A method, system, device, medium and product for monitoring surface roughness of a part in a numerical control machining process

Abstract

The application discloses a kind of monitoring methods, systems, equipment, medium and product of part surface roughness in numerical control machining process, it is related to numerical control machining field, the method includes: real-time acquisition three-dimensional vibration signal and sound signal of part in numerical control machining process;Characteristic extraction is carried out to three-dimensional vibration signal and sound signal, obtain multiple characteristic values;Based on the multiple characteristic values, using recursive feature elimination method and cross-validation method, obtain optimal feature subset;Based on optimal feature subset, using long short-term memory network and Kolmogorov-Arnold network, obtain part surface roughness, the application can monitor the surface roughness of part in real time, and by the correlation of optimal feature subset and part surface roughness, improve the monitoring precision of part surface roughness.

Description

Technical Field

[0001] This application relates to the field of CNC machining, and in particular to a method, system, equipment, medium, and product for monitoring the surface roughness of parts during CNC machining. Background Technology

[0002] Due to the demands for high efficiency and high quality, achieving high speed, high precision, and high material removal efficiency through machining has become an important goal in production. Furthermore, given the importance of tribological properties, fatigue strength, and assembly accuracy of mechanical products, machined parts with high surface finishes are also highly sought after. Dimensional accuracy and surface quality are affected by various factors, such as cutting parameters, tool geometry and vibration, machining dynamics, and the machining environment. Machining vibrations caused by improper machining conditions and parameters can negatively impact surface quality.

[0003] Surface roughness is a crucial parameter in surface quality assessment systems. It can be influenced by various factors, some of which are controllable, such as basic cutting parameter settings, while others are uncontrollable, such as cutting vibration and tool wear. However, due to the complex machining environment and the coupled effects of multiple factors such as cutting fluid, tool vibration, and chips during machining, direct online in-situ monitoring of workpiece surface roughness is impossible. Surface roughness measurement can only be performed after machining is completed, requiring the machine to be stopped and the part removed, significantly reducing machining efficiency.

[0004] Furthermore, the time-varying dynamic characteristics of CNC machining of precision parts determine the complexity of state recognition. When machining parameters, tools, materials, or processes change, the surface roughness monitoring model needs to be retrained, which is obviously detrimental to improving production efficiency. Therefore, the difficulty in obtaining real-time state data for surface roughness monitoring of CNC-machined precision parts, and the complex adjustment process for different machining conditions, have become key technical bottlenecks restricting the development of quality monitoring technology for CNC machining of precision parts. Summary of the Invention

[0005] The purpose of this application is to provide a method, system, equipment, medium, and product for monitoring the surface roughness of parts during CNC machining, which can monitor the surface roughness of parts in real time and improve the monitoring accuracy of surface roughness.

[0006] To achieve the above objectives, this application provides the following solution:

[0007] Firstly, this application provides a method for monitoring the surface roughness of parts during CNC machining, including:

[0008] Real-time acquisition of three-dimensional vibration and sound signals of parts during CNC machining;

[0009] Feature extraction is performed on the triaxial vibration signal and the sound signal to obtain multiple feature values; the multiple feature values ​​include multiple feature values ​​of the triaxial vibration signal and multiple feature values ​​of the sound signal;

[0010] Based on the aforementioned multiple feature values, the optimal feature subset is obtained by employing recursive feature elimination and cross-validation methods.

[0011] Based on the optimal feature subset, the surface roughness of the part is obtained by using a long short-term memory network and a Kolmogorov-Arnold network.

[0012] Secondly, this application provides a system for monitoring the surface roughness of parts during CNC machining, comprising:

[0013] The signal acquisition module is used to collect the three-dimensional vibration and sound signals of the parts during CNC machining in real time.

[0014] The feature value acquisition module, connected to the signal acquisition module, is used to extract features from the triaxial vibration signal and the sound signal to obtain multiple feature values; the multiple feature values ​​include multiple feature values ​​of the triaxial vibration signal and multiple feature values ​​of the sound signal;

[0015] The optimal feature subset acquisition module is connected to the feature value acquisition module and is used to obtain the optimal feature subset based on the multiple feature values ​​by using recursive feature elimination and cross-validation methods.

[0016] The surface roughness value acquisition module is connected to the optimal feature subset acquisition module. It is used to obtain the surface roughness of the part based on the optimal feature subset, using a long short-term memory network and a Kolmogorov-Arnold network.

[0017] Thirdly, this application provides a computer device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the above-described method for monitoring the surface roughness of parts during CNC machining.

[0018] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-described method for monitoring the surface roughness of parts during CNC machining.

[0019] Fifthly, this application provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method for monitoring the surface roughness of parts during CNC machining.

[0020] According to the specific embodiments provided in this application, this application has the following technical effects:

[0021] This application provides a method, system, equipment, medium, and product for monitoring the surface roughness of parts during CNC machining. By real-time acquisition of triaxial vibration and sound signals of parts during CNC machining, the surface roughness of parts can be monitored in real time. Further feature extraction is performed on the triaxial vibration and sound signals. A recursive feature elimination method is used to eliminate several low-weight features, and a cross-validation method is used to avoid feature redundancy or filter out important features to obtain an optimal feature subset. Based on the optimal feature subset, the correlation between surface roughness and features is determined through a long short-term memory network and a Kolmogorov-Arnold network, thereby improving the monitoring accuracy of part surface roughness. Attached Figure Description

[0022] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 This is an application environment diagram of a method for monitoring the surface roughness of parts during CNC machining, according to an embodiment of this application.

[0024] Figure 2 A flowchart illustrating a method for monitoring the surface roughness of a part during CNC machining, provided as an embodiment of this application;

[0025] Figure 3 A schematic diagram illustrating the principle of the recursive feature elimination method and cross-validation method provided in another embodiment of this application;

[0026] Figure 4 A schematic diagram of a long short-term memory network provided for another embodiment of this application;

[0027] Figure 5 A schematic diagram of a Long Short-Term Memory network and a Kolmogorov-Arnold network is provided for another embodiment of this application;

[0028] Figure 6 A schematic diagram showing the results of comparing predicted values ​​with actual values ​​according to another embodiment of this application;

[0029] Figure 7 A schematic diagram of the functional modules of a system for monitoring the surface roughness of parts during CNC machining, provided for another embodiment of this application;

[0030] Figure 8 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0031] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0032] To make the objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0033] The method for monitoring the surface roughness of parts during CNC machining provided in this application embodiment can be applied to, for example... Figure 1 In the application environment shown, terminal 102 communicates with server 104 via a network. A data storage system can store the data that server 104 needs to process. The data storage system can be set up independently, integrated into server 104, or placed in the cloud or on another server. Terminal 102 can send the triaxial vibration signal and sound signal to be processed to server 104. After receiving the triaxial vibration signal and sound signal, server 104 performs feature extraction on the triaxial vibration signal and sound signal to obtain multiple feature values. Based on the multiple feature values, a recursive feature elimination method and cross-validation method are used to obtain an optimal feature subset. Based on the optimal feature subset, a long short-term memory network and a Kolmogorov-Arnold network are used to obtain the surface roughness of the part. Server 104 can feed back the obtained surface roughness of the part to terminal 102. Furthermore, in some embodiments, the method for monitoring the surface roughness of the part during CNC machining can also be implemented independently by server 104 or terminal 102.

[0034] The terminal 102 can be, but is not limited to, various desktop computers, laptops, smartphones, tablets, IoT devices, and portable wearable devices. IoT devices can include smart speakers, smart TVs, smart air conditioners, and smart in-vehicle devices. Portable wearable devices can include smartwatches, smart bracelets, and head-mounted devices. The server 104 can be implemented using a standalone server or a server cluster composed of multiple servers, or it can be a cloud server.

[0035] In one exemplary embodiment, such as Figure 2 As shown, a method for monitoring the surface roughness of parts during CNC machining is provided. This method is executed by a computer device, specifically a terminal or server, or both. In this embodiment, the method is applied to... Figure 1Taking the server 104 in [the example] as an illustration, it includes the following steps 201 to step 204. Among them:

[0036] Step 201, collect the three-way vibration signal and sound signal of the part in real time during the numerical control machining process.

[0037] Specifically, the three-way vibration signal and sound signal of the part during the numerical control machining process can be collected by a three-way vibration sensor and a sound sensor.

[0038] Step 202, extract features from the three-way vibration signal and the sound signal to obtain multiple feature values; the multiple feature values include multiple feature values of the three-way vibration signal and multiple feature values of the sound signal.

[0039] Step 203, based on the multiple feature values, use the recursive feature elimination method and the cross-validation method to obtain the optimal feature subset.

[0040] Step 204, based on the optimal feature subset, use the long short-term memory network and the Kolmogorov-Arnold network to obtain the surface roughness of the part.

[0041] In another exemplary embodiment of the present application, step 203 specifically includes: for the Nth iteration, based on the (N - 1)th feature subset, use the recursive feature elimination method to obtain the weight of each feature value in the (N - 1)th feature subset; and use the cross-validation method to obtain the average score of the (N - 1)th feature subset; 0 < N; when N = 1, the 0th feature subset includes all feature values.

[0042] Based on the weight of each feature value in the (N - 1)th feature subset, sort the multiple feature values in the (N - 1)th feature subset, select a specified number of feature values with the top rankings as the Nth feature subset; and use the cross-validation method to obtain the average score of the Nth feature subset; determine whether the average score of the Nth feature subset is less than the average score of the (N - 1)th feature subset; if so, stop the iteration and take the (N - 1)th feature subset as the optimal feature subset; otherwise, perform the (N + 1)th iteration.

[0043] In another exemplary embodiment of the present application, as Figure 3As shown, the zeroth feature subset includes features 1, 2, 3, 4, and 5; cross-validation yields 80% of the zeroth feature subset. The recursive feature elimination method trains the model using all features in the zeroth feature subset to obtain the weight of each feature value. This model is a supervised learning model, which in this application can be a support vector machine. Features 1, 3, 4, and 5 with high weights are selected as the first feature subset; cross-validation yields 90% of the first feature subset. This process is repeated iteratively, selecting features 1, 3, and 5 with high weights as the second feature subset; cross-validation yields 85% of the second feature subset. Based on the average score of each feature subset, the first feature subset is selected as the optimal feature subset.

[0044] In another exemplary embodiment of this application, the plurality of feature values ​​include a plurality of time-domain feature values ​​and a plurality of frequency-domain feature values; the plurality of time-domain feature values ​​include a maximum value, a minimum value, an average value, a peak-to-peak value, a kurtosis factor, a skewness factor, and a standard deviation; the plurality of frequency-domain feature values ​​include a power spectrum average value, a power spectrum amplitude, a power spectrum power value, and a power spectrum variance value.

[0045] Specifically, the formula for calculating the average value M is: ,in, Let be the true value of the i-th sample, and n be the total number of samples; the average value represents the central tendency of the signal. The formula for calculating the peak-to-peak value PP is: Peak-to-peak value represents the overall range of signal variation. Standard deviation. The calculation formula is: ; For the signal's first The sample value is M; M is the signal mean; the standard deviation represents the degree of signal fluctuation. The formula for calculating the kurtosis factor Kur is: Kurtosis represents the peak value of the probability density distribution curve of a signal at its mean. Skewness factor Ske is calculated using the following formula: The skewness factor represents the degree of asymmetry between the probability density distribution curve of a signal and its mean value.

[0046] In another exemplary embodiment of this application, feature extraction is performed on the triaxial vibration signal and the sound signal to obtain multiple feature values, and then the multiple feature values ​​are normalized to obtain multiple processed feature values.

[0047] In another exemplary embodiment of this application, step 204 specifically includes:

[0048] Multiple feature values ​​from the optimal feature subset are used as a feature value sequence and input into a long short-term memory network to obtain the hidden state of the feature value sequence.

[0049] The hidden state of the feature value sequence is input into the Kolmogorov-Arnold network to obtain the surface roughness of the part.

[0050] In another exemplary embodiment of this application, the method for monitoring the surface roughness of a part during CNC machining further includes:

[0051] The Long Short-Term Memory Network and the Kolmogorov-Arnold Network were trained.

[0052] When training the Long Short-Term Memory (LSTM) network and the Kolmogorov-Arnold network, a subset of historical optimal features is used as input to both networks, and historical part surface roughness is used as a label to output the predicted part surface roughness. The historical part surface roughness is obtained through offline measurement.

[0053] The mean absolute error, root mean square error, and mean absolute percentage error between historical and predicted part surface roughness are used as the optimization objectives for the Long Short-Term Memory (LSTM) network and the Kolmogorov-Arnold network. The specific formulas are as follows:

[0054] .

[0055] .

[0056] .

[0057] in, The root mean square error, The mean absolute error value. The mean absolute percentage error is given by n, where n is the total number of samples. Let i be the true value of the i-th sample. Let be the predicted value for the i-th sample. During training, 80% of the data is used to train the neural network, and 20% of the data is used for the post-training network performance verification step.

[0058] Among them, Long Short-Term Memory Network (LSTM) is a type of recurrent neural network that introduces a gating mechanism to more effectively capture and process long-term dependencies in sequential data, such as... Figure 4As shown. LSTM can solve many time series tasks that feedforward networks cannot, and it exhibits excellent performance in prediction. These gates determine the degree of information flow through the sigmoid activation function (output range (0,1), representing probability values): The Forget Gate determines which information from the cell state at the previous time step should be forgotten, and its output value is between 0 and 1 (0 represents complete forgetting, 1 represents complete retention), calculated as follows: ;in, It is a sigmoid activation function. and These are the weights and biases of the forget gate, respectively. and This is a concatenation of the current input and the previous hidden state. This is the output of the forget gate.

[0059] The input gate consists of two parts: the decision gate and the candidate input gate. The decision gate determines which parts of the most recently input data should be added to the cell state at the current moment. The candidate input gates provide new candidate information that can be added to the cell state, and the input gates determine which new information is important enough to be written to the cell state.

[0060] ; ;in, The output of the input gate, and These are the weights and biases of the input gate, respectively. The output is the candidate cell state. and These are the weights and biases of the candidate cell state, respectively.

[0061] ;in, For the output of the output gate, and These represent the weights and biases of the output gate, respectively. This indicates how the output gate selects which cell state data should be sent as hidden states to subsequent layers or used for final prediction at any given time.

[0062] Based on the above gating mechanism, the LSTM updates the cell state and calculates the hidden state at the current time step, thus updating the cell state. ;in, This represents the current cell state. This is the cell state from the previous time step. Hidden state. ;in, This represents the hidden state at the current moment. The results of hidden state calculation and cell state update generated by LSTM based on the aforementioned gate control mechanism are shown. The cell state can be compressed to the range (-1, 1) using the tanh function for subsequent calculations. The symbols... This indicates the multiplication of the value at the corresponding index.

[0063] Kolmogorov-Arnold Networks (KANs) are a novel neural network architecture that represents functions through learnable nonlinear transformations and summation operations at nodes, achieving higher accuracy in mathematical and physical problems with fewer parameters. In KANs, each connection is no longer a simple weight but a learnable one-dimensional function. These functions are typically parameterized using splines or other smoothing functions, allowing the network to learn the nonlinear transformation best suited to the data in each dimension. However, KAN networks are slow because different activation functions cannot be computed in batches. This paper constructs a regression combinatorial model by combining LSTM. The LSTM layer extracts the temporal dependencies in the input sequence, and the hidden state at the last time step is used as the input to the KAN layer for further processing to generate the final output. The LSTM-KAN network architecture is as follows: Figure 5 As shown.

[0064] This application enables online, in-situ monitoring of part surface roughness during CNC machining. When operating parameters, tools, or materials change, only model retraining is required, eliminating the need for repeated data transmission, model debugging, training, and observation. This improves efficiency, expands the scope of application, and enables roughness monitoring for flexible manufacturing. Addressing the issue of variable feature evolution patterns due to the time-varying dynamic characteristics of the cutting system during CNC machining of precision optical parts, this application effectively filters signal feature values ​​by considering the correlation with prediction results during feature extraction, thereby improving computational efficiency and accuracy.

[0065] In another exemplary embodiment of this application, taking the CNC machining of titanium alloy parts as an example, a series of milling tests were conducted on a Mikron HEM 1000 vertical milling machining center. Carbide end mills were selected, and the material being machined was TC4 alloy. The milling experiments were conducted under different combinations of cutting parameters and tool wear levels. To facilitate machining and considering that tool wear is not sensitive to the cutting width, the cutting width was set to 8 mm, and the single pass depth was 100 mm. To ensure the diversity and comprehensiveness of cutting parameter combinations, the Taguchi experimental design was selected. Other cutting parameters were determined based on the suggestions of the production department and preliminary experiments, as shown in Table 1.

[0066] Table 1. Range of operating parameters during processing.

[0067]

[0068] Time-domain and frequency-domain feature extraction was performed on the collected multi-source signals. Feature value selection was conducted using a recursive feature elimination (RFE) method, with multiple rounds of training. After each round, several low-weight features were eliminated, and the next round of training was performed based on the new feature set. To avoid feature redundancy or filtering out important features, after obtaining feature rankings, model training and cross-validation (CV) were performed on feature subsets with different numbers of features. Finally, the feature subset with the highest average score was selected. Model training was then performed based on the selected feature subset and surface roughness data. After model training, the operating parameters were changed, and CNC machining experiments were conducted again. Multi-source signals were collected for online in-situ monitoring, and the detected values ​​were compared with the true values. Figure 6 As shown in the figure, the detection results are MAE: 0.03662, RMSE: 0.04746, and MAPE: 5.8816%. It can be seen that the detection accuracy is high and meets the requirements of online original text monitoring.

[0069] Based on the same inventive concept, embodiments of this application also provide a system for monitoring the surface roughness of parts during CNC machining. For example... Figure 7 As shown, the surface roughness monitoring system for parts during CNC machining includes:

[0070] The signal acquisition module 301 is used to collect the current three-dimensional vibration signal and the current sound signal of the part during CNC machining.

[0071] The feature value acquisition module 302 is connected to the signal acquisition module 301 and is used to extract features from the triaxial vibration signal and the sound signal to obtain multiple feature values; the multiple feature values ​​include multiple feature values ​​of the triaxial vibration signal and multiple feature values ​​of the sound signal.

[0072] The optimal feature subset acquisition module 303 is connected to the feature value acquisition module and is used to obtain the optimal feature subset based on the multiple feature values ​​by using recursive feature elimination and cross-validation methods.

[0073] The surface roughness value acquisition module 304 is connected to the optimal feature subset acquisition module 303, and is used to obtain the surface roughness of the part based on the optimal feature subset, using a long short-term memory network and a Kolmogorov-Arnold network.

[0074] Specifically, this application employs vibration and sound sensors to collect vibration and sound signals during the milling process. The vibration sensor is magnetically attached to the spindle, perpendicular to the feed direction, and is treated with anti-static measures to prevent signal interference. The sound sensor is placed on one side of the spindle and is waterproofed to prevent cutting fluid infiltration. Both sensors transmit the sampled signals to a PC via a data acquisition card and collect data for further processing. The sampling frequency is 10kHz, and signal acquisition is triggered by a command to ensure that all acquired signals are valid during the machining process.

[0075] In one exemplary embodiment, a computer device is provided, which may be a server or a terminal, and its internal structure diagram may be as follows. Figure 8 As shown, this computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The database stores video tag processing data. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communicating with external terminals via a network. When executed by the processor, the computer program implements a method for monitoring the surface roughness of parts during CNC machining.

[0076] Those skilled in the art will understand that Figure 8 The structures shown are merely block diagrams of some structures related to the present application and do not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than shown in the figures, or combine certain components, or have different component arrangements. In an exemplary embodiment, a computer device is provided, including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0077] In one exemplary embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, implements the above-described method for monitoring the surface roughness of parts during CNC machining.

[0078] In one exemplary embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the above-described method for monitoring the surface roughness of parts during CNC machining.

[0079] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0080] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0081] The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0082] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0083] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method of monitoring the surface roughness of a part during a numerically controlled machining process, characterized in that, The method for monitoring the surface roughness of a part in a numerical control machining process comprises: real-time acquisition of three-direction vibration signals and sound signals of the part in the numerical control machining process; feature extraction of the three-direction vibration signals and the sound signals to obtain a plurality of feature values; the plurality of feature values include a plurality of feature values of the three-direction vibration signals and a plurality of feature values of the sound signals; based on the plurality of feature values, recursive feature elimination and cross-validation are used to obtain an optimal feature subset, specifically including: for the Nth iteration, based on the (N-1)th feature subset, the recursive feature elimination is used to obtain the weight of each feature value in the (N-1)th feature subset; and the cross-validation is used to obtain the average score of the (N-1)th feature subset; 0 < N; when N = 1, the zeroth feature subset includes all feature values; based on the weight of each feature value in the (N-1)th feature subset, the plurality of feature values in the (N-1)th feature subset are sorted, and the top-ranked specified number of feature values are selected as the Nth feature subset; and the cross-validation is used to obtain the average score of the Nth feature subset; it is judged whether the average score of the Nth feature subset is less than the average score of the (N-1)th feature subset; if yes, the iteration is stopped, and the (N-1)th feature subset is taken as the optimal feature subset; otherwise, the (N+1)th iteration is performed; based on the optimal feature subset, a long short-term memory network and a Kolmogorov-Arnold network are used to obtain the surface roughness of the part, specifically including: the plurality of feature values in the optimal feature subset are taken as a feature value sequence, which is input into the long short-term memory network to obtain the hidden state of the feature value sequence; the hidden state of the feature value sequence is input into the Kolmogorov-Arnold network to obtain the surface roughness of the part.

2. The method of claim 1, wherein, The plurality of feature values include a plurality of time domain feature values and a plurality of frequency domain feature values; the plurality of time domain feature values include maximum value, minimum value, average value, peak-to-peak value, kurtosis factor, skewness factor and standard deviation; and the plurality of frequency domain feature values include power spectrum average value, power spectrum amplitude value, power spectrum power value and power spectrum variance value.

3. The method of claim 1, wherein, After the feature extraction of the three-direction vibration signals and the sound signals to obtain the plurality of feature values, the method for monitoring the surface roughness of a part in a numerical control machining process further comprises: normalizing the plurality of feature values to obtain a plurality of processed feature values.

4. The method of claim 1, wherein, The method for monitoring the surface roughness of a part in a numerical control machining process further comprises: training of the long short-term memory network and the Kolmogorov-Arnold network; during the training of the long short-term memory network and the Kolmogorov-Arnold network, a historical optimal feature subset is taken as the input of the long short-term memory network and the Kolmogorov-Arnold network, a historical surface roughness of the part is taken as a label, and a predicted surface roughness of the part is output; and the average absolute error value, the root mean square error and the average absolute percentage error between the historical surface roughness of the part and the predicted surface roughness of the part are taken as the optimization target of the long short-term memory network and the Kolmogorov-Arnold network.

5. A system for monitoring surface roughness of a part in a numerically controlled machining process, applying the method for monitoring surface roughness of a part in a numerically controlled machining process according to any one of claims 1 to 4, characterized in that, The monitoring system for the surface roughness of a part in a numerical control machining process comprises: The signal acquisition module is configured to collect three-dimensional vibration signals and sound signals of a part in a numerical control machining process in real time. The feature value acquisition module is connected with the signal acquisition module and configured to extract features from the three-dimensional vibration signals and the sound signals to obtain a plurality of feature values, wherein the plurality of feature values include a plurality of feature values of the three-dimensional vibration signals and a plurality of feature values of the sound signals. The optimal feature subset acquisition module is connected with the feature value acquisition module and configured to obtain an optimal feature subset based on the plurality of feature values by using a recursive feature elimination method and a cross-validation method, and specifically includes the following steps: For the Nth iteration, the recursive feature elimination method is used to obtain the weight of each feature value in the (N-1)th feature subset based on the (N-1)th feature subset, and the cross-validation method is used to obtain the average score of the (N-1)th feature subset; 0 < N; when N = 1, the zeroth feature subset includes all feature values. Based on the weight of each feature value in the (N-1)th feature subset, the plurality of feature values in the (N-1)th feature subset are sorted, and the top ranked specified number of feature values are selected as the Nth feature subset; the cross-validation method is used to obtain the average score of the Nth feature subset; it is judged whether the average score of the Nth feature subset is less than the average score of the (N-1)th feature subset; if yes, the iteration is stopped, and the (N-1)th feature subset is taken as the optimal feature subset; otherwise, the (N+1)th iteration is performed. The surface roughness value acquisition module is connected with the optimal feature subset acquisition module and configured to obtain the surface roughness of the part based on the optimal feature subset by using a long short-term memory network and a Kolmogorov-Arnold network, and specifically includes the following steps: The plurality of feature values in the optimal feature subset are taken as a feature value sequence, which is input into the long short-term memory network to obtain the hidden state of the feature value sequence. The hidden state of the feature value sequence is input into the Kolmogorov-Arnold network to obtain the surface roughness of the part.

6. A computer device comprising: The memory, the processor, and the computer program stored in the memory and executable on the processor, characterized in that the processor executes the computer program to implement the monitoring method for the surface roughness of the part in the numerical control machining process according to any one of claims 1-4.

7. A computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to implement the monitoring method for the surface roughness of the part in the numerical control machining process according to any one of claims 1-4.

8. A computer program product comprising a computer program, characterized in that, The computer program is executed by the processor to implement the monitoring method for the surface roughness of the part in the numerical control machining process according to any one of claims 1-4.

Images