Precision manufacturing multi-axis linkage machine tool calibration method and device

Abstract

The present invention relates to the technical field of machine tool calibration, and discloses a calibration method and device for a precision manufacturing multi-axis linkage machine tool. The method includes: collecting the machine tool operation data and multi-modal sensor data of the multi-axis linkage machine tool and performing error prediction to obtain error prediction data; measuring the positions of each axis of the machine tool to obtain position data with nanometer-level resolution; obtaining the real-time state representation of the machine tool through a multi-branch deep neural network; constructing a multi-level hypergraph based on the real-time state representation of the machine tool and performing feature processing to obtain a fault feature vector; performing reinforcement learning through a deep Q network to obtain an adaptive compensation strategy; generating the linkage compensation values for each axis based on the adaptive compensation strategy, and performing real-time adjustment and fault prevention analysis on the multi-axis linkage machine tool to generate a fault prevention control plan. The present invention improves the accuracy retention ability and operation reliability of the multi-axis linkage machine tool, and can be applied to different types of multi-axis linkage machine tools.

Description

Technical Field

[0001] The present invention relates to the technical field of machine tool calibration, and particularly to a calibration method and device for a precision manufacturing multi-axis linkage machine tool. Background Art

[0002] With the development of modern manufacturing towards high precision and high efficiency, multi-axis linkage machine tools, as key equipment for high-end equipment manufacturing, their machining accuracy and operation stability directly affect product quality. Traditional machine tool calibration methods mainly rely on manual experience and single-sensor data, making it difficult to achieve precise error compensation and real-time adjustment, resulting in a gradual decline in the accuracy of machine tools during long-term operation and frequent failures.

[0003] Currently, multi-axis linkage machine tools in actual operation face technical problems such as the difficulty in effectively fusing multi-source heterogeneous data, the difficulty in maintaining nanoscale position accuracy, and the difficulty in accurately predicting multi-axis collaborative errors. Especially under the influence of external factors such as temperature changes and vibration interference, the coupling errors between the axes of the machine tool will accumulate over time, and traditional single compensation strategies cannot meet the accuracy requirements under complex working conditions. In addition, existing machine tool fault diagnosis methods are often passive, lacking the ability of active prevention and intelligent decision-making, making it difficult to detect potential faults in a timely manner and take effective preventive measures. This leads to high machine tool maintenance costs and reduced equipment utilization rates. Summary of the Invention

[0004] The present invention provides a calibration method and device for a precision manufacturing multi-axis linkage machine tool, which improves the accuracy retention ability and operation reliability of the multi-axis linkage machine tool and can be applied to different types of multi-axis linkage machine tools.

[0005] In a first aspect, the present invention provides a calibration method for a precision manufacturing multi-axis linkage machine tool, and the calibration method for the precision manufacturing multi-axis linkage machine tool includes:

[0006] Collect the machine tool operation data and multi-modal sensor data of the multi-axis linkage machine tool, and perform error prediction on the machine tool operation data through a multi-axis linkage error prediction model to obtain error prediction data;

[0007] Measure the positions of the axes of the machine tool using an atomic interferometer and a grating scale, and obtain nanoscale resolution position data through Kalman filter processing;

[0008] Input the multi-modal sensor data and the nanoscale resolution position data into a multi-branch deep neural network, and fuse them through an attention mechanism to obtain a real-time state representation of the machine tool;

[0009] Construct a multi-level hypergraph based on the real-time state representation of the machine tool, and input the multi-level hypergraph into a hierarchical hypergraph model for feature processing to obtain a fault feature vector;

[0010] Taking the fault feature vector and the error prediction data as state input, performing reinforcement learning through a deep Q network to obtain an adaptive compensation strategy;

[0011] Based on the adaptive compensation strategy, the linkage compensation value of each axis is generated, and the multi-axis linkage machine tool is adjusted in real time and fault prevention analysis is performed to generate a fault prevention control plan.

[0012] In a second aspect, the present invention provides a precision manufacturing multi-axis linkage machine tool calibration device, the precision manufacturing multi-axis linkage machine tool calibration device comprising:

[0013] An acquisition module is used to acquire machine tool operation data and multi-modal sensor data of a multi-axis linkage machine tool, and to perform error prediction on the machine tool operation data through a multi-axis linkage error prediction model to obtain error prediction data;

[0014] The measurement module is used to measure the position of each axis of the machine tool using an atomic interferometer and a grating ruler, and obtain nanometer-level resolution position data through Kalman filtering;

[0015] A fusion module, used for inputting the multimodal sensor data and the nanometer-level resolution position data into a multi-branch deep neural network, and fusing them through an attention mechanism to obtain a real-time state representation of the machine tool;

[0016] A feature processing module, used for constructing a multi-level hypergraph based on the real-time state representation of the machine tool, and inputting the multi-level hypergraph into a hierarchical hypergraph model for feature processing to obtain a fault feature vector;

[0017] A reinforcement learning module, used for taking the fault feature vector and the error prediction data as state inputs, performing reinforcement learning through a deep Q network, and obtaining an adaptive compensation strategy;

[0018] A generation module is used to generate the linkage compensation value of each axis based on the adaptive compensation strategy, and to perform real-time adjustment and fault prevention analysis on the multi-axis linkage machine tool to generate a fault prevention control plan.

[0019] The third aspect of the present invention provides a computer device, comprising: a memory and at least one processor, wherein the memory stores instructions; the at least one processor calls the instructions in the memory so that the computer device executes the above-mentioned precision manufacturing multi-axis linkage machine tool calibration method.

[0020] A fourth aspect of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores instructions, which, when executed on a computer, enable the computer to execute the above-mentioned precision manufacturing multi-axis linkage machine tool calibration method.

[0021] In the technical solution provided by the present invention, through the fusion of multi-modal sensor data and nano-scale resolution position data, combined with the multi-head self-attention mechanism, high-precision perception and feature extraction of the machine tool state are achieved, improving the accuracy and real-time performance of state monitoring. A multi-level hypergraph structure is used to model the physical layer, motion layer, and functional layer of the machine tool, and feature processing is carried out through the hierarchical hypergraph model, effectively capturing the correlation relationships between different levels and enhancing the expression ability of fault features. Based on the reinforcement learning method of the deep Q network, dynamic optimization of the adaptive compensation strategy is realized, enabling the compensation value to be adaptively adjusted according to the real-time state of the machine tool, significantly improving the compensation effect. By constructing a virtual model through digital twin technology and optimizing the fault prevention plan in the virtual environment, early warning and active prevention of faults are achieved, reducing the equipment maintenance cost and extending the service life of the machine tool. The present invention improves the accuracy retention ability and operation reliability of multi-axis linkage machine tools and can be applied to different types of multi-axis linkage machine tools. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly introduced below. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0023] Figure 1 It is a schematic diagram of the steps of the precision manufacturing multi-axis linkage machine tool calibration method in the embodiments of the present invention;

[0024] Figure 2 It is a schematic diagram of the structure of the precision manufacturing multi-axis linkage machine tool calibration device in the embodiments of the present invention;

[0025] Figure 3 It is a schematic block diagram of the structure of the computer device in the embodiments of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0026] An embodiment of the present invention provides a calibration method and device for a precision manufacturing multi-axis linkage machine tool. Terms such as "first", "second", "third", "fourth", etc. (if any) in the specification, claims and above-mentioned drawings of the present invention are used to distinguish similar objects and do not necessarily need to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances so that the embodiments described here can be implemented in an order different from that illustrated or described here. In addition, the term "comprising" or "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device comprising a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0027] For ease of understanding, the specific process of the embodiment of the present invention will be described below. Please refer to Figure 1 , an embodiment of the calibration method for a precision manufacturing multi-axis linkage machine tool in the embodiment of the present invention includes:

[0028] Step S1: Collect the machine tool operation data and multi-modal sensor data of the multi-axis linkage machine tool, and perform error prediction on the machine tool operation data through a multi-axis linkage error prediction model to obtain error prediction data;

[0029] It can be understood that the execution subject of the present invention can be a calibration device for a precision manufacturing multi-axis linkage machine tool, or a terminal or a server. Specifically, it is not limited here. The embodiment of the present invention is described by taking the server as the execution subject as an example.

[0030] Specifically, the position, speed, and acceleration of each axis of the multi-axis linkage machine tool are collected in real time to obtain the machine tool operation data. At the same time, the temperature distribution, vibration characteristics, and stress state of the machine tool are collected through a temperature sensor array, a three-axis acceleration sensor, and a strain gauge respectively to obtain multi-modal sensor data. The real-time data collection enables the system to comprehensively analyze and record the operation status of the machine tool, including the environmental impact factors of the machine tool, such as temperature changes, vibration, and stress distribution. The machine tool operation data and multi-modal sensor data are time-synchronized to ensure that all data can be effectively used together to capture the real machine tool operation status and environmental changes. The sliding window method is used to segment these data. The size of the sliding window is set to 1 second, and an overlap rate of 50% is set. The overlapping process enhances the capture of data continuity and improves the sensitivity of the prediction model to short-term changes. The data within each window is normalized using standardization processing to ensure that various data features are on the same scale, enabling the subsequent error prediction model to effectively process this data and obtain the first standardized input data. The first standardized input data is input into the first input layer of the multi-axis linkage error prediction model for one-dimensional convolution processing. This input layer contains N parallel one-dimensional convolutional neural networks, where N represents the number of axes of the machine tool, so that each axis of the machine tool has a corresponding one-dimensional convolutional network dedicated to processing its data to ensure the independence and exclusivity of each axis data. Each one-dimensional convolutional neural network contains 3 convolutional layers. The kernel sizes of the convolutional layers are 5, 3, and 3 respectively, and the number of kernels is 32, 64, and 128 respectively. The input data is convolved through convolutional kernels of different sizes and numbers to extract features of different scales in the operation data. BatchNormalization and ReLU activation functions are connected to each convolutional layer. BatchNormalization is used to accelerate the training process and stabilize the network performance, while the ReLU activation function introduces non-linearity, enabling the network to learn complex features and obtain the first convolutional feature vector. The first convolutional feature vector is fused with cross-axis features through 2 three-dimensional convolutional layers. The kernel size of the three-dimensional convolutional layer is 3x3x3, and the number of kernels is 256 and 512 respectively. Three-dimensional convolution can perform convolution operations on the data of multiple axes simultaneously to capture the mutual influence and linkage features between different axes. After each three-dimensional convolutional layer, MaxPooling and Dropout operations are connected. MaxPooling is used to reduce the dimension of the features, reduce the computational amount, and extract important feature information at the same time. Dropout is used to prevent overfitting and enhance the generalization ability of the model. After being processed by two three-dimensional convolutional layers, the second convolutional feature vector is obtained. The second convolutional feature vector is input into the multi-head self-attention mechanism layer, and the self-attention mechanism is used to analyze the correlation of features between different axes.The number of heads of the multi-head self-attention mechanism is set to 8, and the dimension of each head is 64. The correlation weights between different axes are calculated through scaled dot-product attention. The attention mechanism can assign different weights to each feature, thereby strengthening the attention to key features in the overall state, performing feature recombination, and obtaining a weighted convolutional feature vector. The weighted convolutional feature vector is used to capture temporal features, effectively capturing the long-term dependencies in the feature vector and obtaining a feature vector with temporal characteristics. The temporal feature vector is input into the first fully connected layer, which contains 3 hidden layers. The number of neurons in the hidden layers is 256, 128, and 64 respectively, and the feature dimensionality reduction and abstraction are gradually carried out. A linear activation function is used in the last hidden layer to map the features to the output space, with an output dimension of N, corresponding to the predicted error values of N machine tool axes, and error prediction data is obtained.

[0031] Step S2: Use an atomic interferometer and a grating scale to measure the positions of each axis of the machine tool, and obtain position data with nanometer-level resolution through Kalman filter processing;

[0032] Specifically, high-precision grating scales are installed on each axis of a multi-axis machine tool. The resolution of the grating scale is 1 nanometer, and the sampling frequency is 10 kilohertz. By preliminarily measuring the positions of each axis, grating scale position data is obtained. As a high-precision displacement measurement tool, the grating scale can provide accurate position information for each axis of the machine tool. An atomic interferometer system is fixed on the reference plane of the machine tool to precisely measure the positions of each axis of the machine tool relative to the reference plane. The atomic interferometer system includes a laser light source with stable frequency, a beam splitter, multiple mirrors, and a photodetector. Through these optical components, the atomic interferometer system can precisely measure the change in position using the interference effect, with a measurement resolution of up to 0.1 nanometer and a sampling frequency of 100 kilohertz. This makes the atomic interferometer position data have a higher resolution than the grating scale position data and can provide an accurate description of the relative positions of each axis of the machine tool. The grating scale position data and the atomic interferometer position data are time-synchronized to ensure that data from different sources have a consistent time reference, forming synchronized position data. Through the synchronization process, measurement errors and inconsistencies caused by data time offsets are effectively eliminated. Wavelet transform is performed on the synchronized position data to extract the multi-scale position features of the data. Wavelet transform can analyze the data in both the time and frequency domains and extract the position change features at different scales. The multi-scale features help to more comprehensively characterize the dynamic characteristics of the machine tool position change, including the smooth change and transient features of the position at different time scales. Based on the multi-scale position features, a state space model is constructed to describe the dynamic behavior of the system. In the state space model, the state variables include position, velocity, and acceleration, while the observed variable is the measured value of the position. In this way, the position change of the machine tool is described using a linear state space representation. This model can well capture the dynamic relationship between position, velocity, and acceleration and transform it into a form that can be used for filter processing. Based on this linear state space representation, a Kalman filter is created to estimate the state. The Kalman filter is a recursive algorithm for dynamic system state estimation, including a state prediction equation and a measurement update equation. When initializing the Kalman filter, the covariance matrix P of the state estimate and the process noise covariance matrix Q are set to ensure that the Kalman filter can reasonably balance the relationship between prediction and measurement and achieve optimal state estimation. The synchronized position data is input into the Kalman filter, and the system state is estimated through recursive calculation. During the recursive process, the Kalman filter uses the state prediction equation to predict the state at the next moment based on the current state, and then uses the measurement update equation to correct the prediction result in combination with the measurement data, thereby gradually updating the state estimate and the error covariance. This process can effectively fuse the data of the grating scale and the atomic interferometer, make full use of the complementary advantages of both, and improve the accuracy and resolution of position estimation. After Kalman filtering, the filtered position estimate value is obtained.Perform an inverse wavelet transform on the filtered position estimate to reconstruct position data with nanoscale resolution. Through the inverse wavelet transform, the multi-scale features are restored to the position information at the original scale.

[0033] Step S3: Input the multi-modal sensor data and the nanoscale resolution position data into a multi-branch deep neural network, and fuse them through an attention mechanism to obtain a real-time state representation of the machine tool;

[0034] Specifically, time alignment and normalization processing are performed on the multimodal sensor data and nanoscale resolution position data to obtain the second normalized input data. Time alignment can ensure that all data from different sources are synchronized under the same time reference, avoiding information loss or inconsistency caused by data time offset. Normalization processing maps various types of data to the same dimension for the efficient processing of the subsequent feature extraction network. The second normalized input data is divided into four branches: temperature data, vibration data, stress data, and position data, and are respectively input into four parallel feature extraction networks for feature extraction. Each feature extraction network contains 3 one-dimensional convolutional layers, where the sizes of the convolutional kernels are 5, 3, and 3 respectively, and the numbers of convolutional kernels are 32, 64, and 128 respectively. Through these convolutional layers, the model can gradually perform convolutional operations on the input data and extract features of different scales in the data. BatchNormalization and ReLU activation functions are connected to each convolutional layer, where BatchNormalization is used to stabilize the training process of the network and avoid gradient vanishing or gradient explosion, while the ReLU activation function introduces non-linearity, enabling the network to have the ability to learn complex features, and obtaining the primary feature vectors of each branch. Global average pooling and global max pooling operations are performed on the primary feature vectors of each branch to extract global feature information from each feature vector, respectively reflecting the average characteristics and extreme value characteristics of the overall data. Through these two pooling operations, the feature dimension is effectively reduced, and information at different levels is captured, improving the robustness and expressiveness of the features. After concatenating the pooled feature vectors of each branch, a global feature vector is formed. Based on the global feature vector, the correlation weights between different modalities are calculated through the scaled dot-product attention mechanism to obtain the attention-weighted features. The role of the scaled dot-product attention mechanism is to perform correlation analysis on different modalities in the global features, thereby assigning appropriate weights to the data of each modality to better represent the importance of the data of each modality. In this way, the attention mechanism can enhance the attention to key features, suppress redundant or unimportant features, and make the overall feature representation more focused on the information crucial for state judgment. Temporal feature extraction is performed on the attention-weighted features to capture the temporal context relationship of each modality feature, obtaining the temporal context vector. The temporal context vector is input into the cross-modal transformer encoder for fusion processing. The cross-modal transformer encoder contains 6 encoder layers, and each encoder layer contains a multi-head self-attention sublayer and a feed-forward neural network sublayer. The multi-head self-attention sublayer can perform attention calculations with multiple different heads on the input data to capture the complex relationships between different modality features, while the feed-forward neural network sublayer performs feature mapping and abstraction on the features after attention processing. Through the stacking of multiple encoder layers, the transformer can perform deep fusion and extraction of cross-modal features to obtain the fused feature representation.Perform residual connection and layer normalization operations on the fused feature representation, where the residual connection is used to retain the information of the input features and prevent the problem of vanishing gradients during the training of deep networks, while layer normalization is to further stabilize the training process. Feature mapping is performed on the fused feature representation through the non-linear activation function ReLU to obtain the intermediate feature representation. The intermediate feature representation is input into the second fully connected layer, which contains 3 hidden layers, and each hidden layer contains 512, 256, and 128 neurons respectively, abstracting and reducing the dimensionality of the input features layer by layer to extract important information in the features. The Softmax activation function is used in the last hidden layer to map the features to the real-time state representation of the machine tool. This output representation can reflect the comprehensive state of the multi-axis linkage machine tool at the current moment and contains the information after data fusion from different modalities.

[0035] Step S4: Construct a multi-level hypergraph based on the real-time state representation of the machine tool, and input the multi-level hypergraph into the hierarchical hypergraph model for feature processing to obtain the fault feature vector;

[0036] Specifically, the real-time state representation of the machine tool is divided into three levels: the physical layer, the motion layer, and the function layer. Each level contains N nodes, where N is the number of axes of the machine tool, thus obtaining the first node set. Attribute assignment is performed on each node in the first node set. The attributes of the physical layer nodes include temperature, vibration, and stress. The attributes of the motion layer nodes include position, velocity, and acceleration. The attributes of the function layer nodes include machining accuracy and motion synchronization, thus obtaining the second node set. Based on the second node set, a multi-level hypergraph structure is constructed. When constructing the hypergraph, the similarity between nodes is calculated by the K-nearest neighbor algorithm, and the threshold of the similarity is set to 0.8. Hyperedge connections are established between nodes with similarity greater than the threshold to obtain the initial hypergraph structure. Through the K-nearest neighbor algorithm, nodes similar in a specific attribute space are effectively found and connected by hyperedges. The threshold of the similarity is set to 0.8 to ensure strong similarity between the selected nodes, making the constructed hyperedge connections more reasonable and accurate. Intra-layer edge weight calculation is performed on the initial hypergraph structure. The cosine similarity is used to calculate the correlation between node attributes, and the correlation is used as the edge weight to obtain the weighted intra-layer hypergraph. The cosine similarity is used to measure the cosine value of the angle between two vectors. The closer its value is to 1, the higher the similarity between the two node attributes. The calculation of the intra-layer edge weight by the cosine similarity reflects the attribute relationship between nodes, making the intra-layer hypergraph more realistically represent the association between nodes. Inter-layer edge weight calculation is performed on the weighted intra-layer hypergraph. The Pearson correlation coefficient is used to calculate the correlation between node attributes of different layers, and the calculated correlation is used as the inter-layer edge weight to obtain the target multi-level hypergraph. The Pearson correlation coefficient is used to measure the linear correlation degree between two variables, and its value ranges from -1 to 1, which can well reflect the positive or negative correlation relationship between nodes of different levels. Through the weighting of the inter-layer edges, the multi-level hypergraph can represent the relationships within each layer of nodes and capture the interaction relationships between different layers, constructing a more comprehensive multi-level network structure. The target multi-level hypergraph is input into the hierarchical hypergraph model for feature processing. The hierarchical hypergraph model contains 3 hypergraph convolutional layers, and the output channel numbers of each convolutional layer are 64, 128, and 256 respectively. The activation function is LeakyReLU. The hypergraph convolutional layer is used to perform convolutional operations on node features in the hypergraph structure to extract deeper features. The introduction of the LeakyReLU activation function effectively avoids the "death" of neurons, enabling the network to still retain a part of the activation output when facing negative inputs, enhancing the network's non-linear expression ability and overall feature extraction ability. Through the stacking of 3 hypergraph convolutional layers, the model gradually abstracts and integrates the input features to extract higher-level hierarchical features. Cross-layer feature fusion is performed on the hierarchical features to obtain cross-layer fusion features. Information from different levels is integrated to capture the interaction and influence of the machine tool state between different levels.Through cross-layer fusion, the feature representation includes the information within each layer and the interaction relationships between layers. The cross-layer fusion features are input into the third fully connected layer, which contains two hidden layers with the number of neurons in the hidden layers being 128 and 64 respectively. The input features are gradually reduced in dimension and abstracted to further extract and retain the information that is most crucial for fault prediction. The Tanh activation function is used in the last layer of the fully connected layer to map the features to the final output space with an output dimension of 32, thereby obtaining the fault feature vector. The output range of the Tanh activation function is from -1 to 1, which can effectively perform non-linear mapping on the features, ensure the balance of the numerical distribution of the output, and better adapt to subsequent classification or regression tasks.

[0037] Step S5: Use the fault feature vector and the error prediction data as states as inputs, and perform reinforcement learning through a deep Q-network to obtain an adaptive compensation strategy;

[0038] Specifically, the fault feature vector is concatenated with the error prediction data to obtain a state vector. The state vector is normalized to obtain a standardized state input, which improves the stability and speed of learning and reduces the training difficulty caused by differences in data ranges. The standardized state input is input into the second input layer of the deep Q-network. This input layer is a fully connected layer containing 256 neurons and uses the ReLU activation function. Through the processing of the fully connected layer, the high-dimensional state vector is converted into a feature representation. The ReLU activation function enables the model to have the ability to learn non-linear relationships, so that the model can better extract the complex associations between different states and obtain an initial feature representation. The initial feature representation is input into a deep feature extraction module composed of 3 residual blocks. Each residual block contains two convolutional layers and a short-circuit connection. The size of the convolutional kernel is 3x3, and the number of output channels is 128, 256, and 512 respectively. Through the residual blocks, the model can learn deep features while retaining the initial information. The short-circuit connection can effectively alleviate the problem of gradient disappearance in the deep network, enabling information to be better transmitted between network layers and maintaining the model's sensitivity to the initial features. After each convolutional layer, the model gradually captures more complex and abstract features in the state vector to obtain deep features. The deep features are input into the attention layer, and the correlation weights between the features are calculated through the self-attention mechanism to obtain attention-weighted features. The self-attention mechanism analyzes the mutual relationships between different features, assigns higher weights to important features, highlights the features that are most critical for the current state evaluation and action selection, enables the model to more accurately focus on the key feature information, and enhances the ability to understand the machine tool state. The attention-weighted features are subjected to temporal modeling to obtain temporal context features. Temporal modeling captures the evolution relationship of features over time, thereby reflecting the dynamic changes of the machine tool state. The temporal context features are input into the advantage function estimator. The advantage function estimator contains two fully connected layers with the number of neurons being 128 and 64 respectively. The last layer uses the linear activation function to output the Q-values corresponding to different compensation actions to obtain the action value estimation. The Q-value is an expected estimate of the rewards brought by different actions in the current state. The temporal context features are abstracted through two fully connected layers to extract high-level features related to action selection. The introduction of the linear activation function enables the output Q-values to vary within any range, better distinguishing the values of different actions. Based on the obtained action value estimation, the ε-greedy strategy is adopted to select the compensation action, where the ε value is initially set to 0.9 and gradually decreases according to the exponential decay rule. The ε-greedy strategy is a strategy that balances exploration and exploitation. In the initial stage, the model tends to randomly select compensation actions to explore the optimal strategy. As the training progresses, the ε value gradually decreases, and the model will more often select the currently known optimal actions, thus realizing the transition from exploration to exploitation and ensuring the convergence and optimality of the model.In this way, a compensation action sequence is obtained for real-time adjustment of the operating state of the machine tool. To improve the training effect of the deep Q-network, experience replay is performed on the compensation action sequence. The role of experience replay is to save the compensation actions and the corresponding state changes. During subsequent training, a batch of experience samples are randomly sampled for updating the parameters of the Q-network. This method can break the temporal correlation of the data, increase the independence of the data, and effectively improve the training efficiency and stability of the model. Through temporal difference learning, the parameters of the Q-network are gradually updated, enabling the model to more accurately evaluate the value of different actions, and finally an adaptive compensation strategy is obtained.

[0039] Step S6: Generate the coupling compensation values for each axis based on the adaptive compensation strategy, and perform real-time adjustment and fault prevention analysis on the multi-axis linkage machine tool to generate a fault prevention control plan.

[0040] Specifically, the adaptive compensation strategy is input into the compensation value generator, and the initial compensation value of each axis is obtained by inverse kinematics calculation. The calculation process of inverse kinematics is to deduce the required displacement or rotation of each axis according to the target state and the geometric model of the machine tool to achieve the expected posture correction. The initial compensation value is nonlinearly optimized to overcome the errors and uncertainties in the kinematic model and obtain the linkage compensation value of each axis. The nonlinear optimization process uses a nonlinear programming algorithm to ensure the accuracy and linkage of the compensation value, thereby minimizing the error and improving the compensation effect. The multi-axis linkage machine tool is adjusted in real time based on the linkage compensation value of each axis, and the operating parameters of each axis are adjusted to ensure that the machine tool can maintain high-precision processing performance under different conditions. While completing the real-time adjustment, the adjusted machine tool operation data is continuously collected to monitor the operating status of the machine tool and ensure the effectiveness of the compensation strategy. The collected data is fast Fourier transformed to obtain the frequency domain characteristics of the machine tool in the adjusted operating state. Fast Fourier transform converts the operating signal of the machine tool from the time domain to the frequency domain, which is convenient for analyzing abnormal vibration or periodic characteristics in the operation process of the machine tool. The frequency domain features are input into a multi-class support vector machine for fault prediction. The multi-class support vector machine is an effective classification algorithm used to classify input features to determine whether there is a fault or abnormality in the current state of the machine tool. Among them, the kernel function uses the radial basis function, which can handle nonlinear data well and adapt to complex feature distribution, thereby improving the accuracy of fault prediction. Through the calculation of the support vector machine, the probability distribution of the fault type is obtained, and the faults of the machine tool are classified and predicted based on the distribution. The threshold judgment of the probability distribution of the fault type is performed, and the threshold of the fault warning is set to 0.8. When the predicted probability of a certain type of fault exceeds this threshold, the fault warning signal is triggered. After the fault warning signal is triggered, the fault tracing analysis program is immediately started. The main task of the program is to find out the specific cause of the warning signal. First, a fault causal relationship graph is constructed, and the root cause of the fault is located through probabilistic reasoning. The fault causal relationship graph is a model based on a graph structure, in which nodes represent different states or parameters of the system, and edges represent the causal relationship between these states or parameters. By performing probabilistic reasoning on the fault causal relationship graph, it is possible to find out which nodes have a strong causal relationship, so as to accurately locate the root cause of the fault and obtain the fault tracing result. Based on the fault tracing results, a virtual model of the machine tool is constructed using digital twin technology. Fault prevention solutions are executed and optimized in a virtual environment. Digital twin technology can realistically simulate the operating performance of machine tools in different states by building a mirror model of the machine tool in a virtual environment. In the virtual model, different control solutions are simulated and compared based on the fault tracing results to find the optimal fault prevention solution. Based on the verification results in the digital twin environment, a fault prevention control solution is formed and applied to the actual machine tool control, thereby achieving real-time adjustment and fault prevention analysis of multi-axis linkage machine tools.Meanwhile, through real-time adjustment based on the adaptive compensation strategy, the errors during the operation of the machine tool are effectively compensated, ensuring high precision and high consistency in the machining process.

[0041] In the embodiments of the present invention, through the fusion of multi-modal sensor data and nano-level resolution position data, combined with the multi-head self-attention mechanism, high-precision perception and feature extraction of the machine tool state are realized, improving the accuracy and real-time performance of state monitoring. A multi-level hypergraph structure is used to model the physical layer, motion layer, and functional layer of the machine tool. Through feature processing with the hierarchical hypergraph model, the correlation relationships between different levels are effectively captured, enhancing the expression ability of fault features. Based on the reinforcement learning method of the deep Q-network, the dynamic optimization of the adaptive compensation strategy is realized, enabling the compensation value to be adaptively adjusted according to the real-time state of the machine tool, significantly improving the compensation effect. By constructing a virtual model through digital twin technology and optimizing the fault prevention plan in the virtual environment, early warning and active prevention of faults are realized, reducing the equipment maintenance cost and extending the service life of the machine tool. The present invention improves the accuracy retention ability and operation reliability of multi-axis linkage machine tools and can be applied to different types of multi-axis linkage machine tools.

[0042] In a specific embodiment, the process of executing step S1 may specifically include the following steps:

[0043] Collect the real-time position, speed, and acceleration of each axis of the multi-axis linkage machine tool to obtain the machine tool operation data, and collect the temperature distribution, vibration characteristics, and stress state of the machine tool through a temperature sensor array, a three-axis acceleration sensor, and a strain gauge to obtain multi-modal sensor data;

[0044] Synchronize the time of the machine tool operation data and the multi-modal sensor data, segment the data using the sliding window method, with a window size of 1 second and an overlap rate of 50%, and then perform normalization processing on the data within each window to obtain the first normalized input data;

[0045] Input the first normalized input data into the first input layer of the multi-axis linkage error prediction model for one-dimensional convolution processing to obtain the first convolution feature vector. The first input layer includes N parallel one-dimensional convolutional neural networks, where N is the number of machine tool axes. Each one-dimensional convolutional neural network includes 3 convolutional layers, with convolutional kernel sizes of 5, 3, and 3 respectively, and convolutional kernel numbers of 32, 64, and 128 respectively. After each convolutional layer, BatchNormalization and ReLU activation functions are connected;

[0046] Extract features from the first convolution feature vector, perform cross-axis feature fusion through 2 three-dimensional convolutional layers, with convolutional kernel sizes of 3x3x3 and convolutional kernel numbers of 256 and 512 respectively. After each layer, MaxPooling and Dropout operations are connected to obtain the second convolution feature vector;

[0047] Input the second convolutional feature vector into the multi-head self-attention mechanism layer with 8 heads and a dimension of 64 for each head. Calculate the correlation weights between different axes through scaled dot-product attention and perform feature recombination to obtain a weighted convolutional feature vector;

[0048] Capture the temporal features and long-term dependencies of the weighted convolutional feature vector to obtain a temporal feature vector, and input the temporal feature vector into the first fully connected layer. The first fully connected layer contains 3 hidden layers with the number of neurons being 256, 128, and 64 respectively. The last layer uses a linear activation function, and the output dimension is N, corresponding to the predicted error values of N axes, to obtain error prediction data.

[0049] Specifically, the real-time position, speed, and acceleration of each axis of the multi-axis linkage machine tool are collected to obtain the operation data of the machine tool, which reflects the dynamic motion state of the machine tool at different time points. At the same time, the environmental state information of the machine tool is obtained through a variety of sensors. The temperature sensor array accurately measures the temperature distribution of different parts of the machine tool, the three-axis acceleration sensor is used to measure the vibration characteristics during the operation of the machine tool, and the strain gauge is used to measure the stress state of the machine tool. These data together constitute multi-modal sensor data. The operation data of the machine tool and the multi-modal sensor data are time-synchronized to ensure that all data are processed under the same time reference, eliminating information deviation caused by time inconsistency. The sliding window method is used to segment the data, the window size is 1 second, and the overlap rate is 50%. The role of sliding window segmentation is to ensure the continuity of the data and enhance the sensitivity to short-term changes through overlapping. The data within each window is normalized to eliminate the differences between different feature dimensions, obtaining the first normalized input data. The first normalized input data is input into the first input layer of the multi-axis linkage error prediction model for one-dimensional convolution processing. The first input layer contains N parallel one-dimensional convolutional neural networks, where N is the number of axes of the machine tool. The data of each axis is passed through an independent convolutional network for feature extraction to ensure that the features of each axis can be independently extracted and learned. Each one-dimensional convolutional neural network contains 3 convolutional layers, the sizes of the convolutional kernels are 5, 3, and 3 respectively, and the numbers of convolutional kernels are 32, 64, and 128 respectively. The purpose of the convolution operation is to extract local features from the input data. A convolutional kernel size of 5 means performing a sliding convolution on the data for 5 time steps, which can capture features in a larger range, while the subsequent 3 and 3 convolutional kernels can capture more subtle feature changes. After each convolutional layer, BatchNormalization and the ReLU activation function are connected. BatchNormalization is used to normalize the convolutional output, thereby accelerating the training process of the network and improving the stability of the network; the ReLU activation function is used to introduce non-linearity, enabling the network to learn more complex feature relationships. After convolution processing, the first convolutional feature vector is obtained. The first convolutional feature vector is input into two three-dimensional convolutional layers for cross-axis feature fusion. The convolutional kernel size of the three-dimensional convolution is 3x3x3, and the numbers of convolutional kernels are 256 and 512 respectively. The three-dimensional convolution performs convolution operations on the features in both the spatial and time dimensions, capturing the linkage relationships between different axes. Through three-dimensional convolution, the associated features are effectively extracted, enhancing the understanding of the overall state of the system. MaxPooling and Dropout operations are connected to each three-dimensional convolutional layer. MaxPooling is used to reduce the dimension of the features, reduce the computational amount, and at the same time retain important feature information; Dropout is used to prevent overfitting and enhance the generalization ability of the model. After the processing of these two three-dimensional convolutional layers, the second convolutional feature vector is obtained.Input the second convolutional feature vector into the multi-head self-attention mechanism layer. The number of heads in the multi-head self-attention mechanism is 8, and the dimension of each head is 64. The role of the self-attention mechanism is to calculate the correlation weights between different axes through scaled dot-product attention, thereby realizing the recombination of features. The self-attention mechanism calculates the attention weights through the following formula:.

[0050] ;

[0051] Among them, is the query matrix, is the key matrix, is the value matrix, is the dimension of the key vector. Through this formula, the correlation between different axes is obtained, and the features are weighted and recombined using the correlation to obtain the weighted convolutional feature vector. Temporal feature extraction is performed on the weighted convolutional feature vector to capture the long-term dependencies in the features. Input the temporal feature vector into the first fully connected layer. The fully connected layer contains 3 hidden layers, and the number of neurons is 256, 128, and 64 respectively. The role of the fully connected layer is to further integrate and abstract the features to extract the most representative feature information. Use a linear activation function in the last layer to map the features to the output space. The output dimension is N, corresponding to the predicted error values of N axes, and the error prediction data is obtained.

[0052] In a specific embodiment, the process of executing step S2 may specifically include the following steps:

[0053] Install high-precision grating scales on each axis of the multi-axis linkage machine tool. The resolution of the grating scale is 1 nm, and the sampling frequency is 10 kHz. Perform preliminary measurements on the positions of each axis to obtain the grating scale position data;

[0054] Fix the atom interferometer system on the machine tool reference plane. The atom interferometer system includes a frequency-stable laser light source, a beam splitter, a mirror, and a photodetector. The measurement resolution of the atom interferometer is 0.1 nm, and the sampling frequency is 100 kHz. Perform precise measurements on the positions of each axis relative to the reference plane to obtain the atom interferometer position data;

[0055] Synchronize the grating scale position data and the atom interferometer position data in time to obtain the synchronized position data, and perform wavelet transform on the synchronized position data to obtain multi-scale position features;

[0056] Construct a state space model based on the multi-scale position features. The state variables include position, velocity, and acceleration, and the observation variable is the position measurement value to obtain a linear state space representation;

[0057] Create a Kalman filter according to the linear state - space representation. The Kalman filter includes a state prediction equation and a measurement update equation. Initialize the state - estimation covariance matrix P and the process - noise covariance matrix Q to obtain the Kalman - filter parameters.

[0058] Input the synchronized position data into the Kalman filter. Through recursive calculations, perform state estimation and error - covariance update to obtain the filtered position estimate, and perform inverse wavelet transform on the filtered position estimate to reconstruct the nanoscale - resolution position data.

[0059] Specifically, install high - precision grating scales on each axis of the machine tool for preliminary position measurement. The resolution of the grating scale is 1 nanometer, and the sampling frequency is 10 kHz. Accurately measure the positions of each axis of the machine tool in real - time to obtain the grating - scale position data. As a high - precision position - measurement tool, the grating scale can provide position feedback for each axis, enabling the operating conditions of the machine tool to be accurately described. To obtain higher - precision data, fix an atomic interferometer system on the machine - tool reference plane for accurately measuring the positions of each axis relative to the reference plane. The system includes a frequency - stabilized laser light source, a beam splitter, a mirror, and a photodetector, with a measurement resolution of 0.1 nanometer and a sampling frequency of 100 kHz. Through the interference - measurement principle of the atomic interferometer, obtain high - precision position data. Synchronize the grating - scale position data and the atomic - interferometer position data to ensure that all position data are processed under the same time reference to obtain the synchronized position data, which contains the position information of the machine tool at each moment. Perform wavelet transform on the synchronized position data to extract multi - scale position features. Wavelet transform is a tool for analyzing the changes of a signal in the time and frequency domains, which can decompose the signal into detail and trend components at different scales. Assume the synchronized position data is , and the wavelet transform is expressed as:

[0060] ;

[0061] where, are wavelet coefficients, is the scale parameter, representing the resolution of the signal, is the translation parameter, representing the time position of the signal, is the mother wavelet function. Through wavelet transform, decompose the position data into multi - scale position features, effectively representing the dynamic changes of the machine tool at different time scales. Based on the multi - scale position features, construct a state - space model for describing the dynamic behavior of the system. In the state - space model, the state variables include position, velocity, and acceleration, while the observation variable is the position measurement value. The purpose of the state - space model is to describe the motion of the machine tool as a dynamic system, where the relationships between position, velocity, and acceleration are represented by a mathematical model. The state equation is expressed as:

[0062] ;

[0063] ;

[0064] wherein, is the state vector at time , including position, velocity and acceleration, is the state transition matrix, describing the change of state over time, is the control matrix, indicating the influence of the input of the system on the state, is the control input, is the process noise, representing the uncertainty of the model, is the observation vector, is the observation matrix, is the observation noise. Through the state space model, the motion process of the machine tool is described by a linear system. Based on the linear state space model, a Kalman filter is created to perform optimal estimation of the system state. The Kalman filter is a recursive algorithm, including a state prediction equation and a measurement update equation. The state prediction equation is used to estimate the state at the next moment based on the current state, and the measurement update equation is used to correct the prediction result by combining the measurement data. The state prediction and measurement update processes of the Kalman filter are expressed as the following two equations:

[0065] ;

[0066] ;

[0067] wherein, is the prediction of the state at the next moment, is the updated state estimate, is the Kalman gain, indicating the degree of correction of the measurement to the prediction. Initialize the state estimation covariance matrix and the process noise covariance matrix , so as to make a reasonable trade-off of the uncertainty of the state estimation in the recursive calculation. Input the synchronous position data into the Kalman filter, and perform state estimation and update of the error covariance through recursive calculation to obtain the filtered position estimation value. The Kalman filter combines the advantages of multi-source data, removes noise and reduces measurement errors at the same time, and obtains a more accurate position estimation value. Perform inverse wavelet transform on the filtered position estimation value to reconstruct the position data with nanoscale resolution. The inverse wavelet transform restores the features in the wavelet domain back to the time domain to obtain the original position data with high precision.

[0068] In a specific embodiment, the process of executing step S3 may specifically include the following steps:

[0069] Perform temporal alignment and normalization on multimodal sensor data and nanoscale resolution position data to obtain second standardized input data;

[0070] Divide the second standardized input data into four branches: temperature data, vibration data, stress data, and position data, and input them into four parallel feature extraction networks respectively. Each feature extraction network contains 3 one-dimensional convolutional layers, with the convolutional kernel sizes being 5, 3, 3 respectively, and the number of convolutional kernels being 32, 64, 128 respectively. After each layer, a BatchNormalization and a ReLU activation function are connected to obtain the primary feature vectors of each branch;

[0071] Perform global average pooling and global max pooling operations on the primary feature vectors of each branch to obtain the pooled feature vectors of each branch, and concatenate the pooled feature vectors of each branch to obtain the global feature vector;

[0072] Based on the global feature vector, calculate the correlation weights between different modalities through the scaled dot-product attention mechanism to obtain the attention-weighted features, and perform temporal feature extraction on the attention-weighted features to obtain the temporal context vector;

[0073] Input the temporal context vector into the cross-modal transformer encoder. The cross-modal transformer encoder contains 6 encoder layers, and each layer contains a multi-head self-attention sublayer and a feed-forward neural network sublayer to obtain the fused feature representation. Perform residual connection and layer normalization on the fused feature representation, and perform feature mapping through the non-linear activation function ReLU to obtain the intermediate feature representation;

[0074] Input the intermediate feature representation into the second fully connected layer. The second fully connected layer contains 3 hidden layers, with the number of neurons being 512, 256, 128 respectively. The Softmax activation function is used in the last layer to output the real-time state representation of the machine tool.

[0075] Specifically, time alignment and normalization are performed on multi-modal sensor data and position data with nanoscale resolution. The multi-modal sensor data includes information such as temperature, vibration, and stress, and the position data with nanoscale resolution provides precise spatial position information. Time alignment enables all data from different sources to have the same time reference to ensure the accuracy of subsequent analysis and eliminate biases caused by different sampling times. Normalization, on the other hand, transforms all features to the same dimension, giving them the same mean and variance, reducing the impact of differences in the numerical ranges of different features on model training. After completing time alignment and normalization, the second normalized input data is obtained. The second normalized input data is divided into four branches: temperature data, vibration data, stress data, and position data, which are respectively input into four parallel feature extraction networks. Each feature extraction network contains three one-dimensional convolutional layers with kernel sizes of 5, 3, and 3 respectively, and the number of convolutional kernels is 32, 64, and 128 respectively. The main purpose of the convolutional operation is to extract local features in the data. For example, in temperature data, it extracts the change pattern of temperature over time, or in vibration data, it captures specific vibration frequencies. The first convolutional kernel has a size of 5, which means the convolutional operation extracts features over a relatively long time span, while the subsequent convolutional kernels with a size of 3 capture more subtle time features. After each convolutional layer, BatchNormalization and the ReLU activation function are connected. BatchNormalization is used to normalize the output of the convolution to ensure the stability and efficiency of the model training process, and the ReLU activation function introduces non-linearity, enabling the network to learn more complex features. Through the processing of the convolutional layers, the primary feature vectors of each branch are obtained. Global average pooling and global max pooling operations are performed on the primary feature vectors of each branch. Global average pooling compresses the features by calculating the average value of each feature channel, while global max pooling takes the maximum value of each feature channel. Summarize the features at different scales, retain the most representative information, and reduce the dimension of the features to obtain the pooled feature vectors of each branch. The pooled feature vectors of each branch are concatenated to obtain the global feature vector. Based on the global feature vector, the correlation weights between different modalities are calculated through the scaled dot product attention mechanism to obtain the attention-weighted features. The scaled dot product attention mechanism calculates the similarity between features through the following formula:

[0076] ;

[0077] where, 、 and represent the query, key, and value matrices respectively, which are obtained by linearly transforming the input features; represents the dimension of the key vector, and by dividing the dot product result by Scale it to prevent the dot product value from becoming too large as the dimension increases. Convert the dot product result into weights through the softmax function, and then multiply it with the value matrix to obtain weighted features, enabling the model to focus on the internal relationships between different modality features and enhancing the attention to key features. Extract temporal features from the attention-weighted features to obtain a temporal context vector, capturing the evolution of each modality feature in the time dimension, especially the long-term dependencies of dynamic changes during the operation of the machine tool. Input the temporal context vector into the cross-modal transformer encoder. The cross-modal transformer encoder consists of 6 encoder layers, and each encoder layer contains a multi-head self-attention sub-layer and a feed-forward neural network sub-layer. The multi-head self-attention mechanism processes the input in parallel through multiple heads, and each head focuses on different aspects of the features, thereby enhancing the model's expressive ability for the input features. For each head, calculate the self-attention using the following formula:

[0078] ;

[0079] where represents the number of attention heads, is the linear transformation matrix of the output. By concatenating the outputs of all heads and then performing a linear transformation, a fused feature representation is obtained. Each encoder layer contains a feed-forward neural network sub-layer for performing linear transformation and non-linear mapping on the features output by the self-attention sub-layer. The encoder layer includes residual connections and layer normalization operations. The residual connection alleviates the problem of gradient vanishing in deep networks, ensuring that the model can still retain the original feature information at deeper levels, while layer normalization can improve the training stability and convergence speed. Through multiple processes of the encoder layer, features from different modalities are effectively fused to obtain a more comprehensive fused feature representation. After the fused feature representation is processed, it is then subjected to feature mapping through the non-linear activation function ReLU to obtain an intermediate feature representation. The intermediate feature representation contains the feature information of temperature, vibration, stress, and position data under multi-modal interaction, which is a high-level description of the current operating state of the machine tool. Input the intermediate feature representation into the second fully connected layer. This fully connected layer contains 3 hidden layers, and the number of neurons in each hidden layer is 512, 256, and 128 respectively. The input features are further abstracted layer by layer to extract the information most relevant to the machine tool state. The gradual reduction of the dimension in the hidden layer can reduce the computational amount while retaining the most meaningful features. The Softmax activation function is used in the last layer of the fully connected layer to map the intermediate features to the output space to obtain the real-time state representation of the machine tool. The Softmax activation function converts the feature values into a probability distribution, making the output represent the probabilities of different states, thereby helping operators or control systems to make real-time judgments and adjustments on the machine tool state.

[0080] In a specific embodiment, the process of executing step S4 may specifically include the following steps:

[0081] Divide the real-time state representation of the machine tool into three levels: the physical layer, the motion layer, and the function layer. Each level contains N nodes, where N is the number of machine tool axes, to obtain the first node set;

[0082] Assign attributes to the nodes in the first node set. The physical layer node attributes include temperature, vibration, and stress. The motion layer node attributes include position, velocity, and acceleration. The function layer node attributes include machining accuracy and motion synchronization, to obtain the second node set;

[0083] Construct a multi-level hypergraph structure based on the second node set. Calculate the similarity between nodes through the K-nearest neighbor algorithm. Set the similarity threshold to 0.8. Establish hyperedge connections between nodes with similarity greater than the threshold to obtain the initial hypergraph structure;

[0084] Calculate the intra-layer edge weights of the initial hypergraph structure. Use cosine similarity to calculate the correlation between node attributes and use the correlation as the edge weight to obtain the weighted intra-layer hypergraph;

[0085] Calculate the inter-layer edge weights of the weighted intra-layer hypergraph. Use the Pearson correlation coefficient to calculate the correlation between node attributes of different layers and use the correlation as the inter-layer edge weight to obtain the target multi-level hypergraph;

[0086] Input the target multi-level hypergraph into the hierarchical hypergraph model. The hierarchical hypergraph model contains 3 hypergraph convolutional layers. The output channel numbers of each hypergraph convolutional layer are 64, 128, and 256 respectively. The activation function is LeakyReLU to obtain the hierarchical features;

[0087] Perform cross-layer feature fusion on the hierarchical features to obtain the cross-layer fusion features, and input the cross-layer fusion features into the third fully connected layer. The third fully connected layer contains two hidden layers. The number of neurons is 128 and 64 respectively. The last layer uses the Tanh activation function and the output dimension is 32 to obtain the fault feature vector.

[0088] Specifically, the real-time state representation of the machine tool is divided into three levels: the physical layer, the motion layer, and the function layer. Each level contains N nodes, where N is the number of axes of the machine tool. These three levels reflect the operating states of the machine tool at different levels: the physical layer describes physical information related to the environment, such as temperature, vibration, and stress; the motion layer is related to the motion parameters of the machine tool, including position, speed, and acceleration; the function layer represents the machining performance of the machine tool, such as machining accuracy and motion synchronization between axes. Through layering, the first node set is obtained, and each node corresponds to the state of a certain axis at a certain level. Attribute assignment is performed on the nodes in the first node set. The node attributes of the physical layer include temperature, vibration, and stress, and these parameters directly affect the mechanical performance and stability of the machine tool; the node attributes of the motion layer include position, speed, and acceleration, which describe the dynamic behavior of each axis of the machine tool; while the node attributes of the function layer include machining accuracy and motion synchronization, and these attributes are related to the actual machining effect of the machine tool. By performing attribute assignment on the nodes, the second node set is obtained, which describes the state characteristics of the machine tool at the physical, motion, and function levels. Based on the second node set, a multi-level hypergraph structure is constructed. A hypergraph is a graph structure that can connect multiple nodes. Different from the traditional simple edge connection, the relationship between multiple nodes is described through hyperedge connection. To determine which nodes need to be connected by hyperedges, the K-nearest neighbor algorithm is used to calculate the similarity between nodes. The K-nearest neighbor algorithm can effectively find similar nodes in the attribute space. By setting the similarity threshold to 0.8, hyperedge connections are established between nodes with a similarity greater than the threshold, and the initial hypergraph structure is obtained. The level of similarity reflects the degree of similarity between node attributes. Through the hyperedge connection between nodes with high similarity, the linkage relationship between similar attributes in the machine tool state is captured. The initial hypergraph structure is processed to calculate the weights of the intra-layer edges, and the cosine similarity is used to calculate the correlation between node attributes. The formula for cosine similarity is as follows:

[0089] CosineSimilarity ;

[0090] where and are the attribute vectors of two nodes, represents the dot product of the vectors, and are the norms of the vectors respectively. Calculate the similarity between each pair of node attributes through cosine similarity, and use these similarities as the weights of the intra-layer edges to obtain a weighted intra-layer hypergraph. The value range of cosine similarity is between 0 and 1. The closer its value is to 1, the more similar the attributes between two nodes are. The weighted intra-layer hypergraph can more accurately describe the correlation of the machine tool state within the same level. Calculate the weights of the inter-layer edges for the weighted intra-layer hypergraph. Use the Pearson correlation coefficient to calculate the correlation between the node attributes of different layers. The formula for the Pearson correlation coefficient is:

[0091] ;

[0092] where, and are the attribute vectors of two nodes respectively, and are the means of the vectors respectively, is the dimension of the vector. Through the Pearson correlation coefficient, measure the linear correlation between the nodes of different layers. The value of the correlation coefficient is between -1 and 1. The closer its absolute value is to 1, the closer the relationship between two nodes is. Use the correlation as the weight of the inter-layer edge to obtain the target multi-level hypergraph. Input the target multi-level hypergraph into the hierarchical hypergraph model for feature extraction. The hierarchical hypergraph model contains 3 hypergraph convolutional layers. The output channel numbers of each hypergraph convolutional layer are 64, 128, and 256 respectively. The activation function of each layer is LeakyReLU. The role of the hypergraph convolutional layer is to perform a convolutional operation on the node features in the hypergraph to extract the high-order features between the nodes. Through hypergraph convolution, effectively aggregate the information between the nodes and hyperedges, and capture complex multi-node relationships. The LeakyReLU activation function is used to introduce non-linearity. The formula for LeakyReLU is:

[0093] ;

[0094] where, is a positive number less than 1, usually taking 0.01, It is used to handle the case where the input is negative. Through the activation function, it effectively avoids the "death" of neurons, enhances the non-linear expression ability of the network and the robustness of feature extraction. It performs cross-layer feature fusion on hierarchical features, integrating information from different levels to capture the interaction and influence of the machine tool state between different levels. Through the fusion operation, cross-layer fusion features are obtained. The cross-layer fusion features are input into the third fully connected layer, which contains two hidden layers. The number of neurons in the hidden layers is 128 and 64 respectively. The input features are abstracted and dimension-reduced layer by layer to extract the information most relevant to the fault features. The Tanh activation function is used in the last layer of the fully connected layer to map the features to the output space with an output dimension of 32, obtaining the fault feature vector. The output range of the Tanh activation function is from -1 to 1, making the feature values evenly distributed within a certain range. The formula is:

[0095] ;

[0096] Through Tanh activation, it ensures that the numerical values of the features are kept within a stable range, which helps subsequent feature analysis and fault judgment.

[0097] In a specific embodiment, the process of executing step S5 may specifically include the following steps:

[0098] The fault feature vector and the error prediction data are concatenated to obtain a state vector, and the state vector is normalized to obtain a normalized state input;

[0099] The normalized state input is input into the second input layer of the deep Q network. The second input layer is a fully connected layer with 256 neurons. The ReLU activation function is used to obtain an initial feature representation;

[0100] Feature extraction is performed on the initial feature representation. Deep feature learning is carried out through 3 residual blocks. Each residual block contains two convolutional layers and a shortcut connection. The convolutional kernel size is 3x3, and the output channel numbers are 128, 256, and 512 respectively to obtain deep features;

[0101] The deep features are input into the attention layer. The self-attention mechanism is used to calculate the correlation weights between features to obtain attention-weighted features, and temporal modeling is performed on the attention-weighted features to obtain temporal context features;

[0102] The temporal context features are input into the advantage function estimator, which contains two fully connected layers with 128 and 64 neurons respectively. The linear activation function is used in the last layer to output the Q value to obtain the action value estimation;

[0103] Based on action value estimation, the ε-greedy strategy is adopted to select compensation actions. The ε value is initially set to 0.9 and gradually decreased according to the exponential decay rule to obtain a sequence of compensation actions. Then, experience replay is performed on the sequence of compensation actions, and a batch of empirical samples are randomly sampled. The Q-network parameters are updated through temporal difference learning to obtain an adaptive compensation strategy.

[0104] Specifically, the fault feature vector and error prediction data are concatenated to obtain a state vector. The state vector integrates the fault information and error prediction of the system, and can comprehensively reflect the current operating state and potential deviations of the machine tool. The state vector is normalized to standardize all data, eliminate the dimensional differences between different features, and make each feature within the same scale range, improving the training stability and efficiency of the subsequent deep neural network to obtain a standardized state input. The standardized state input is fed into the second input layer of the deep Q-network, which is a fully connected layer containing 256 neurons and uses the ReLU activation function. The role of the fully connected layer is to perform linear transformation and non-linear activation on the input state vector to obtain an initial feature representation. By using the ReLU activation function, non-linear features are introduced, enabling the model to learn the complex relationship between states and actions. The definition of the ReLU activation function is as follows:

[0105] ;

[0106] Among them, is the input feature, is the output feature. When the input is greater than 0, the output is equal to the input itself, and when the input is less than or equal to 0, the output is 0. The use of ReLU effectively solves the problem of gradient disappearance in traditional activation functions and enhances the learning ability of deep neural networks for complex features. Deep feature extraction is performed on the initial feature representation, and 3 residual blocks are used to achieve deep feature learning. Each residual block contains two convolutional layers and a shortcut connection, thereby maintaining information flow in the deep network and preventing the problem of gradient disappearance caused by excessive depth. The convolutional kernel size of the convolutional layer is 3x3, and the number of output channels is 128, 256, and 512 respectively. The gradually increasing number of output channels can capture more feature details. The residual block adds the input directly to the output through a shortcut connection, and the formula is as follows:

[0107] ;

[0108] Among them, represents the input feature, represents the weight of the convolutional layer, represents the output feature after passing through two convolutional layers, is the output of the residual block. Through the shortcut connection, the model retains the feature information of the input, reducing the risk of gradient vanishing, and thus can effectively learn deep features. The deep features are input into the attention layer, and the self-attention mechanism is used to calculate the correlation weights between features. The self-attention mechanism performs global correlation calculation on the input features, enabling the model to focus on the important relationships between features. The calculation formula for the attention weights is as follows:

[0109] ;

[0110] where, , and represent the query, key, and value matrices respectively. These matrices are obtained by linear transformation of the input features, represents the dimension of the key vector, which is used to scale the dot product result to prevent the dot product value from becoming too large as the dimension increases. The softmax function converts the dot product result into a probability distribution to obtain the correlation weights between features, and weights the features according to the weights to finally obtain the attention-weighted features. Temporal modeling is performed on the attention-weighted features to capture the time-dependent relationship of the machine tool state, obtaining temporal context features, enabling the model to consider the dynamic changes of the machine tool state over time, especially long-term trend changes, and better predicting and compensating for the errors of the machine tool. The temporal context features are input into the advantage function estimator, which contains two fully connected layers with the number of neurons being 128 and 64 respectively. The fully connected layers are used to abstract the input features to extract the most important information related to the compensation action. A linear activation function is used in the last layer to output the Q values corresponding to different actions, and the formula is as follows:

[0111] ;

[0112] where, represents the value of executing action in state , represents the state value function, represents the advantage function, which is used to estimate the advantage of a certain action relative to other actions in a specific state. In this way, the model evaluates the values of different compensation actions to obtain the action value estimate. Based on the action value estimate, an -greedy strategy is used to select the compensation action. The core idea of the As the value gradually decreases, the model more often selects the currently known optimal action, thus achieving the transition from exploration to exploitation. Specifically, - The selection rule of the greedy strategy is expressed as:

[0113] ;

[0114] where, represents the action selected at time , is the value estimate of different actions in the current state . Through this strategy, a compensated action sequence is obtained. The compensated action sequence is subjected to experience replay. The compensated actions and the corresponding state changes are saved, and a batch of empirical samples are randomly sampled during the training process for training. In this way, the temporal correlation of the data is broken, the independence of the data is increased, and the generalization ability of the model and the stability of the training are improved. During the experience replay process, the parameters of the deep Q-network are updated through temporal difference learning. Temporal difference learning combines the advantages of dynamic programming and Monte Carlo methods, and gradually adjusts the parameters during the estimation of the state, enabling the model to more accurately predict future states and action rewards. The specific temporal difference update formula is as follows:

[0115] ;

[0116] where, is the learning rate, which controls the step size of parameter update; is the immediate reward after executing the action at time ; is the discount factor, which measures the importance of future rewards; is the next state reached after executing the action; represents the value estimate of the optimal action in the next state. Through this recursive calculation, the strategy of the deep Q-network is gradually improved to obtain a more accurate adaptive compensation strategy.

[0117] In a specific embodiment, the process of executing step S6 may specifically include the following steps:

[0118] Input the adaptive compensation strategy into the compensation value generator, obtain the initial compensation values of each axis through inverse kinematics calculation, and perform non-linear optimization on the initial compensation values of each axis to obtain the coupled-axis compensation values of each axis;

[0119] Based on the coupled-axis compensation values of each axis, the multi-axis linkage machine tool is adjusted in real time, and the operation data of the adjusted machine tool is collected. The fast Fourier transform is performed on the operation data of the adjusted machine tool to obtain the frequency domain characteristics;

[0120] Input the frequency-domain features into a multi-class support vector machine for fault prediction. The radial basis function is selected as the kernel function to obtain the probability distribution of fault types. Then, a threshold judgment is made on the probability distribution of fault types. Set the fault warning threshold to 0.8. When the probability of a certain type of fault exceeds the threshold, a fault warning signal is triggered.

[0121] Based on the fault warning signal, start the fault traceability analysis program, construct a fault causality graph, locate the fault root cause through probability reasoning to obtain the fault traceability result. And according to the fault traceability result, use digital twin technology to construct a virtual model of the machine tool, execute and optimize the fault prevention plan in the virtual environment to obtain the fault prevention control plan.

[0122] Specifically, input the adaptive compensation strategy into the compensation value generator, and obtain the initial compensation values of each axis through inverse kinematic calculation. Inverse kinematics refers to calculating the motion parameters of each axis according to the expected end state to obtain appropriate compensation values for each axis. Assume that the end state of the machine tool is given by the target position The calculation process of inverse kinematics is described as:

[0123] ;

[0124] Among them, is the vector of angle or displacement compensation values for each axis, is the inverse kinematic solution function, is the expected end position information. Through inverse kinematic solution, the initial compensation value of each axis is obtained. Perform non-linear optimization on the initial compensation value. Considering the system constraints, further adjust the initial compensation value so that the final compensation value can achieve the optimal correction effect. Non-linear optimization is achieved by minimizing a certain error function, and the error function is defined as the sum of the squares of the errors between the actual position and the expected position:

[0125] ;

[0126] Among them, is the error function, is the actual position of the th axis, is the expected position, The number of axes. By optimizing the error function, the compensation values for each axis that minimize the error are found, and finally the coupled-axis compensation values for each axis are obtained. These compensation values are used to adjust the movement of each axis, making the end state of the machine tool closer to the desired position and improving the machining accuracy. Based on the coupled-axis compensation values of each axis, the multi-axis machine tool is adjusted in real time to ensure the accuracy and stability of the machine tool during operation. During the real-time adjustment process, the operating data of the adjusted machine tool is collected simultaneously to monitor the compensation effect and the operating state of the machine tool. The collected data includes the displacement, speed, vibration signal, etc. of each axis. To analyze the frequency characteristics during the operation of the machine tool, the fast Fourier transform is performed on the adjusted operating data to transform the data from the time domain to the frequency domain and extract the frequency domain characteristics. The formula for the fast Fourier transform is:

[0127] ;

[0128] where, represents the frequency domain signal with frequency , is the time domain signal of the th sampling point, is the total number of sampling points. Through the fast Fourier transform, the frequency spectrum of the signal is obtained, thereby analyzing the vibration frequency and other frequency domain characteristics during the operation of the machine tool. The frequency domain characteristics are input into a multi-class support vector machine (SVM) for fault prediction. The multi-class support vector machine is a classification tool that can classify the input frequency domain characteristic data to determine whether there is a fault in the machine tool and the possible fault types. The radial basis function is selected as the kernel function, and the expression of the radial basis function is:

[0129] ;

[0130] where, is the similarity between the input feature vectors and , is a parameter of the kernel function, which is used to control the range of similarity. Through the radial basis function, the support vector machine can map the input features to a high-dimensional space, find the optimal hyperplane in this space to separate different fault types, and obtain the probability distribution of fault types. A threshold judgment is performed on the probability distribution of fault types, and the threshold for fault warning is set to 0.8. When the probability of a certain type of fault exceeds this threshold, the system will trigger a fault warning signal. The fault warning signal can timely remind the operator or control system that there is a certain serious fault in the machine tool and immediate measures need to be taken for treatment. When the fault warning signal is triggered, start the fault traceability analysis program to find the root cause of the fault. Construct a fault causal relationship graph. The fault causal relationship graph is a graph structure used to describe the causal relationship between variables in the system, where nodes represent variables in the system and edges represent the causal relationship between variables. Through this graph structure, describe the propagation path of the fault. Based on the fault causal relationship graph, use the probability reasoning method to locate the root cause of the fault. Probability reasoning is a reasoning method based on Bayes' theorem. By calculating the probabilities of different fault paths, find the most likely fault cause. According to the fault traceability result, use digital twin technology to construct a virtual model of the machine tool. Digital twin is a simulation technology that constructs a digital model corresponding to the actual machine tool in a virtual environment and truly simulates the operating behavior of the machine tool in different states. In the virtual environment, based on the result of fault traceability, execute and optimize the fault prevention plan. This process includes simulating different preventive measures in the virtual model, analyzing their effects, and selecting the optimal plan. In this way, without affecting the operation of the actual machine tool, fully verify and optimize the fault prevention plan to obtain the fault prevention control plan.

[0131] The calibration method of the precision manufacturing multi-axis linkage machine tool in the embodiment of the present invention has been described above. Next, the calibration device of the precision manufacturing multi-axis linkage machine tool in the embodiment of the present invention will be described. Please refer to Figure 2 , an embodiment of the calibration device of the precision manufacturing multi-axis linkage machine tool in the embodiment of the present invention includes:

[0132] An acquisition module, configured to acquire the machine tool operation data and multi-modal sensor data of the multi-axis linkage machine tool, and perform error prediction on the machine tool operation data through a multi-axis linkage error prediction model to obtain error prediction data;

[0133] A measurement module, configured to measure the positions of each axis of the machine tool by using an atomic interferometer and a grating scale, and obtain nanometer-level resolution position data through Kalman filter processing;

[0134] A fusion module, configured to input the multi-modal sensor data and the nanometer-level resolution position data into a multi-branch deep neural network, and perform fusion through an attention mechanism to obtain a real-time state representation of the machine tool;

[0135] A feature processing module, which is used to construct a multi-level hypergraph based on the real-time state representation of the machine tool, and input the multi-level hypergraph into a hierarchical hypergraph model for feature processing to obtain a fault feature vector;

[0136] A reinforcement learning module, which is used to take the fault feature vector and error prediction data as state inputs, and perform reinforcement learning through a deep Q-network to obtain an adaptive compensation strategy;

[0137] A generation module, which is used to generate the coupled-axis compensation values of each axis based on the adaptive compensation strategy, perform real-time adjustment and fault prevention analysis on the multi-axis linkage machine tool, and generate a fault prevention control plan.

[0138] Through the collaborative cooperation of the above-mentioned various components, through the fusion of multi-modal sensor data and nanometer-level resolution position data, combined with the multi-head self-attention mechanism, the high-precision perception and feature extraction of the machine tool state are realized, and the accuracy and real-time performance of state monitoring are improved. The multi-level hypergraph structure is used to model the physical layer, motion layer and function layer of the machine tool, and feature processing is performed through the hierarchical hypergraph model, effectively capturing the correlation relationships between different levels and enhancing the expression ability of fault features. Based on the reinforcement learning method of the deep Q-network, the dynamic optimization of the adaptive compensation strategy is realized, enabling the compensation value to be adaptively adjusted according to the real-time state of the machine tool, and significantly improving the compensation effect. By constructing a virtual model through digital twin technology and optimizing the fault prevention plan in the virtual environment, the early warning and active prevention of faults are realized, the equipment maintenance cost is reduced, and the service life of the machine tool is extended. The present invention enhances the accuracy retention ability and operation reliability of the multi-axis linkage machine tool and can be applied to different types of multi-axis linkage machine tools.

[0139] Referring to Figure 3 , in the embodiment of the present invention, a computer device is further provided. The computer device may be a server, and its internal structure may be as Figure 3 shown. The computer device includes a processor, a memory, a display screen, an input device, a network interface and a database connected through a system bus. Among them, the processor of the computer design is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store the corresponding data in this embodiment. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, the above method is implemented.

[0140] Those skilled in the art can understand, Figure 3The structure shown is only a block diagram of some structures related to the solution of the present invention, and does not constitute a limitation on the computer device to which the solution of the present invention is applied.

[0141] An embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the above method is implemented. It can be understood that the computer-readable storage medium in this embodiment can be a volatile readable storage medium or a non-volatile readable storage medium.

[0142] Those of ordinary skill in the art can understand that all or part of the processes in the above method embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the above method embodiments. Among them, any reference to a memory, storage, database, or other medium provided by the present invention and used in the embodiments can include non-volatile and / or volatile memories. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM, etc.

[0143] Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the above-described systems, systems, and units can refer to the corresponding processes in the foregoing method embodiments, and will not be described herein again.

[0144] When the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as USB flash drives, mobile hard disks, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical discs.

[0145] As described above, the above embodiments are only used to illustrate the technical solution of the present invention, rather than to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A precision manufacturing multi-axis linkage machine tool calibration method, characterized in that: The method comprises: Collecting machine tool operation data and multi-modal sensor data of a multi-axis linkage machine tool, and performing error prediction on the machine tool operation data through a multi-axis linkage error prediction model to obtain error prediction data; The position of each axis of the machine tool is measured using an atomic interferometer and a grating ruler, and the position data with nanometer resolution is obtained through Kalman filtering. Inputting the multimodal sensor data and the nanometer-level resolution position data into a multi-branch deep neural network, fusing them through an attention mechanism, and obtaining a real-time state representation of the machine tool; Constructing a multi-level hypergraph based on the real-time state representation of the machine tool, and inputting the multi-level hypergraph into a hierarchical hypergraph model for feature processing to obtain a fault feature vector; Taking the fault feature vector and the error prediction data as state input, performing reinforcement learning through a deep Q network to obtain an adaptive compensation strategy; Based on the adaptive compensation strategy, the linkage compensation value of each axis is generated, and the multi-axis linkage machine tool is adjusted in real time and fault prevention analysis is performed to generate a fault prevention control plan.

2. The precision manufacturing multi-axis linkage machine tool calibration method according to claim 1, characterized in that: The collecting of machine tool operation data and multi-modal sensor data of the multi-axis linkage machine tool, and performing error prediction on the machine tool operation data through a multi-axis linkage error prediction model to obtain error prediction data, includes: The real-time position, speed and acceleration of each axis of the multi-axis linkage machine tool are collected to obtain the machine tool operation data. The temperature distribution, vibration characteristics and stress state of the machine tool are collected through the temperature sensor array, three-axis acceleration sensor and strain gauge to obtain multi-modal sensor data. Performing time synchronization on the machine tool operation data and the multimodal sensor data, segmenting the data using a sliding window method, with a window size of 1 second and an overlap rate of 50%, and then standardizing the data in each window to obtain first standardized input data; Input the first standardized input data into the first input layer of the multi-axis linkage error prediction model for one-dimensional convolution processing to obtain a first convolution feature vector, wherein the first input layer includes N parallel one-dimensional convolutional neural networks, where N is the number of machine tool axes, and each one-dimensional convolutional neural network includes 3 convolutional layers, and the convolution kernel sizes are 5, 3, and 3 respectively, and the number of convolution kernels is 32, 64, and 128 respectively, and each convolution layer is followed by BatchNormalization and ReLU activation functions; Extract features from the first convolution feature vector, perform cross-axis feature fusion through two three-dimensional convolution layers, the convolution kernel size is 3x3x3, the number of convolution kernels is 256 and 512 respectively, and each layer is followed by MaxPooling and Dropout operations to obtain a second convolution feature vector; The second convolution feature vector is input into a multi-head self-attention mechanism layer, with 8 heads and 64 dimensions of each head. The correlation weights between different axes are calculated by scaling the dot product attention and the features are reorganized to obtain a weighted convolution feature vector. The time series features and long-term dependencies of the weighted convolution feature vector are captured to obtain a time series feature vector, and the time series feature vector is input into the first fully connected layer. The first fully connected layer includes 3 hidden layers with 256, 128, and 64 neurons respectively. The last layer uses a linear activation function with an output dimension of N, corresponding to the prediction error values ​​of N axes, to obtain error prediction data.

3. The precision manufacturing multi-axis linkage machine tool calibration method according to claim 2, characterized in that: The atomic interferometer and grating ruler are used to measure the position of each axis of the machine tool, and the nanometer-level resolution position data is obtained through Kalman filtering, including: A high-precision grating ruler is installed on each axis of the multi-axis linkage machine tool. The grating ruler resolution is 1nm and the sampling frequency is 10kHz. The position of each axis is preliminarily measured to obtain the grating ruler position data. An atom interferometer system is fixed on the machine tool reference surface, wherein the atom interferometer system includes a frequency-stable laser light source, a beam splitter, a reflector and a photodetector. The measurement resolution of the atom interferometer is 0.1 nm, and the sampling frequency is 100 kHz. The position of each axis relative to the reference surface is accurately measured to obtain the position data of the atom interferometer; Performing time synchronization on the grating ruler position data and the atomic interferometer position data to obtain synchronized position data, and performing wavelet transformation on the synchronized position data to obtain multi-scale position features; A state space model is constructed based on the multi-scale position feature, where the state variables include position, velocity and acceleration, and the observation variable is the position measurement value, to obtain a linear state space representation; Creating a Kalman filter according to the linear state space representation, wherein the Kalman filter includes a state prediction equation and a measurement update equation, initializing a state estimation covariance matrix P and a process noise covariance matrix Q, and obtaining Kalman filter parameters; The synchronized position data is input into the Kalman filter, and state estimation and error covariance update are performed through recursive calculation to obtain a filtered position estimate value, and an inverse wavelet transform is performed on the filtered position estimate value to reconstruct the nanometer-level resolution position data.

4. The precision manufacturing multi-axis linkage machine tool calibration method according to claim 3 is characterized in that: The step of inputting the multimodal sensor data and the nanometer-level resolution position data into a multi-branch deep neural network and fusing them through an attention mechanism to obtain a real-time state representation of the machine tool includes: Performing time alignment and normalization processing on the multimodal sensor data and the nanometer-level resolution position data to obtain second standardized input data; The second standardized input data is divided into four branches: temperature data, vibration data, stress data, and position data, and input into four parallel feature extraction networks respectively. Each feature extraction network contains three one-dimensional convolutional layers, and the convolution kernel sizes are 5, 3, and 3 respectively. The number of convolution kernels is 32, 64, and 128 respectively. Each layer is followed by BatchNormalization and ReLU activation functions to obtain the primary feature vector of each branch. Perform global average pooling and global maximum pooling operations on the primary feature vector of each branch to obtain the pooled feature vector of each branch, and concatenate the pooled feature vectors of each branch to obtain the global feature vector; Based on the global feature vector, the correlation weights between different modalities are calculated by a scaled dot product attention mechanism to obtain an attention weighted feature, and temporal feature extraction is performed on the attention weighted feature to obtain a temporal context vector; Inputting the temporal context vector into a cross-modal transformer encoder, the cross-modal transformer encoder comprising 6 encoder layers, each layer comprising a multi-head self-attention sublayer and a feed-forward neural network sublayer, obtaining a fused feature representation, performing residual connection and layer normalization on the fused feature representation, and performing feature mapping through a nonlinear activation function ReLU to obtain an intermediate feature representation; The intermediate feature representation is input into the second fully connected layer, which includes three hidden layers with 512, 256 and 128 neurons respectively. The last layer uses the Softmax activation function to output the real-time status representation of the machine tool.

5. The precision manufacturing multi-axis linkage machine tool calibration method according to claim 4, characterized in that: The method of constructing a multi-level hypergraph based on the real-time state representation of the machine tool and inputting the multi-level hypergraph into a hierarchical hypergraph model for feature processing to obtain a fault feature vector includes: The real-time state representation of the machine tool is divided into three levels: a physical layer, a motion layer, and a functional layer, each level includes N nodes, where N is the number of axes of the machine tool, to obtain a first node set; Assigning attributes to the nodes in the first node set, where the physical layer node attributes include temperature, vibration and stress, the motion layer node attributes include position, velocity and acceleration, and the functional layer node attributes include processing accuracy and motion synchronization, to obtain a second node set; Building a multi-level hypergraph structure based on the second node set, calculating the similarity between nodes by using a K-nearest neighbor algorithm, setting a similarity threshold to 0.8, establishing hyperedge connections between nodes with similarities greater than the threshold, and obtaining an initial hypergraph structure; Calculating the intra-layer edge weights of the initial hypergraph structure, using cosine similarity to calculate the correlation between node attributes, and using the correlation as the edge weight to obtain a weighted intra-layer hypergraph; Calculating the inter-layer edge weights of the weighted intra-layer hypergraph, using the Pearson correlation coefficient to calculate the correlation between the attributes of nodes in different layers, and using the correlation as the inter-layer edge weight to obtain the target multi-level hypergraph; Input the target multi-level hypergraph into a hierarchical hypergraph model, wherein the hierarchical hypergraph model comprises three hypergraph convolutional layers, the number of output channels of each hypergraph convolutional layer is 64, 128 and 256 respectively, and the activation function is LeakyReLU, to obtain hierarchical features; The hierarchical features are fused across layers to obtain cross-layer fusion features, and the cross-layer fusion features are input into the third fully connected layer, wherein the third fully connected layer includes two hidden layers with 128 and 64 neurons respectively. The last layer uses a Tanh activation function with an output dimension of 32 to obtain a fault feature vector.

6. The precision manufacturing multi-axis linkage machine tool calibration method according to claim 5, characterized in that: The method uses the fault feature vector and the error prediction data as state inputs and performs reinforcement learning through a deep Q network to obtain an adaptive compensation strategy, including: splicing the fault feature vector and the error prediction data to obtain a state vector, and normalizing the state vector to obtain a standardized state input; Input the standardized state into the second input layer of the deep Q network, where the second input layer is a fully connected layer with 256 neurons and a ReLU activation function to obtain an initial feature representation; Extracting features from the initial feature representation, performing deep feature learning through three residual blocks, each residual block comprising two convolutional layers and a short-circuit connection, a convolution kernel size of 3x3, and output channel numbers of 128, 256, and 512, respectively, to obtain deep features; Input the deep features into the attention layer, use the self-attention mechanism to calculate the correlation weights between the features to obtain attention-weighted features, and perform time series modeling on the attention-weighted features to obtain time series context features; The temporal context features are input into the advantage function estimator, which includes two fully connected layers with 128 and 64 neurons respectively. The last layer uses a linear activation function and outputs a Q value to obtain an action value estimate; Based on the action value estimation, the ε-greedy strategy is adopted to select the compensation action. The ε value is initially set to 0.9 and gradually reduced according to the exponential decay rule to obtain a compensation action sequence. The compensation action sequence is then empirically replayed, batch experience samples are randomly sampled, and the Q network parameters are updated through temporal difference learning to obtain an adaptive compensation strategy.

7. The precision manufacturing multi-axis linkage machine tool calibration method according to claim 6, characterized in that: The method of generating the linkage compensation value of each axis based on the adaptive compensation strategy, performing real-time adjustment and fault prevention analysis on the multi-axis linkage machine tool, and generating a fault prevention control plan includes: The adaptive compensation strategy is input into the compensation value generator, the initial compensation value of each axis is obtained by inverse kinematics calculation, and the initial compensation value of each axis is nonlinearly optimized to obtain the linkage compensation value of each axis; Based on the linkage compensation values ​​of each axis, the multi-axis linkage machine tool is adjusted in real time, and the adjusted machine tool operation data is collected, and the adjusted machine tool operation data is fast Fourier transformed to obtain frequency domain features; The frequency domain features are input into a multi-class support vector machine for fault prediction. The radial basis function is selected as the kernel function to obtain the probability distribution of the fault type, and a threshold judgment is performed on the probability distribution of the fault type. The fault warning threshold is set to 0.

8. When the probability of a certain type of fault exceeds the threshold, a fault warning signal is triggered; Based on the fault warning signal, a fault tracing analysis program is started, a fault cause-effect relationship diagram is constructed, the root cause of the fault is located through probabilistic reasoning, and a fault tracing result is obtained. Based on the fault tracing result, a machine tool virtual model is constructed using digital twin technology, and a fault prevention plan is executed and optimized in a virtual environment to obtain a fault prevention control plan.

8. A precision manufacturing multi-axis linkage machine tool calibration device, characterized in that: The device is used to perform the precision manufacturing multi-axis linkage machine tool calibration method according to any one of claims 1 to 7, and comprises: An acquisition module is used to acquire machine tool operation data and multi-modal sensor data of a multi-axis linkage machine tool, and to perform error prediction on the machine tool operation data through a multi-axis linkage error prediction model to obtain error prediction data; The measurement module is used to measure the position of each axis of the machine tool using an atomic interferometer and a grating ruler, and obtain nanometer-level resolution position data through Kalman filtering; A fusion module, used for inputting the multimodal sensor data and the nanometer-level resolution position data into a multi-branch deep neural network, and fusing them through an attention mechanism to obtain a real-time state representation of the machine tool; A feature processing module, used for constructing a multi-level hypergraph based on the real-time state representation of the machine tool, and inputting the multi-level hypergraph into a hierarchical hypergraph model for feature processing to obtain a fault feature vector; A reinforcement learning module, used for taking the fault feature vector and the error prediction data as state inputs, performing reinforcement learning through a deep Q network, and obtaining an adaptive compensation strategy; A generation module is used to generate the linkage compensation value of each axis based on the adaptive compensation strategy, and to perform real-time adjustment and fault prevention analysis on the multi-axis linkage machine tool to generate a fault prevention control plan.

9. A computer device, characterized in that: It comprises a memory and a processor, wherein the memory stores a computer program that can be run on the processor, and is characterized in that when the processor executes the computer program, the precision manufacturing multi-axis linkage machine tool calibration method described in any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the processor is enabled to execute the precision manufacturing multi-axis linkage machine tool calibration method according to any one of claims 1 to 7.

Images