude and vibration frequency, but also statistical features and kurtosis in time domain and frequency domain. Time domain features are to directly extract some statistical features from the original vibration data, such as mean, variance, skewness, kurtosis, etc., which reflect the volatility and asymmetry of the vibration signal and can identify irregular vibrations in the system; frequency domain features are to convert the vibration signal from time domain to frequency domain through Fourier transform, and analyze the spectral characteristics of vibration, including main frequency, frequency bandwidth and power spectrum density; kurtosis is to calculate the sharpness of the vibration signal, and can detect abnormal fluctuations in the operation of the spindle. Regarding temperature feature extraction, the key temperature features include the temperature change rate and the long-term temperature change trend. The rate of change of the spindle temperature reflects the change of the spindle load. The long-term trend of the temperature is monitored. If the temperature continues to rise, it may indicate that there are hidden dangers in the spindle. Regarding torque feature extraction, the amplitude and frequency of load fluctuations can be analyzed to extract the load fluctuation amplitude and load fluctuation frequency features. A large load fluctuation amplitude may mean problems such as tool wear, uneven workpiece material, or unstable fixtures. For the frequency component of load fluctuations, if an abnormal frequency occurs, it may indicate that the spindle has a fault or external interference. Regarding speed feature extraction, the speed feature mainly includes the speed fluctuation amplitude. A large speed fluctuation may indicate a fault in the spindle drive system.

[0027] S240, integrating the multi-dimensional feature data through principal component analysis method.

[0028] In this embodiment, the multi-dimensional feature data fusion process is as follows: S241, performing standardization processing on the multi-dimensional feature data, including: Calculate the mean of multidimensional feature data : ; in, N is the number of multi-dimensional feature data; Calculate the standard deviation of multidimensional feature data : ; Standardize multidimensional feature data by mean and standard deviation: ; in, For the data set i Multi-dimensional feature data, is the mean of the multidimensional feature data, is the standard deviation of the multidimensional feature data, For the data set i Data points obtained by standardization of multi-dimensional feature data; Standardized data points The mean of is 0 and the variance is 1; S242, construct a covariance matrix, solve the eigenvalues ​​and eigenvectors, including: Construct the covariance matrix: ; in, Z For data points The multidimensional feature matrix composed of n is the number of samples, each sample for Z The transposed matrix of is the covariance matrix, which represents the covariance between each feature; Construct the characteristic polynomial from the covariance matrix: ; Where I is the identity matrix, is the characteristic polynomial; By solving the characteristic polynomial , get all the eigenvalues ​​of the covariance matrix , m is the total number of eigenvalues; For each eigenvalue , by solving the system of equations , and obtain the corresponding feature vector , k =1, 2, 3, ..., m ; S243, sorting the eigenvalues ​​from large to small, and selecting a plurality of principal components preset previously; S244, projecting the multidimensional feature data onto a plurality of selected principal components to obtain multidimensional feature data after dimensionality reduction.

[0029] In the five-axis machine tool spindle state analysis method based on multi-dimensional data of the present application, in S300, based on the multi-dimensional feature data obtained above, the spindle health state is determined by a machine learning model, and the specific process includes: S310, dividing the multidimensional feature data into a training set, a validation set, and a test set; S320, using the deep learning framework Keras to build a fully connected neural network model; S330, training the fully connected neural network model through the training set, performing model selection and hyperparameter tuning on the fully connected neural network model through the validation set, and evaluating the fully connected neural network model through the test set; S340. Determine the health status of the spindle through a fully connected neural network model, where the types of health status of the spindle include healthy, slightly abnormal, and seriously abnormal.

[0030] In a preferred embodiment of the present application, the multidimensional feature data is divided into a training set, a validation set and a test set; wherein the training set is used to train the model, accounting for 70% of the total data, the validation set is used for model selection and hyperparameter tuning, accounting for 15% of the total data, and the test set is used to evaluate the performance of the final model, accounting for 15% of the total data. The fully connected neural network model consists of multiple fully connected layers, each layer has a certain number of neurons, and nonlinear transformation is performed through an activation function; wherein the size of the input layer depends on the number of features; according to the complexity of the problem, multiple hidden layers are used, and the hidden layers usually have 128, 256 neurons, or more; the number of neurons in the output layer corresponds to the number of classifications of the health status of the spindle, and in this embodiment, there are 3 classifications: healthy, slightly abnormal, and severely abnormal. The model is trained through the training set, the performance during the training process is monitored to avoid overfitting, the trained model is deployed to the actual system, and the model is updated regularly to maintain high efficiency.

[0031] The five-axis machine tool spindle state analysis method based on multidimensional data of the present application, in S400, based on the current spindle health state, combined with historical experience data and regression model, predicts the impact data of the current spindle health state on workpiece forming, determines whether it will affect the subsequent workpiece forming, and analyzes the types of defects that may occur during the workpiece forming process. In the preferred implementation of the present application, the specific process is: S410, predicting the influence data of the spindle health status on the workpiece forming according to the empirical data, the influence data including the health status score and multiple processing parameters; S420, the workpiece quality is calculated based on the health status score and various processing parameters input into the regression model: ; Where Y is the mass of the workpiece, is the intercept, is the corresponding regression coefficient, is a health status score or a processing parameter; intercept It represents the expected value of workpiece quality Y when the health status score and all processing parameters are 0; regression coefficient Indicates the influence of the health status score or each processing parameter on the workpiece quality Y; the health status score or processing parameter Indicates factors that may affect the quality of the workpiece.

[0032] S430, determining the type of defects occurring during the workpiece forming process according to the workpiece quality.

[0033] Among them, the processing parameters include cutting speed, feed rate, cutting depth, tool wear, coolant status, etc. The health status score and processing parameters related to the possible molding defects of the workpiece are substituted into the regression model corresponding to the workpiece quality, and the workpiece quality is calculated through the regression model. Finally, according to the workpiece quality, the types of defects that may occur during the workpiece molding process are analyzed, such as unqualified surface roughness, excessive dimensional deviation, increased tool wear, etc.

[0034] In the five-axis machine tool spindle state analysis method based on multidimensional data of the present application, if the prediction analysis shows that the defect type determined by the spindle health state may have a negative impact on the subsequent workpiece forming, it is necessary to put forward processing adjustment suggestions. Specifically, in S500, the adjustment strategy generated according to the defect type includes: Adjust the spindle speed or feed speed; Adjust the spindle load; Optimize the cooling system or replace the lubricating oil; Replace or repair faulty parts.

[0035] Among them, overload operation can be avoided by adjusting the spindle load, and the spindle temperature can be reduced by optimizing the cooling system or replacing the lubricating oil.

[0036] The five-axis machine tool spindle status analysis method based on multi-dimensional data of the present application, in S600, generates adjustment instructions according to the adjustment strategy, and outputs the adjustment instructions to the operator or the machine tool control system in real time to ensure that the spindle always runs in the best working state. Through continuous monitoring and data feedback, the adjustment strategy is continuously optimized to improve the service life of the spindle and the workpiece processing quality.

[0037] The five-axis machine tool spindle status analysis method based on multi-dimensional data of the present application can comprehensively evaluate the spindle health status and discover potential problems in advance by real-time monitoring and analysis of multi-dimensional feature data through a machine learning model; predict the impact of the spindle health status on subsequent workpiece forming based on empirical data, and identify the defect type in combination with a regression model to avoid forming defects caused by spindle failures; avoid machining errors caused by spindle abnormalities through early adjustment, improve workpiece machining quality, and reduce downtime and production losses; and help reduce the spindle workload, reduce spindle failures, and extend the spindle service life through real-time monitoring and intelligent adjustment.

[0038] Based on the above-mentioned five-axis machine tool spindle state analysis method based on multidimensional data, the second aspect of the present application provides a five-axis machine tool spindle state analysis system based on multidimensional data, including: Data acquisition module, used to collect the status data of the spindle in real time through a variety of sensors; A data processing module is used to extract features from the preprocessed state data to obtain multi-dimensional feature data; A status analysis module is used to determine the health status of the spindle through a machine learning model based on multi-dimensional feature data; The defect prediction module is used to predict the influence of the spindle health status on the workpiece forming process based on the empirical data, and determine the defect type occurring during the workpiece forming process through the regression model based on the influence data; An adjustment module, used to generate adjustment strategies based on defect types; The output module is used to generate adjustment instructions according to the adjustment strategy and process the workpiece according to the adjustment instructions.

[0039] The five-axis machine tool spindle status analysis system based on multi-dimensional data of the present application ensures that the vibration, temperature, torque, speed and other sensors installed on the spindle have been correctly calibrated and can transmit data stably, and each sensor is correctly connected to the data acquisition module to ensure the accuracy and real-time performance of data transmission; the data acquisition module collects the vibration, temperature, torque, speed and other key status parameters of the spindle in real time through the sensors installed on the spindle, and transmits the collected status data to the data processing module; the data processing module processes the collected status data, including time synchronization of the collected various status data, and uses multi-dimensional data fusion technology to extract multi-dimensional feature data related to the health status of the spindle; the status analysis module uses a machine learning model to perform real-time analysis on the health status of the spindle based on the extracted multi-dimensional feature data, and obtains the spindle health status classification and health status. The defect prediction module predicts whether the current spindle health status will affect the subsequent forming of the workpiece based on the current spindle health status, combined with historical experience data and regression models, and analyzes the types of defects that may occur during the forming process of the workpiece; if the prediction analysis shows that the defect type determined according to the spindle health status may have a negative impact on the subsequent forming of the workpiece, the adjustment module will propose an adjustment strategy based on the predicted defect type; the output module generates adjustment instructions according to the adjustment strategy. The adjustment instructions can be output in the form of visual charts, reports, etc., to provide intuitive and clear reference information for machine tool operators and maintenance personnel. The adjustment instructions can also be directly output to the machine tool control system in the form of electronic instructions to adjust the processing parameters, paths or logic to achieve control of workpiece processing. The spindle health status can also be viewed in real time through the monitoring interface of the output module.

[0040] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed in the present application should be included in the protection scope of the present application. Therefore, the protection scope of the present application shall be based on the protection scope of the claims.

Claims

1. A five-axis machine tool spindle state analysis method based on multidimensional data, characterized in that: include: S100, collects spindle status data in real time through multiple sensors; S200, extracting features from the preprocessed state data to obtain multi-dimensional feature data; S300, determining the health status of the spindle through a machine learning model based on the multi-dimensional feature data; S400, predicting the influence data of the spindle health status on the workpiece forming according to the empirical data, and determining the defect type occurring in the workpiece forming process through a regression model based on the influence data; S500, generating an adjustment strategy according to the defect type; S600: Generate an adjustment instruction according to the adjustment strategy, and process the workpiece according to the adjustment instruction.

2. The five-axis machine tool spindle state analysis method based on multidimensional data according to claim 1 is characterized in that: The state data includes vibration data, temperature data, torque data and rotation speed data.

3. The five-axis machine tool spindle state analysis method based on multidimensional data according to claim 2 is characterized in that: In S200, feature extraction is performed on the preprocessed state data to obtain multi-dimensional feature data, including: S210, sorting the state data according to timestamps, and synchronizing the data by interpolation; S220, preprocessing the state data, wherein the preprocessing method at least includes cleaning, denoising, and filtering; S230, extracting multidimensional feature data related to the health status of the spindle from the status data; S240: Fusing the multi-dimensional feature data by using a principal component analysis method.

4. The five-axis machine tool spindle state analysis method based on multidimensional data according to claim 3 is characterized in that: In S230, multi-dimensional feature data related to the spindle health status is extracted from the status data, including: Extracting vibration characteristic data related to the health status of the spindle from the vibration data, including vibration amplitude and vibration frequency; Extracting temperature characteristic data related to the health status of the spindle from the temperature data, including a temperature change rate and a temperature change trend; Extracting torque characteristic data related to the health status of the spindle from the torque data, including load fluctuation amplitude and load fluctuation frequency; Speed ​​characteristic data related to the spindle health status is extracted from the speed data, including the speed fluctuation amplitude.

5. The five-axis machine tool spindle state analysis method based on multidimensional data according to claim 4 is characterized in that: In S240, the multi-dimensional feature data is fused by a principal component analysis method, including: S241, performing standardization processing on the multi-dimensional feature data, including: Calculate the mean of multi-dimensional feature data ; Calculate the standard deviation of multidimensional feature data ; Standardize multidimensional feature data by mean and standard deviation: ； in, For the data set i Multi-dimensional feature data, is the mean of the multidimensional feature data, is the standard deviation of the multidimensional feature data, For the data set i Data points obtained by standardization of multi-dimensional feature data; S242, construct a covariance matrix, solve the eigenvalues ​​and eigenvectors, including: Construct the covariance matrix: ； in, Z For data points The multidimensional feature matrix composed of n is the number of samples, for Z The transposed matrix of is the covariance matrix; Construct the characteristic polynomial from the covariance matrix: ； Where I is the identity matrix, is the characteristic polynomial; By solving the characteristic polynomial , get all the eigenvalues ​​of the covariance matrix , m is the total number of eigenvalues; For each eigenvalue , by solving the system of equations , and obtain the corresponding feature vector , k =1, 2, 3, ..., m ; S243, sorting the eigenvalues ​​from large to small, and selecting a plurality of principal components preset previously; S244: Project the multidimensional feature data onto a plurality of selected principal components to obtain multidimensional feature data after dimensionality reduction.

6. The five-axis machine tool spindle state analysis method based on multidimensional data according to claim 5 is characterized in that: In S300, based on the multi-dimensional feature data, determining the health status of the spindle through a machine learning model includes: S310, dividing the multidimensional feature data into a training set, a validation set, and a test set; S320, using the deep learning framework Keras to build a fully connected neural network model; S330, training the fully connected neural network model through the training set, performing model selection and hyperparameter tuning on the fully connected neural network model through the validation set, and evaluating the fully connected neural network model through the test set; S340. Determine the health status of the spindle through the fully connected neural network model, where the types of the health status of the spindle include healthy, slightly abnormal, and seriously abnormal.

7. The five-axis machine tool spindle state analysis method based on multidimensional data according to claim 6 is characterized in that: In S400, the influence data of the spindle health status on the workpiece forming is predicted according to the empirical data, and based on the influence data, the defect type occurring in the workpiece forming process is determined by a regression model, including: S410, predicting the influence data of the spindle health status on workpiece forming according to the empirical data, wherein the influence data includes a health status score and a plurality of processing parameters; S420, inputting the health status score and the multiple processing parameters into a regression model to calculate the workpiece quality: ； Where Y is the mass of the workpiece, is the intercept, is the corresponding regression coefficient, is a health status score or a processing parameter; S430, determining the type of defects occurring during the workpiece forming process according to the workpiece quality.

8. The five-axis machine tool spindle state analysis method based on multidimensional data according to claim 7 is characterized in that: The processing parameters at least include cutting speed, feed rate, cutting depth, tool wear, and coolant status.

9. The five-axis machine tool spindle state analysis method based on multidimensional data according to claim 8, characterized in that: In S500, the adjustment strategy generated according to the defect type includes: Adjust the spindle speed or feed speed; Adjust the spindle load; Optimize the cooling system or replace the lubricating oil; Replace or repair faulty parts.

10. A five-axis machine tool spindle state analysis system based on multidimensional data, based on the five-axis machine tool spindle state analysis method based on multidimensional data according to any one of claims 1 to 9, characterized in that: include: Data acquisition module, used to collect the status data of the spindle in real time through a variety of sensors; A data processing module, used for extracting features from the preprocessed state data to obtain multi-dimensional feature data; A state analysis module, used to determine the health state of the spindle through a machine learning model based on the multi-dimensional feature data; A defect prediction module is used to predict the influence data of the spindle health status on the workpiece forming according to the empirical data, and determine the defect type occurring in the workpiece forming process through a regression model based on the influence data; An adjustment module, used for generating an adjustment strategy according to the defect type; The output module is used to generate an adjustment instruction according to the adjustment strategy and process the workpiece according to the adjustment instruction.

Images