Ultra-precision cutting machine tool processing product index prediction analysis method and system thereof
Patent Information
- Application Number
- CN202611084703.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-07-21
- Publication Date
- 2026-09-25
AI Technical Summary
[0003]现有技术中，对于超精密加工机床的运行参数调整一般都是通过采集机床的相关数据，然后由人工根据实际经验来判断当前数据状态下工件表面的如粗糙度等指标的发展趋势，然后反向来根据判断结果来调整机床工艺参数，或者根据实际产品的粗糙度等指标的监测结果来调整机床的工艺参数，上述的两种方式，第一种主管因素极强，而且完全受限于工作人员的经验，从而导致工艺参数调整反复，最终产品质量并不能得到保障，次品率高，而后一种以产品的粗糙度等指标检测后再进行调整，具有滞后性，并且同样需要反复调整，效率低，次品率高，导致成本增加
[0045]本发明的有益效果：通过本发明，通过建立相应的产品指标预测模型并对其完成训练后，实时获取相应的机床运行参数来进行产品指标预测，再通过预测结果来指导机床工艺参数的调整，这个过程中能够实时跟踪机床的运行状态，从而确保工艺参数调整的准确性，避免传统技术中主观因素的影响，而且具有较高的效率，能够有效确保最终切削产品表面的质量，降低因为工艺参数调整所带来的额外成本。
Smart Images
Figure CN122820006A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to a method for predicting the indicators of machined products, and more particularly to a method and system for predicting and analyzing the indicators of products machined by ultra-precision cutting machine tools. Background Technology
[0002] As core equipment in the field of advanced manufacturing, how to achieve stable operation and high-precision machining of ultra-precision machine tools under complex working conditions has become a key technical problem that urgently needs to be solved. This is because machine tools are affected by a variety of factors during actual operation, such as spindle vibration, ambient temperature fluctuations, and changes in cutting force. These factors can cause deviations in process parameters, which in turn affect the contour accuracy and surface roughness of the workpiece. Therefore, it is necessary to adjust the process parameters of the machine tool during the machining process to meet the requirements of the workpiece product specifications.
[0003] In existing technologies, the adjustment of operating parameters for ultra-precision machine tools generally involves collecting relevant machine tool data, then manually judging the development trend of workpiece surface indicators such as roughness based on practical experience, and then adjusting the machine tool process parameters accordingly. Alternatively, the process parameters can be adjusted based on the monitoring results of the roughness and other indicators of the actual product. The first method is highly subjective and entirely dependent on the experience of the operators, leading to repeated adjustments of process parameters and ultimately failing to guarantee product quality, resulting in a high defect rate. The second method, which adjusts the parameters after detecting the roughness and other indicators of the product, is lagging and also requires repeated adjustments, resulting in low efficiency, a high defect rate, and increased costs.
[0004] Therefore, in order to solve the above-mentioned technical problems, it is urgent to propose a new technical approach. Summary of the Invention
[0005] In view of this, the purpose of this invention is to provide a method and system for predicting and analyzing the product indicators of ultra-precision cutting machine tools. By establishing a corresponding product indicator prediction model and training it, the corresponding machine tool operating parameters are acquired in real time to predict the product indicators. The prediction results are then used to guide the adjustment of machine tool process parameters. In this process, the machine tool's operating status can be tracked in real time, thereby ensuring the accuracy of process parameter adjustments, avoiding the influence of subjective factors in traditional technologies, and having high efficiency. It can effectively ensure the quality of the final machined product surface and reduce the additional costs caused by process parameter adjustments.
[0006] This invention provides a method and system for predicting and analyzing the performance indicators of products machined by ultra-precision cutting machine tools, comprising the following steps:
[0007] S1. Obtain the operating parameters of the ultra-precision cutting machine tool, the operating parameters including dynamic parameters and static parameters;
[0008] S2. Preprocess the operating parameters of the ultra-precision cutting machine tool;
[0009] S3. Construct a product indicator prediction model and input the preprocessed running parameters into the product indicator prediction model for training;
[0010] S4. Obtain the real-time operating parameters of the ultra-precision cutting machine tool, and input the preprocessed real-time operating parameters into the trained product index prediction model to obtain the product index prediction results.
[0011] Furthermore, the product indicator prediction model includes a feature extraction unit, a Transformer network, a feature extraction module, an average pooling module, an ECA module, and an output module;
[0012] The feature extraction unit inputs the preprocessed operating parameters, and the output features of the feature extraction unit are input into the Transformer network for processing. The output features of the Transformer network are input into the feature extraction module, and the output features of the feature extraction module are input into the average pooling module. The output features of the average pooling module are input into the ECA module, and the output features of the ECA module are input into the output module and processed by the output module to output the product indicator prediction result.
[0013] Furthermore, the feature extraction unit includes a first extraction module, a first max pooling module, a second extraction module, and a second max pooling module;
[0014] The input terminal of the first extraction module serves as the input terminal of the feature extraction unit. The output features of the first extraction module are input to the first max pooling module. The output features of the first max pooling module are input to the second extraction module. The output features of the second extraction module are input to the second max pooling module. The output terminal of the second max pooling module serves as the output terminal of the feature extraction unit.
[0015] Furthermore, the first extraction module and the second extraction module have the same structure;
[0016] The first extraction module includes convolution module I, convolution module II, convolution module III, global average pooling module I, global average pooling module II, ReLU activation function module I, ReLU activation function module II, and SiLU activation function module;
[0017] The input of convolution module I serves as the input of the first extraction module. The output of convolution module I is connected to the input of ReLU activation function module I. The output of ReLU activation function module I is connected to the input of global average pooling module I. The output of global average pooling module I is connected to the input of convolution module II. The output of convolution module II is connected to the input of ReLU activation function module II. The output of ReLU activation function module II is connected to the input of convolution module III. The output of convolution module III is connected to the input of SiLU activation function module. The output of SiLU activation function module serves as the output of the first extraction module.
[0018] Among them, the convolution scales of convolution module I, convolution module II, and convolution module III are 3×3, 5×5, and 7×7, respectively.
[0019] Furthermore, the feature extraction module includes convolution module IV, convolution module V, convolution module VI, batch normalization module, and SiLU activation function module I;
[0020] The input of convolution module IV serves as the input of the first extraction module. The output of convolution module IV is connected to the input of convolution module V. The output of convolution module V is connected to the input of the batch normalization module. The output of the batch normalization module is connected to the input of SiLU activation function module I. The output of SiLU activation function module I is connected to the input of convolution module VI. The output of convolution module VI serves as the output of the first extraction module.
[0021] Among them, the convolution scales of convolution module I, convolution module II and convolution module III are 1×1, 3×3 and 1×1 respectively.
[0022] Furthermore, the output module includes a fully connected layer and a softmax activation function module;
[0023] The input of the fully connected layer is connected to the input of the ECA module, and the output of the fully connected layer is connected to the input of the softmax activation function module. The softmax activation function module outputs the product indicator prediction results.
[0024] Furthermore, the static parameters include machine tool guideway positioning accuracy, repeatability, and spindle rotation accuracy;
[0025] The dynamic parameters include cutting tool temperature, machine tool component vibration data, acoustic emission signals during the cutting process, and cutting force.
[0026] Furthermore, the preprocessing of the operating parameters includes:
[0027] Denoising is used to remove abnormal data from operating parameters;
[0028] The dynamic parameters in the denoised operating parameters are transformed using Morlet wavelets.
[0029] Accordingly, the present invention also provides a predictive analysis system for product indicators of ultra-precision cutting machine tools, including a sensor module and a processing module;
[0030] The sensor module is used to acquire the operating parameters of the ultra-precision cutting machine tool and input them to the processing module; the sensor module includes a temperature sensor, a cutting force sensor, a vibration sensor, and an acoustic emission sensor.
[0031] The processing module is used to receive the operating parameters output by the sensor module and output the product indicator prediction results.
[0032] The processing module includes a preprocessing module and a product indicator prediction unit;
[0033] The preprocessing module is used to denoise the operating parameters and then perform Morlet wavelet transform processing. The output data of the preprocessing module is input to the product indicator prediction unit.
[0034] The product indicator prediction unit includes a feature extraction unit, a Transformer network, a feature extraction module, an average pooling module, an ECA module, and an output module.
[0035] The feature extraction unit inputs preprocessed operating parameters, the output features of the feature extraction unit are input into the Transformer network for processing, the output features of the Transformer network are input into the feature extraction module, the output features of the feature extraction module are input into the average pooling module, the output features of the average pooling module are input into the ECA module, and the output features of the ECA module are input into the output module and processed by the output module to output the product indicator prediction result.
[0036] The first extraction module and the second extraction module have the same structure;
[0037] The first extraction module includes convolution module I, convolution module II, convolution module III, global average pooling module I, global average pooling module II, ReLU activation function module I, ReLU activation function module II, and SiLU activation function module;
[0038] The input of convolution module I serves as the input of the first extraction module. The output of convolution module I is connected to the input of ReLU activation function module I. The output of ReLU activation function module I is connected to the input of global average pooling module I. The output of global average pooling module I is connected to the input of convolution module II. The output of convolution module II is connected to the input of ReLU activation function module II. The output of ReLU activation function module II is connected to the input of convolution module III. The output of convolution module III is connected to the input of SiLU activation function module. The output of SiLU activation function module serves as the output of the first extraction module.
[0039] Among them, the convolution scales of convolution module I, convolution module II and convolution module III are 3×3, 5×5 and 7×7 respectively;
[0040] The feature extraction module includes convolution module IV, convolution module V, convolution module VI, batch normalization module, and SiLU activation function module I;
[0041] The input of convolution module IV serves as the input of the first extraction module. The output of convolution module IV is connected to the input of convolution module V. The output of convolution module V is connected to the input of the batch normalization module. The output of the batch normalization module is connected to the input of SiLU activation function module I. The output of SiLU activation function module I is connected to the input of convolution module VI. The output of convolution module VI serves as the output of the first extraction module.
[0042] Among them, the convolution scales of convolution module I, convolution module II and convolution module III are 1×1, 3×3 and 1×1 respectively.
[0043] Furthermore, the output module includes a fully connected layer and a softmax activation function module;
[0044] The input of the fully connected layer is connected to the input of the ECA module, and the output of the fully connected layer is connected to the input of the softmax activation function module. The softmax activation function module outputs the product indicator prediction results.
[0045] The beneficial effects of this invention are as follows: By establishing and training a corresponding product indicator prediction model, the invention can obtain the corresponding machine tool operating parameters in real time to predict product indicators. The prediction results can then guide the adjustment of machine tool process parameters. In this process, the machine tool's operating status can be tracked in real time, thereby ensuring the accuracy of process parameter adjustments, avoiding the influence of subjective factors in traditional technologies, and achieving high efficiency. This effectively ensures the quality of the final cut product surface and reduces the additional costs caused by process parameter adjustments. Attached Figure Description
[0046] The present invention will be further described below with reference to the accompanying drawings and embodiments:
[0047] Figure 1 This is a schematic diagram of the process of the present invention.
[0048] Figure 2 This is a schematic diagram of the product indicator prediction module of the present invention.
[0049] Figure 3 This is a schematic diagram of the feature extraction unit structure of the present invention.
[0050] Figure 4 This is a schematic diagram of the structure of the first extraction module of the present invention.
[0051] Figure 5 This is a schematic diagram of the feature extraction module structure of the present invention.
[0052] Figure 6 This is a comparison chart of the predicted results and actual values of the present invention. Detailed Implementation
[0053] The invention will be further described in detail below:
[0054] This invention provides a method and system for predicting and analyzing the performance indicators of products machined by ultra-precision cutting machine tools, comprising the following steps:
[0055] S1. Obtain the operating parameters of the ultra-precision cutting machine tool, the operating parameters including dynamic parameters and static parameters;
[0056] S2. Preprocess the operating parameters of the ultra-precision cutting machine tool;
[0057] S3. Construct a product indicator prediction model and input the preprocessed running parameters into the product indicator prediction model for training;
[0058] S4. Obtain the real-time operating parameters of the ultra-precision cutting machine tool, preprocess the real-time operating parameters, and input them into the trained product index prediction model to obtain the product index prediction result. Through this invention, by establishing and training a corresponding product index prediction model, the corresponding machine tool operating parameters are obtained in real time to predict product indexes. The prediction results then guide the adjustment of machine tool process parameters. This process allows for real-time tracking of the machine tool's operating status, ensuring the accuracy of process parameter adjustments, avoiding the influence of subjective factors in traditional technologies, and achieving high efficiency. It effectively ensures the quality of the final cut product surface and reduces the additional costs associated with process parameter adjustments.
[0059] The product indicators include the surface roughness, dimensional accuracy, and tool wear of the processed products. These indicators can all be achieved through the product indicator prediction model of this invention. When used to guide the adjustment of process parameters, any one indicator or a combination of multiple indicators can be used. Adjusting process parameters based on indicators is an existing technology and will not be elaborated here.
[0060] In this embodiment, the product indicator prediction model includes a feature extraction unit, a Transformer network, a feature extraction module, an average pooling module, an ECA module, and an output module.
[0061] The feature extraction unit receives preprocessed operating parameters as input, and its output features are fed into a Transformer network for processing. The output features of the Transformer network are then fed into a feature extraction module, which in turn feeds into an average pooling module. The average pooling module's output features are fed into an ECA module, and the ECA module's output features are fed into an output module, which processes the data and outputs the product metric prediction result. By employing a Transformer network, the global receptive field can be obtained, and adaptive feature association can be performed. Dynamic weighted aggregation of global contextual information ensures prediction accuracy. The Transformer network is an existing network and will not be described in detail here. The entire prediction model uses existing loss functions, such as mean loss and cross-loss, for training, which can be selected according to actual needs. The ECA module, short for Efficient ChannelAttention, is a channel attention mechanism module that can locally interact across channels to strengthen key features and suppress redundant information.
[0062] In this embodiment, the feature extraction unit includes a first extraction module, a first max pooling module, a second extraction module, and a second max pooling module;
[0063] The input terminal of the first extraction module serves as the input terminal of the feature extraction unit. The output features of the first extraction module are input to the first max pooling module. The output features of the first max pooling module are input to the second extraction module. The output features of the second extraction module are input to the second max pooling module. The output terminal of the second max pooling module serves as the output terminal of the feature extraction unit.
[0064] Specifically: the first extraction module and the second extraction module have the same structure;
[0065] The first extraction module includes convolution module I, convolution module II, convolution module III, global average pooling module I, global average pooling module II, ReLU activation function module I, ReLU activation function module II, and SiLU activation function module;
[0066] The input of convolution module I serves as the input of the first extraction module. The output of convolution module I is connected to the input of ReLU activation function module I. The output of ReLU activation function module I is connected to the input of global average pooling module I. The output of global average pooling module I is connected to the input of convolution module II. The output of convolution module II is connected to the input of ReLU activation function module II. The output of ReLU activation function module II is connected to the input of global average pooling module II. The output of global average pooling module II is connected to the input of convolution module III. The output of convolution module III is connected to the input of SiLU activation function module. The output of SiLU activation function module serves as the output of the first extraction module.
[0067] Specifically, the convolutional scales of convolutional modules I, II, and III are 3×3, 5×5, and 7×7, respectively. This structure effectively ensures a sufficient receptive field, thereby capturing spatial features at different scales and guaranteeing the accuracy of the prediction results.
[0068] In this embodiment, the feature extraction module includes convolution module IV, convolution module V, convolution module VI, batch normalization module, and SiLU activation function module I;
[0069] The input of convolution module IV serves as the input of the first extraction module. The output of convolution module IV is connected to the input of convolution module V. The output of convolution module V is connected to the input of the batch normalization module. The output of the batch normalization module is connected to the input of SiLU activation function module I. The output of SiLU activation function module I is connected to the input of convolution module VI. The output of convolution module VI serves as the output of the first extraction module.
[0070] Among them, the convolution scales of convolution module IV, convolution module V and convolution module VI are 1×1, 3×3 and 1×1 respectively; through the above structure, the feature expression ability can be effectively maintained, cross-channel information fusion can be achieved, and the computational load can be reduced.
[0071] The output module includes a fully connected layer and a softmax activation function module;
[0072] The input of the fully connected layer is connected to the input of the ECA module, and the output of the fully connected layer is connected to the input of the softmax activation function module. The softmax activation function module outputs the product indicator prediction results.
[0073] In this embodiment, the static parameters include machine tool guideway positioning accuracy, repeatability positioning accuracy, and spindle rotation accuracy; these static parameters are generally determined and tested before production.
[0074] The dynamic parameters include cutting tool temperature, machine tool component vibration data, acoustic emission signals during the cutting process, and cutting force. Combining these dynamic and static parameters ensures the accuracy of product performance predictions.
[0075] In this embodiment, the preprocessing of the operating parameters includes:
[0076] Denoising is used to remove abnormal data from operating parameters; existing filtering methods are used for denoising.
[0077] The dynamic parameters in the denoised operating parameters are transformed using Morlet wavelets. This step, by selecting the scale parameter, decomposes the one-dimensional time series signal into wavelet coefficients of multiple different frequency scales. Each scale corresponds to a specific frequency range, effectively capturing transient features, periodic changes, and local anomalies in the signal, providing accurate data support for the subsequent learning and training of the prediction model and the final prediction results.
[0078] Accordingly, the present invention also provides a product index prediction and analysis system for products processed by ultra-precision cutting machine tools, including a sensor module and a processing module;
[0079] The sensor module is used to acquire the operating parameters of the ultra-precision cutting machine tool and input them to the processing module; the sensor module includes a temperature sensor, a cutting force sensor, a vibration sensor, and an acoustic emission sensor.
[0080] The temperature sensor used is an existing non-contact temperature sensor.
[0081] The vibration sensor is an accelerometer such as the Kistler 8763B050AT, used to detect vibration signals at two positions: the lathe spindle and the tool post.
[0082] The cutting force sensor is a Kistler 9119AA1 sensor, which is set at the clamping position of the tool holder;
[0083] The acoustic emission sensor is used to acquire a certain acoustic emission signal generated by the friction and impact force between the tool and the workpiece when they come into contact. The Vallen Systeme-VS900-RIC acoustic emission sensor is used.
[0084] The processing module is used to receive the operating parameters output by the sensor module and output the product indicator prediction results.
[0085] The processing module includes a preprocessing module and a product indicator prediction unit;
[0086] The preprocessing module is used to denoise the operating parameters and then perform Morlet wavelet transform processing (of course, the wavelet transform here is performed on the dynamic parameters). The output data of the preprocessing module is input into the product indicator prediction unit.
[0087] The product indicator prediction unit includes a feature extraction unit, a Transformer network, a feature extraction module, an average pooling module, an ECA module, and an output module.
[0088] The feature extraction unit inputs preprocessed operating parameters, the output features of the feature extraction unit are input into the Transformer network for processing, the output features of the Transformer network are input into the feature extraction module, the output features of the feature extraction module are input into the average pooling module, the output features of the average pooling module are input into the ECA module, and the output features of the ECA module are input into the output module and processed by the output module to output the product indicator prediction result.
[0089] The first extraction module and the second extraction module have the same structure;
[0090] The first extraction module includes convolution module I, convolution module II, convolution module III, global average pooling module I, global average pooling module II, ReLU activation function module I, ReLU activation function module II, and SiLU activation function module;
[0091] The input of convolution module I serves as the input of the first extraction module. The output of convolution module I is connected to the input of ReLU activation function module I. The output of ReLU activation function module I is connected to the input of global average pooling module I. The output of global average pooling module I is connected to the input of convolution module II. The output of convolution module II is connected to the input of ReLU activation function module II. The output of ReLU activation function module II is connected to the input of convolution module III. The output of convolution module III is connected to the input of SiLU activation function module. The output of SiLU activation function module serves as the output of the first extraction module.
[0092] Among them, the convolution scales of convolution module I, convolution module II and convolution module III are 3×3, 5×5 and 7×7 respectively;
[0093] The feature extraction module includes convolution module IV, convolution module V, convolution module VI, batch normalization module, and SiLU activation function module I;
[0094] The input of convolution module IV serves as the input of the first extraction module. The output of convolution module IV is connected to the input of convolution module V. The output of convolution module V is connected to the input of the batch normalization module. The output of the batch normalization module is connected to the input of SiLU activation function module I. The output of SiLU activation function module I is connected to the input of convolution module VI. The output of convolution module VI serves as the output of the first extraction module.
[0095] Among them, the convolution scales of convolution module I, convolution module II and convolution module III are 1×1, 3×3 and 1×1 respectively.
[0096] Wherein: the output module includes a fully connected layer and a softmax activation function module;
[0097] The input of the fully connected layer is connected to the input of the ECA module, and the output of the fully connected layer is connected to the input of the softmax activation function module. The softmax activation function module outputs the product indicator prediction results.
[0098] like Figure 6 As shown in the figure, the blue lines represent the predicted values, the red dots represent the actual values, and the gray area represents the prediction error range. It can be seen from the figure that the predicted values and the actual values are highly consistent at most points in time. Even in areas with large numerical fluctuations, the model can still accurately track the actual trend.
[0099] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.
Claims
1. A method for predicting and analyzing the performance indicators of products machined by ultra-precision cutting machine tools, characterized in that: Includes the following steps: S1. Obtain the operating parameters of the ultra-precision cutting machine tool, the operating parameters including dynamic parameters and static parameters; S2. Preprocess the operating parameters of the ultra-precision cutting machine tool; S3. Construct a product indicator prediction model and input the preprocessed running parameters into the product indicator prediction model for training; S4. Obtain the real-time operating parameters of the ultra-precision cutting machine tool, and input the preprocessed real-time operating parameters into the trained product index prediction model to obtain the product index prediction results.
2. The method for predicting and analyzing the indicators of products processed by ultra-precision cutting machine tools according to claim 1, characterized in that: The product indicator prediction model includes a feature extraction unit, a Transformer network, a feature extraction module, an average pooling module, an ECA module, and an output module. The feature extraction unit inputs the preprocessed operating parameters, and the output features of the feature extraction unit are input into the Transformer network for processing. The output features of the Transformer network are input into the feature extraction module, and the output features of the feature extraction module are input into the average pooling module. The output features of the average pooling module are input into the ECA module, and the output features of the ECA module are input into the output module and processed by the output module to output the product indicator prediction result.
3. The method for predicting and analyzing the indicators of products processed by ultra-precision cutting machine tools according to claim 2, characterized in that: The feature extraction unit includes a first extraction module, a first max pooling module, a second extraction module, and a second max pooling module; The input terminal of the first extraction module serves as the input terminal of the feature extraction unit. The output features of the first extraction module are input to the first max pooling module. The output features of the first max pooling module are input to the second extraction module. The output features of the second extraction module are input to the second max pooling module. The output terminal of the second max pooling module serves as the output terminal of the feature extraction unit.
4. The method for predicting and analyzing the indicators of products processed by ultra-precision cutting machine tools according to claim 3, characterized in that: The first extraction module and the second extraction module have the same structure; The first extraction module includes convolution module I, convolution module II, convolution module III, global average pooling module I, global average pooling module II, ReLU activation function module I, ReLU activation function module II, and SiLU activation function module; The input of convolution module I serves as the input of the first extraction module. The output of convolution module I is connected to the input of ReLU activation function module I. The output of ReLU activation function module I is connected to the input of global average pooling module I. The output of global average pooling module I is connected to the input of convolution module II. The output of convolution module II is connected to the input of ReLU activation function module II. The output of ReLU activation function module II is connected to the input of convolution module III. The output of convolution module III is connected to the input of SiLU activation function module. The output of SiLU activation function module serves as the output of the first extraction module. Among them, the convolution scales of convolution module I, convolution module II, and convolution module III are 3×3, 5×5, and 7×7, respectively.
5. The method for predicting and analyzing the indicators of products processed by ultra-precision cutting machine tools according to claim 2, characterized in that: The feature extraction module includes convolution module IV, convolution module V, convolution module VI, batch normalization module, and SiLU activation function module I; The input of convolution module IV serves as the input of the first extraction module. The output of convolution module IV is connected to the input of convolution module V. The output of convolution module V is connected to the input of the batch normalization module. The output of the batch normalization module is connected to the input of SiLU activation function module I. The output of SiLU activation function module I is connected to the input of convolution module VI. The output of convolution module VI serves as the output of the first extraction module. Among them, the convolution scales of convolution module I, convolution module II and convolution module III are 1×1, 3×3 and 1×1 respectively.
6. The method for predicting and analyzing the indicators of products processed by ultra-precision cutting machine tools according to claim 2, characterized in that: The output module includes a fully connected layer and a softmax activation function module; The input of the fully connected layer is connected to the input of the ECA module, and the output of the fully connected layer is connected to the input of the softmax activation function module. The softmax activation function module outputs the product indicator prediction results.
7. The method for predicting and analyzing the indicators of products processed by ultra-precision cutting machine tools according to claim 1, characterized in that: The static parameters include machine tool guideway positioning accuracy, repeatability, and spindle rotation accuracy. The dynamic parameters include cutting tool temperature, machine tool component vibration data, acoustic emission signals during the cutting process, and cutting force.
8. The method for predicting and analyzing the indicators of products processed by ultra-precision cutting machine tools according to claim 1, characterized in that: Preprocessing of operating parameters includes: Denoising is used to remove abnormal data from operating parameters; The dynamic parameters in the denoised operating parameters are transformed using Morlet wavelets.
9. A predictive analysis system for product indicators processed by ultra-precision cutting machine tools, characterized in that: Includes a sensor module and a processing module; The sensor module is used to acquire the operating parameters of the ultra-precision cutting machine tool and input them to the processing module; the sensor module includes a temperature sensor, a cutting force sensor, a vibration sensor, and an acoustic emission sensor. The processing module is used to receive the operating parameters output by the sensor module and output the product indicator prediction results. The processing module includes a preprocessing module and a product indicator prediction unit; The preprocessing module is used to denoise the operating parameters and then perform Morlet wavelet transform processing. The output data of the preprocessing module is input to the product indicator prediction unit. The product indicator prediction unit includes a feature extraction unit, a Transformer network, a feature extraction module, an average pooling module, an ECA module, and an output module. The feature extraction unit inputs preprocessed operating parameters, the output features of the feature extraction unit are input into the Transformer network for processing, the output features of the Transformer network are input into the feature extraction module, the output features of the feature extraction module are input into the average pooling module, the output features of the average pooling module are input into the ECA module, and the output features of the ECA module are input into the output module and processed by the output module to output the product indicator prediction result. The first extraction module and the second extraction module have the same structure; The first extraction module includes convolution module I, convolution module II, convolution module III, global average pooling module I, global average pooling module II, ReLU activation function module I, ReLU activation function module II, and SiLU activation function module; The input of convolution module I serves as the input of the first extraction module. The output of convolution module I is connected to the input of ReLU activation function module I. The output of ReLU activation function module I is connected to the input of global average pooling module I. The output of global average pooling module I is connected to the input of convolution module II. The output of convolution module II is connected to the input of ReLU activation function module II. The output of ReLU activation function module II is connected to the input of convolution module III. The output of convolution module III is connected to the input of SiLU activation function module. The output of SiLU activation function module serves as the output of the first extraction module. Among them, the convolution scales of convolution module I, convolution module II and convolution module III are 3×3, 5×5 and 7×7 respectively; The feature extraction module includes convolution module IV, convolution module V, convolution module VI, batch normalization module, and SiLU activation function module I; The input of convolution module IV serves as the input of the first extraction module. The output of convolution module IV is connected to the input of convolution module V. The output of convolution module V is connected to the input of the batch normalization module. The output of the batch normalization module is connected to the input of SiLU activation function module I. The output of SiLU activation function module I is connected to the input of convolution module VI. The output of convolution module VI serves as the output of the first extraction module. Among them, the convolution scales of convolution module I, convolution module II and convolution module III are 1×1, 3×3 and 1×1 respectively.
10. The predictive analysis system for product indicators of ultra-precision cutting machine tools according to claim 9, characterized in that: The output module includes a fully connected layer and a softmax activation function module; The input of the fully connected layer is connected to the input of the ECA module, and the output of the fully connected layer is connected to the input of the softmax activation function module. The softmax activation function module outputs the product indicator prediction results.