Numerical control tool machining monitoring method

Abstract

The invention discloses a numerical control tool machining monitoring method. The method comprises the following steps that firstly, data collection and primary processing are conducted; 2, constructing a digital twinborn model; 3, adjusting a self-adaptive threshold value; 4, performing prediction maintenance and fault early warning; and 5, optimizing machining parameters in real time. Multi-dimensional data such as temperature, vibration, stress and cutting force are collected in real time through a multi-sensor system, high-frequency sampling and signal denoising technologies are combined, comprehensiveness and accuracy of monitoring data are ensured, and a foundation is laid for follow-up analysis; the virtual machining model is constructed through the digital twinning technology, real-time simulation and synchronization of the machining process are achieved, the current state of the tool can be comprehensively reflected, anomaly detection and pre-judgment are conducted in combination with historical data and real-time data, and the fault monitoring accuracy is improved.

Description

Technical Field

[0001] The invention relates to the technical field of tool processing, and in particular to a numerical control tool processing monitoring method. Background Art

[0002] With the development of modern manufacturing, CNC machine tools have become important tools for high-precision and high-efficiency processing. In the CNC processing process, the status of the tool has a direct impact on the processing quality and efficiency. However, the existing technology still has the following problems in tool monitoring and maintenance:

[0003] 1. Incomplete data collection: Traditional tool monitoring mostly uses a single sensor or low-frequency data collection method, which leads to the loss of key data or delayed collection, and cannot fully reflect the processing status.

[0004] 2. Fixed monitoring threshold: Most existing systems judge tool status based on fixed thresholds, which makes it difficult to adapt to complex and changeable processing conditions and different material properties.

[0005] 3. Lack of intelligent prediction and fault warning: Existing technologies mainly rely on manual experience to judge tool wear or failure risks, and lack data-driven intelligent prediction methods, which often leads to sudden failures.

[0006] 4. Parameter optimization lags: Existing machining parameter adjustments are mostly set manually, and it is difficult to dynamically optimize according to the actual status of the tool and machining requirements in a timely manner.

[0007] In response to the above problems, the introduction of digital twin technology, deep reinforcement learning and time series analysis technology provides new solutions for CNC tool processing monitoring and optimization. These technologies can realize intelligent monitoring and dynamic maintenance of the processing process based on real-time multi-source data collection and combined with the collaborative optimization of virtual and real. Summary of the invention

[0008] The purpose of this section is to summarize some aspects of embodiments of the present invention and briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section and the specification abstract and the invention title of this application to avoid blurring the purpose of this section, the specification abstract and the invention title, and such simplifications or omissions cannot be used to limit the scope of the present invention.

[0009] In view of the above problems existing in the existing CNC tool machining monitoring method, the present invention is proposed.

[0010] Therefore, the purpose of the present invention is to provide a CNC tool processing monitoring method, which constructs a virtual processing model through digital twin technology, realizes real-time simulation and synchronization of the processing process, can comprehensively reflect the current state of the tool, and combines historical data and real-time data for abnormality detection and prediction, thereby improving the accuracy of fault monitoring.

[0011] In order to solve the above technical problems, the present invention provides the following technical solutions: a CNC tool processing monitoring method, comprising the following steps:

[0012] Step 1: Data collection and preliminary processing

[0013] Multi-source data acquisition: During CNC machining, the multi-dimensional data of the tool is collected in real time through a multi-sensor system, including but not limited to: tool temperature, vibration, stress, acoustic data; cutting force, cutting speed, feed rate; surface quality parameters of the workpiece during machining;

[0014] Among them, these data are uploaded to the central monitoring system through the sensor network and collected at a high frequency to ensure the real-time and comprehensiveness of the data;

[0015] Data preprocessing: Use signal processing algorithms to filter and denoise the collected data, remove noise interference during the collection process, and ensure data accuracy;

[0016] Step 2: Digital twin model construction

[0017] Establish a virtual processing model: According to the process parameters, equipment characteristics and historical processing data of CNC tool processing, build a corresponding digital twin model, which can reflect the current processing status in real time and keep pace with the actual processing process;

[0018] Real-time data synchronization: Update the state parameters in the digital twin model through real-time transmission of processing data, so that the virtual model is consistent with the actual tool processing process;

[0019] Abnormal detection and prediction: Based on the virtual model, the machining process is simulated, and real-time data and predictive analysis are combined to determine whether the tool is in an abnormal working state or is about to reach the threshold state;

[0020] Step 3: Adaptive Threshold Adjustment

[0021] Threshold initialization: Initially, according to the tool material and processing conditions, set the basic tool wear threshold and other monitoring parameters, such as the standard limit of temperature and vibration;

[0022] Adaptive adjustment mechanism:

[0023] Through real-time monitoring data and virtual feedback in the digital twin model, the threshold is adaptively adjusted using a deep reinforcement learning algorithm to ensure optimal tool status monitoring and maintenance in different environments.

[0024] Step 4: Predictive maintenance and fault warning

[0025] Historical data analysis: Combine the tool’s historical data with the long-term operation data of the digital twin model and apply time series analysis technology to predict tool wear trends and failure probabilities;

[0026] Fault warning mechanism: When there is a large deviation between the real-time monitoring data of the tool and the expected normal data in the digital twin model, the system automatically issues a fault warning signal and reminds the operator through a visual monitoring interface;

[0027] Step 5: Optimize processing parameters in real time

[0028] Co-optimization of virtual and real world: The digital twin model is not only used for monitoring, but also for real-time optimization of machining parameters. Based on the real-time tool status and machining data, the system simulates the virtual machining path and calculates the optimal tool cutting path, cutting force and feed speed parameters.

[0029] Automatic parameter adjustment: According to the optimization results of the digital twin model, the processing parameters of the CNC machine tool are automatically adjusted to ensure that the tool operates in the best condition and maximize the processing efficiency, processing accuracy and tool life.

[0030] As a preferred solution of the CNC tool machining monitoring method of the present invention, the specific content of the digital twin model in step 2 includes:

[0031] Model input data:

[0032] 1) Process parameters: cutting speed (V), feed rate (f), cutting depth (a);

[0033] 2) Equipment characteristics: tool material, tool geometry, machine tool characteristics;

[0034] 3) Historical processing data: tool wear data, processing accuracy, and cutting force history under different working conditions;

[0035] Model structure:

[0036] Physical model: describes the physical phenomena of the tool during the machining process, such as cutting force and temperature distribution, and uses finite element analysis for simulation;

[0037] Data-driven model: Use machine learning algorithms to process historical data and generate a predictive model for tool wear.

[0038] As a preferred solution of the CNC tool machining monitoring method of the present invention, wherein: the cutting force model: according to the basic formula of cutting force, the cutting force F can be expressed by the following formula:

[0039] F=K t ·A·f·V

[0040] Among them, K t is the cutting constant, A is the cutting area, f is the feed rate, and V is the cutting speed;

[0041] Temperature model: The change of tool temperature T can be described by the heat conduction equation:

[0042]

[0043] Among them, α is the thermal diffusivity, Q is the heat generated during cutting, ρ is the material density, and c is the specific heat capacity;

[0044] Tool wear model: Tool wear can be expressed by the wear rate W, which is calculated according to the following formula:

[0045] W=K·T·f·V n

[0046] Among them, K is the wear coefficient, n is the empirical coefficient, T is the cutting temperature, f is the feed rate, and V is the cutting speed;

[0047] Machine learning model: Use machine self-learning algorithm to train historical data to generate a predictive model of tool status.

[0048] As a preferred solution of the CNC tool machining monitoring method of the present invention, the threshold judgment and analysis process in step 2 is specifically as follows:

[0049] S1. Threshold setting:

[0050] Initial threshold: Set the initial wear threshold W according to the tool material processing conditions. thresh , temperature threshold T thresh and the vibration threshold V thresh , obtained through industry standards or experimental data;

[0051] Adaptive adjustment: Using real-time data and digital twin models, the threshold is dynamically adjusted through machine learning algorithms;

[0052] S2. Real-time monitoring:

[0053] During the machining process, the cutting force, temperature and wear status data of the tool are collected in real time and compared with the set thresholds;

[0054] S3. Abnormal judgment:

[0055] Wear judgment: If the current wear rate W current More than W thresh , then it is judged that the tool wear is abnormal;

[0056] Temperature judgment: If the current tool temperature T current More than T thresh, then it is judged that the tool is overheated;

[0057] Vibration judgment: If the vibration amplitude detected by the vibration sensor exceeds the threshold V thresh , then it is judged that the tool is at risk of failure;

[0058] S4. Early warning mechanism:

[0059] When any parameter exceeds the set threshold, the system automatically sends out a warning signal to prompt the operator to conduct inspection or maintenance, and records relevant data for subsequent analysis and optimization.

[0060] As a preferred solution of the CNC tool machining monitoring method of the present invention, wherein: the machine learning algorithm is used for regression and classification tasks;

[0061] Specifically, the regression model is used to predict the degree of wear:

[0062] Model selection: Select linear regression or random forest regression algorithm;

[0063] Among them, linear regression, model formula:

[0064] W=β0+β1·T+β2·f+β3·V+∈

[0065] W: predicted tool wear; T: cutting temperature; f: feed rate; V: cutting speed; β0, β1, β2, β3: model parameters; ∈: error term;

[0066] Among them, random forest regression, model formula:

[0067]

[0068] W RF : Wear degree predicted by random forest; T i (X): the predicted value of the i-th tree under the input feature X; N: the total number of trees;

[0069] Specifically, the classification model is used to determine whether the tool is in a normal state

[0070] Model Selection: Logistic Regression and Support Vector Machines

[0071] Among them, logistic regression, model formula:

[0072]

[0073] P(Y=1|X): probability that the tool is in a normal state;

[0074] Y: Status label (1 for normal, 0 for abnormal)

[0075] X: Feature vector (cutting temperature, feed rate, cutting speed)

[0076] β0, β1, β2, β3: model parameters;

[0077] Among them, support vector machine, model formula:

[0078]

[0079] w: normal vector of the hyperplane; b: bias term; X: input feature vector.

[0080] As a preferred solution of the CNC tool machining monitoring method of the present invention, the machine learning algorithm training process includes the following steps:

[0081] Step S1, data preparation: collect historical processing data, including cutting temperature, feed rate, cutting speed, wear degree characteristic data, and mark the normal or abnormal state of the tool;

[0082] Step S2, data preprocessing: processing missing values ​​and outliers; feature scaling; data set division (training set, validation set and test set);

[0083] Step S3, model training: For the regression model, the mean square error is used as the loss function for training:

[0084]

[0085] For classification models, the cross entropy loss function of logistic regression is used:

[0086]

[0087] Step S4: Model evaluation:

[0088] Regression Model: Using the Coefficient of Determination R 2 and root mean square error to evaluate model performance;

[0089] Classification model: Use confusion matrix, precision, recall and F1-score indicators to evaluate model performance;

[0090] Step S5, model optimization: According to the evaluation results, adjust the model parameters or select different features for training to improve the prediction performance of the model.

[0091] As a preferred solution of the CNC tool machining monitoring method of the present invention, the deep reinforcement learning algorithm in step 3 adaptively adjusts the threshold value and adopts the Q-function definition:

[0092]

[0093] Q(St , A t ): In state S t Next take action A t Expected return; γ: discount factor, between 0 and 1;

[0094] S t Includes the following: current wear status of the tool; real-time monitoring data of the tool; parameters of the current processing task; past adjustment history; A t Including: increasing the temperature threshold; reducing the temperature threshold; increasing the wear threshold; reducing the wear threshold;

[0095] Reward function R t Used to evaluate the effect of each adjustment:

[0096]

[0097] The reward function should encourage the agent to make effective threshold adjustments to keep the tool state within a safe range;

[0098] Use a neural network to approximate the Q-function:

[0099] Q(S t , A t ;θ)≈Q(S t , A)

[0100] Where θ is the network weight;

[0101] Loss function: Update the Q-value using mean squared error:

[0102]

[0103] Network Update:

[0104] Use experience replay to store tuples of historical states, actions, rewards, and next states, perform batch training, and regularly update the target network to improve learning stability.

[0105] As a preferred solution of the CNC tool machining monitoring method of the present invention, the specific implementation process of the time series analysis technology in step 4 includes the following steps:

[0106] 1) Data collection: extracting long-term operation data and historical wear data of the tool from the digital twin model;

[0107] 2) Data preprocessing:

[0108] Clean up missing values ​​and outliers;

[0109] Perform stationarity test and difference processing on the data;

[0110] Normalization and time windowing;

[0111] 3) Model construction:

[0112] Construct a time series analysis model using the ARIMA model:

[0113]

[0114] Y t : The wear rate at the current moment; Autoregressive coefficient; θ: moving average coefficient; ∈: white noise term;

[0115] 4) Model training:

[0116] Train time series models using historical data and evaluate performance;

[0117] 5) Prediction and evaluation:

[0118] Use the constructed time series model to predict future wear trends;

[0119] Calculate the probability of failure based on the predicted wear rate, and make evaluation and comparison;

[0120] 6) Result analysis:

[0121] Visualize the comparison between the predicted results and the actual values ​​to evaluate the prediction accuracy of the model.

[0122] Beneficial effects of the present invention: 1. Comprehensive and real-time data collection: Real-time collection of multi-dimensional data such as temperature, vibration, stress, cutting force, etc. through a multi-sensor system, combined with high-frequency sampling and signal denoising technology, to ensure the comprehensiveness and accuracy of monitoring data, laying the foundation for subsequent analysis.

[0123] 2. Efficient application of digital twin models: By building a virtual processing model through digital twin technology, real-time simulation and synchronization of the processing process can be achieved, which can fully reflect the current status of the tool, and combine historical data and real-time data for abnormal detection and prediction, thereby improving the accuracy of fault monitoring.

[0124] 3. Adaptive threshold adjustment: Introduce deep reinforcement learning algorithm to dynamically adjust the monitoring threshold according to different processing conditions, tool materials and other factors, so that the tool monitoring system has strong environmental adaptability and avoids misjudgment or missed judgment under fixed threshold mode.

[0125] 4. Intelligent prediction and fault warning: Through time series analysis and deep learning technology, tool wear trends and failure probabilities are predicted, and a high-precision predictive maintenance mechanism is established. When an abnormality occurs, a warning signal is issued in time, significantly reducing the failure rate.

[0126] 5. Real-time optimization of processing parameters: Based on the virtual simulation capability of the digital twin model, the processing parameters are optimized and adjusted to achieve dynamic adjustment of the cutting path, cutting force and feed speed, improve processing accuracy and efficiency, and extend the service life of the tool. BRIEF DESCRIPTION OF THE DRAWINGS

[0127] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative labor. Among them:

[0128] Figure 1 It is a schematic diagram of the structure of the method for monitoring CNC tool machining of the present invention. DETAILED DESCRIPTION

[0129] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0130] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0131] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0132] Secondly, the present invention is described in detail with reference to the schematic diagram. When describing the embodiments of the present invention in detail, for the sake of convenience, the cross-sectional diagrams showing the device structure will not be partially enlarged according to the general scale, and the schematic diagrams are only examples, which should not limit the scope of protection of the present invention. In addition, in actual production, the three-dimensional dimensions of length, width and depth should be included.

[0133] Reference Figure 1 , provides a CNC tool processing monitoring method, comprising the following steps:

[0134] Step 1: Data collection and preliminary processing

[0135] Multi-source data acquisition: During CNC machining, the multi-dimensional data of the tool is collected in real time through a multi-sensor system, including but not limited to: tool temperature, vibration, stress, acoustic data; cutting force, cutting speed, feed rate; surface quality parameters of the workpiece during machining;

[0136] Among them, these data are uploaded to the central monitoring system through the sensor network and collected at a high frequency to ensure the real-time and comprehensiveness of the data;

[0137] Data preprocessing: Use signal processing algorithms to filter and denoise the collected data, remove noise interference during the collection process, and ensure data accuracy;

[0138] Step 2: Digital twin model construction

[0139] Establish a virtual processing model: According to the process parameters, equipment characteristics and historical processing data of CNC tool processing, build a corresponding digital twin model, which can reflect the current processing status in real time and keep pace with the actual processing process;

[0140] Real-time data synchronization: Update the state parameters in the digital twin model through real-time transmission of processing data, so that the virtual model is consistent with the actual tool processing process;

[0141] Abnormal detection and prediction: Based on the virtual model, the machining process is simulated, and real-time data and predictive analysis are combined to determine whether the tool is in an abnormal working state or is about to reach the threshold state;

[0142] Step 3: Adaptive Threshold Adjustment

[0143] Threshold initialization: Initially, according to the tool material and processing conditions, set the basic tool wear threshold and other monitoring parameters, such as the standard limit of temperature and vibration;

[0144] Adaptive adjustment mechanism:

[0145] Through real-time monitoring data and virtual feedback in the digital twin model, the threshold is adaptively adjusted using a deep reinforcement learning algorithm to ensure optimal tool status monitoring and maintenance in different environments.

[0146] Step 4: Predictive maintenance and fault warning

[0147] Historical data analysis: Combine the tool’s historical data with the long-term operation data of the digital twin model and apply time series analysis technology to predict tool wear trends and failure probabilities;

[0148] Fault warning mechanism: When there is a large deviation between the real-time monitoring data of the tool and the expected normal data in the digital twin model, the system automatically issues a fault warning signal and reminds the operator through a visual monitoring interface;

[0149] Step 5: Optimize processing parameters in real time

[0150] Co-optimization of virtual and real world: The digital twin model is not only used for monitoring, but also for real-time optimization of machining parameters. Based on the real-time tool status and machining data, the system simulates the virtual machining path and calculates the optimal tool cutting path, cutting force and feed speed parameters.

[0151] Automatic parameter adjustment: According to the optimization results of the digital twin model, the processing parameters of the CNC machine tool are automatically adjusted to ensure that the tool operates in the best condition and maximize the processing efficiency, processing accuracy and tool life.

[0152] Among them, the specific contents of the digital twin model in step 2 include:

[0153] Model input data:

[0154] 1) Process parameters: cutting speed (V), feed rate (f), cutting depth (a);

[0155] 2) Equipment characteristics: tool material, tool geometry, machine tool characteristics;

[0156] 3) Historical processing data: tool wear data, processing accuracy, and cutting force history under different working conditions;

[0157] Model structure:

[0158] Physical model: describes the physical phenomena of the tool during the machining process, such as cutting force and temperature distribution, and uses finite element analysis for simulation;

[0159] Data-driven model: Use machine learning algorithms to process historical data and generate a predictive model for tool wear.

[0160] Further, cutting force model: According to the basic formula of cutting force, the cutting force F can be expressed by the following formula:

[0161] F=K t ·A·f·V

[0162] Among them, K t is the cutting constant, A is the cutting area, f is the feed rate, and V is the cutting speed;

[0163] Temperature model: The change of tool temperature T can be described by the heat conduction equation:

[0164]

[0165] Among them, α is the thermal diffusivity, Q is the heat generated during cutting, ρ is the material density, and c is the specific heat capacity;

[0166] Tool wear model: Tool wear can be expressed by the wear rate W, which is calculated according to the following formula:

[0167] W=K·T·f·V n

[0168] Among them, K is the wear coefficient, n is the empirical coefficient, T is the cutting temperature, f is the feed rate, and V is the cutting speed;

[0169] Machine learning model: Use machine self-learning algorithm to train historical data to generate a predictive model of tool status.

[0170] Specifically, the threshold judgment and analysis process in step 2 is as follows:

[0171] S1. Threshold setting:

[0172] Initial threshold: Set the initial wear threshold W according to the tool material processing conditions. thresh , temperature threshold T thresh and the vibration threshold V thresh , obtained through industry standards or experimental data;

[0173] Adaptive adjustment: Using real-time data and digital twin models, the threshold is dynamically adjusted through machine learning algorithms;

[0174] S2. Real-time monitoring:

[0175] During the machining process, the cutting force, temperature and wear status data of the tool are collected in real time and compared with the set thresholds;

[0176] S3. Abnormal judgment:

[0177] Wear judgment: If the current wear rate W current More than W thresh , then it is judged that the tool wear is abnormal;

[0178] Temperature judgment: If the current tool temperature T current More than T thresh , then it is judged that the tool is overheated;

[0179] Vibration judgment: If the vibration amplitude detected by the vibration sensor exceeds the threshold V thresh , then it is judged that the tool is at risk of failure;

[0180] S4. Early warning mechanism:

[0181] When any parameter exceeds the set threshold, the system automatically sends out a warning signal to prompt the operator to conduct inspection or maintenance, and records relevant data for subsequent analysis and optimization.

[0182] Among them, machine learning algorithms are used for regression and classification tasks;

[0183] Specifically, the regression model is used to predict the degree of wear:

[0184] Model selection: Select linear regression or random forest regression algorithm;

[0185] Among them, linear regression, model formula:

[0186] W=β0+β1·T+β2·f+β3·V+∈

[0187] W: predicted tool wear; T: cutting temperature; f: feed rate; V: cutting speed; β0, β1, β2, β3: model parameters; ∈: error term;

[0188] Among them, random forest regression, model formula:

[0189]

[0190] W RF : Wear degree predicted by random forest; T i (X): the predicted value of the i-th tree under the input feature X; N: the total number of trees;

[0191] Specifically, the classification model is used to determine whether the tool is in a normal state

[0192] Model Selection: Logistic Regression and Support Vector Machines

[0193] Among them, logistic regression, model formula:

[0194]

[0195] P(Y=1|X): probability that the tool is in a normal state;

[0196] Y: Status label (1 for normal, 0 for abnormal)

[0197] X: Feature vector (cutting temperature, feed rate, cutting speed)

[0198] β0, β1, β2, β3: model parameters;

[0199] Among them, support vector machine, model formula:

[0200]

[0201] w: normal vector of the hyperplane; b: bias term; X: input feature vector.

[0202] Furthermore, the machine learning algorithm training process includes the following steps:

[0203] Step S1, data preparation: collect historical processing data, including cutting temperature, feed rate, cutting speed, wear degree characteristic data, and mark the normal or abnormal state of the tool;

[0204] Step S2, data preprocessing: processing missing values ​​and outliers; feature scaling; data set division (training set, validation set and test set);

[0205] Step S3, model training: For the regression model, the mean square error is used as the loss function for training:

[0206]

[0207] For classification models, the cross entropy loss function of logistic regression is used:

[0208]

[0209] Step S4: Model evaluation:

[0210] Regression Model: Using the Coefficient of Determination R 2 and root mean square error to evaluate model performance;

[0211] Classification model: Use confusion matrix, precision, recall and F1-score indicators to evaluate model performance;

[0212] Step S5, model optimization: According to the evaluation results, adjust the model parameters or select different features for training to improve the prediction performance of the model.

[0213] Among them, the deep reinforcement learning algorithm in step 3 adaptively adjusts the threshold and adopts the Q-function definition:

[0214]

[0215] Q(S t , A t ): In state S t Next take action A t Expected return; γ: discount factor, between 0 and 1;

[0216] S t Includes the following: current wear status of the tool; real-time monitoring data of the tool; parameters of the current processing task; past adjustment history; A t Including: increasing the temperature threshold; reducing the temperature threshold; increasing the wear threshold; reducing the wear threshold;

[0217] Reward function R t Used to evaluate the effect of each adjustment:

[0218]

[0219] The reward function should encourage the agent to make effective threshold adjustments to keep the tool state within a safe range;

[0220] Use a neural network to approximate the Q-function:

[0221] Q(S t , A t ;θ)≈Q(S t , A)

[0222] Where θ is the network weight;

[0223] Loss function: Update the Q-value using mean squared error:

[0224]

[0225] Network Update:

[0226] Use experience replay to store tuples of historical states, actions, rewards, and next states, perform batch training, and regularly update the target network to improve learning stability.

[0227] Specifically, the specific implementation process of the time series analysis technology in step 4 includes the following steps:

[0228] 1) Data collection: extracting long-term operation data and historical wear data of the tool from the digital twin model;

[0229] 2) Data preprocessing:

[0230] Clean up missing values ​​and outliers;

[0231] Perform stationarity test and difference processing on the data;

[0232] Normalization and time windowing;

[0233] 3) Model construction:

[0234] Construct a time series analysis model using the ARIMA model:

[0235]

[0236] Y t : The wear rate at the current moment; Autoregressive coefficient; θ: moving average coefficient; ∈: white noise term;

[0237] 4) Model training:

[0238] Train time series models using historical data and evaluate performance;

[0239] 5) Prediction and evaluation:

[0240] Use the constructed time series model to predict future wear trends;

[0241] Calculate the probability of failure based on the predicted wear rate, and make evaluation and comparison;

[0242] 6) Result analysis:

[0243] Visualize the comparison between the predicted results and the actual values ​​to evaluate the prediction accuracy of the model.

[0244] This embodiment also provides a computer device, which is applicable to a CNC tool processing monitoring method, including: a memory and a processor; the memory is used to store computer executable instructions, and the processor is used to execute computer executable instructions to implement a CNC tool processing monitoring method proposed in the above embodiment.

[0245] The computer device may be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected via a system bus. The processor of the computer device 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 and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covering the display screen, or a key, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse, etc.

[0246] This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, a CNC tool processing monitoring method proposed in the above embodiment is implemented; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (Static Random Access Memory, referred to as SRAM), electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, referred to as EEPROM), erasable programmable read-only memory (Erasable Programmable Read Only Memory, referred to as EPROM), programmable read-only memory (Programmable Red-Only Memory, referred to as PROM), read-only memory (Read-Only Memory, referred to as ROM), magnetic storage, flash memory, magnetic disk or optical disk.

[0247] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A CNC tool processing monitoring method, characterized in that: The following steps are involved: Step 1: Data collection and preliminary processing Multi-source data acquisition: During CNC machining, the multi-dimensional data of the tool is collected in real time through a multi-sensor system, including but not limited to: tool temperature, vibration, stress, acoustic data; cutting force, cutting speed, feed rate; surface quality parameters of the workpiece during machining; Among them, these data are uploaded to the central monitoring system through the sensor network and collected at a high frequency to ensure the real-time and comprehensiveness of the data; Data preprocessing: Use signal processing algorithms to filter and denoise the collected data, remove noise interference during the collection process, and ensure data accuracy; Step 2: Digital twin model construction Establish a virtual processing model: According to the process parameters, equipment characteristics and historical processing data of CNC tool processing, build a corresponding digital twin model, which can reflect the current processing status in real time and keep pace with the actual processing process; Real-time data synchronization: Update the state parameters in the digital twin model through real-time transmission of processing data, so that the virtual model is consistent with the actual tool processing process; Abnormal detection and prediction: Based on the virtual model, the machining process is simulated, and real-time data and predictive analysis are combined to determine whether the tool is in an abnormal working state or is about to reach the threshold state; Step 3: Adaptive Threshold Adjustment Threshold initialization: Initially, according to the tool material and processing conditions, set the basic tool wear threshold and other monitoring parameters, such as the standard limit of temperature and vibration; Adaptive adjustment mechanism: Through real-time monitoring data and virtual feedback in the digital twin model, the threshold is adaptively adjusted using a deep reinforcement learning algorithm to ensure optimal tool status monitoring and maintenance in different environments. Step 4: Predictive maintenance and fault warning Historical data analysis: Combine the tool’s historical data with the long-term operating data of the digital twin model and apply time series analysis technology to predict tool wear trends and failure probabilities; Fault warning mechanism: When there is a large deviation between the real-time monitoring data of the tool and the expected normal data in the digital twin model, the system automatically issues a fault warning signal and reminds the operator through a visual monitoring interface; Step 5: Optimize processing parameters in real time Co-optimization of virtual and real world: The digital twin model is not only used for monitoring, but also for real-time optimization of machining parameters. Based on the real-time tool status and machining data, the system simulates the virtual machining path and calculates the optimal tool cutting path, cutting force and feed speed parameters. Automatic parameter adjustment: According to the optimization results of the digital twin model, the processing parameters of the CNC machine tool are automatically adjusted to ensure that the tool operates in the best condition and maximize the processing efficiency, processing accuracy and tool life.

2. The CNC tool machining monitoring method according to claim 1, characterized in that: The specific contents of the digital twin model in step 2 include: Model input data: 1) Process parameters: cutting speed (V), feed rate (f), cutting depth (a); 2) Equipment characteristics: tool material, tool geometry, machine tool characteristics; 3) Historical processing data: tool wear data, processing accuracy, and cutting force history under different working conditions; Model structure: Physical model: describes the physical phenomena of the tool during the machining process, such as cutting force and temperature distribution, and uses finite element analysis for simulation; Data-driven model: Use machine learning algorithms to process historical data and generate a predictive model for tool wear.

3. The CNC tool machining monitoring method according to claim 2, characterized in that: The cutting force model: According to the basic formula of cutting force, the cutting force F can be expressed by the following formula: F=K t ·A·f·V Among them, K t is the cutting constant, A is the cutting area, f is the feed rate, and V is the cutting speed; Temperature model: The change of tool temperature T can be described by the heat conduction equation: Among them, α is the thermal diffusivity, Q is the heat generated during cutting, ρ is the material density, and c is the specific heat capacity; Tool wear model: Tool wear can be expressed by the wear rate W, which is calculated according to the following formula: W=K·T·f·V n Among them, K is the wear coefficient, n is the empirical coefficient, T is the cutting temperature, f is the feed rate, and V is the cutting speed; Machine learning model: Use machine self-learning algorithm to train historical data to generate a predictive model of tool status.

4. The CNC tool machining monitoring method according to claim 3, characterized in that: The threshold judgment and analysis process in step 2 is specifically as follows: S1. Threshold setting: Initial threshold: Set the initial wear threshold W according to the tool material processing conditions. thresh , temperature threshold T thresh and the vibration threshold V thresh , obtained through industry standards or experimental data; Adaptive adjustment: Using real-time data and digital twin models, the threshold is dynamically adjusted through machine learning algorithms; S2. Real-time monitoring: During the machining process, the cutting force, temperature and wear status data of the tool are collected in real time and compared with the set thresholds; S3. Abnormal judgment: Wear judgment: If the current wear rate W current More than W thresh , then it is judged that the tool wear is abnormal; Temperature judgment: If the current tool temperature T current More than T thresh , then it is judged that the tool is overheated; Vibration judgment: If the vibration amplitude detected by the vibration sensor exceeds the threshold V thresh , then it is judged that the tool is at risk of failure; S4. Early warning mechanism: When any parameter exceeds the set threshold, the system automatically sends out a warning signal to prompt the operator to conduct inspection or maintenance, and records relevant data for subsequent analysis and optimization.

5. The CNC tool machining monitoring method according to claim 3, characterized in that: The machine learning algorithms are used for regression and classification tasks; Specifically, the regression model is used to predict the degree of wear: Model selection: Select linear regression or random forest regression algorithm; Among them, linear regression, model formula: W=β0+β1T+β2·f+β3·V+∈ W: predicted tool wear; T: cutting temperature; f: feed rate; V: cutting speed; β0, β1, β2, β3: model parameters; ∈: error term; Among them, random forest regression, model formula: W RF : Wear degree predicted by random forest; T i (X): the predicted value of the i-th tree under the input feature X; N: the total number of trees; Specifically, the classification model is used to determine whether the tool is in a normal state Model Selection: Logistic Regression and Support Vector Machines Among them, logistic regression, model formula: P(Y=1|X): probability that the tool is in a normal state; Y: Status label (1 for normal, 0 for abnormal) X: Feature vector (cutting temperature, feed rate, cutting speed) β0, β1, β2, β3: model parameters; Among them, support vector machine, model formula: w: normal vector of the hyperplane; b: bias term; X: input feature vector.

6. The CNC tool machining monitoring method according to claim 5, characterized in that: The machine learning algorithm training process includes the following steps: Step S1, data preparation: collect historical processing data, including cutting temperature, feed rate, cutting speed, wear degree characteristic data, and mark the normal or abnormal state of the tool; Step S2, data preprocessing: processing missing values ​​and outliers; feature scaling; data set division (training set, validation set and test set); Step S3, model training: For the regression model, the mean square error is used as the loss function for training: For classification models, the cross entropy loss function of logistic regression is used: Step S4: Model evaluation: Regression Model: Using the Coefficient of Determination R 2 and root mean square error to evaluate model performance; Classification model: Use confusion matrix, precision, recall and F1-score indicators to evaluate model performance; Step S5, model optimization: according to the evaluation results, adjust the model parameters or select different features for training to improve the prediction performance of the model.

7. The CNC tool machining monitoring method according to claim 1, characterized in that: In the step 3, the deep reinforcement learning algorithm adaptively adjusts the threshold and adopts the Q-function definition: Q(S t ,A t ): In state S t Next take action A t Expected return; γ: discount factor, between 0 and 1; S t Includes the following: current wear status of the tool; real-time monitoring data of the tool; parameters of the current processing task; past adjustment history; A t Including: increasing the temperature threshold; reducing the temperature threshold; increasing the wear threshold; reducing the wear threshold; Reward function R t Used to evaluate the effect of each adjustment: The reward function should encourage the agent to make effective threshold adjustments to keep the tool state within a safe range; Use a neural network to approximate the Q-function: Q(S t ,A t ;θ)≈Q(S t ,A) Where θ is the network weight; Loss function: Update the Q-value using mean squared error: Network Update: Use experience replay to store tuples of historical states, actions, rewards, and next states, perform batch training, and regularly update the target network to improve learning stability.

8. The CNC tool machining monitoring method according to claim 5, characterized in that: The specific implementation process of the time series analysis technology in step 4 includes the following steps: 1) Data collection: extracting long-term operation data and historical wear data of the tool from the digital twin model; 2) Data preprocessing: Clean up missing values ​​and outliers; Perform stationarity test and difference processing on the data; Normalization and time windowing; 3) Model construction: Construct a time series analysis model using the ARIMA model: Y t : The wear rate at the current moment; Autoregressive coefficient; θ: moving average coefficient; ∈: white noise term; 4) Model training: Train time series models using historical data and evaluate performance; 5) Prediction and evaluation: Use the constructed time series model to predict future wear trends; Calculate the probability of failure based on the predicted wear rate, and make evaluation and comparison; 6) Result analysis: Visualize the comparison between the predicted results and the actual values ​​to evaluate the prediction accuracy of the model.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of a CNC tool machining monitoring method according to any one of claims 1 to 8 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of a CNC tool machining monitoring method according to any one of claims 1 to 8 are implemented.

Images