Numerical control machine tool cutter regulation and control method and system based on cutting state diagnosis

Abstract

The invention discloses a numerical control machine tool cutter regulation and control method and system based on cutting state diagnosis. The method comprises the following steps that S1, cutting state original data in the machining process of a numerical control machine tool are obtained and preprocessed; s2, inputting the preprocessed cutting state data into the trained cutting state diagnosis model, and outputting the current wear grade of the tool and a machining risk assessment result; and S3, based on the current wear grade of the cutter, the processing risk assessment result and a preset cutter life optimization target, the invention relates to the technical field of numerical control machine tool cutter regulation and control. According to the numerical control machine tool cutter regulation and control method and system based on cutting state diagnosis, a self-adaptive control algorithm is adopted, two optimization targets of cutting efficiency and cutter service life prolonging are combined, it is ensured that under different abrasion states, the optimized machining parameters can maximize the machining efficiency and the cutter service life, and the service life of the cutter is prolonged. The self-adaptive adjustment mode can be continuously optimized according to the actual machining state, and the machining precision and the production efficiency are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of numerical control machine tool cutter regulation, more particularly, to a numerical control machine tool cutter regulation method and system based on cutting state diagnosis. BACKGROUND

[0002] In the numerical control machine tool processing process, the wear state of the cutter directly affects the processing quality and efficiency. The traditional cutter regulation method often relies on fixed cutting parameters, lacks real-time perception and dynamic adjustment ability of the actual cutting state. This leads to the fact that in actual processing, the cutter may fail prematurely due to excessive wear, increasing the frequency and cost of replacing the cutter; or due to unreasonable cutting parameters, resulting in low processing efficiency, which cannot fully exert the performance of the numerical control machine tool.

[0003] With the increasing requirements of manufacturing industry on processing precision and efficiency, the existing cutter regulation method has been difficult to meet the actual production needs. Especially under complex working conditions, the wear process of the cutter is more complex, and the traditional experience judgment and fixed parameter setting method cannot accurately adapt to different processing scenes. Therefore, a cutter regulation method is needed, which can monitor the cutting state in real time, accurately diagnose the wear grade of the cutter, evaluate the processing risk, and dynamically adjust the cutting parameters, so as to improve the service life and processing efficiency of the cutter and reduce the processing cost. SUMMARY

[0004] The purpose of the present application is to provide a numerical control machine tool cutter regulation method and system based on cutting state diagnosis, which solves the problem that the existing cutter regulation method cannot accurately adapt to different processing scenes according to the traditional experience judgment and fixed parameter setting method.

[0005] The present application realizes the above-mentioned purpose through the following technical scheme: a numerical control machine tool cutter regulation method based on cutting state diagnosis, comprising the following steps: S1, acquiring cutting state original data in the numerical control machine tool processing process and preprocessing; S2, inputting the preprocessed cutting state data into the trained cutting state diagnosis model, and outputting the current wear grade of the cutter and the processing risk evaluation result; S3, based on the current wear grade of the cutter, the processing risk evaluation result and the preset cutter life optimization target, generating a cutting parameter regulation strategy through a self-adaptive control algorithm, the cutting parameter at least including one or more of feed speed, cutting depth and spindle speed; S4, issuing the cutting parameter regulation strategy to the numerical control machine tool actuator to complete the dynamic regulation of the cutter cutting process; S5, real-time acquisition of the cutting state feedback data of the regulated numerical control machine tool, iterative optimization of the cutting parameter regulation strategy, until the cutter life optimization target is met or the processing task is completed.

[0006] Further, the step S1 acquires the cutting state original data in the machining process of the numerical control machine tool and performs preprocessing, including: The cutting state original data is collected by a multi-sensor array, the multi-sensor array at least includes one or more of cutting force sensors, vibration sensors, temperature sensors, acoustic emission sensors, and the collected original data constitutes a multi-dimensional time series data set; The multi-dimensional time series data set is preprocessed, and the preprocessing step at least includes one or more of outlier rejection, data smoothing, and normalization processing.

[0007] Further, the data smoothing is smoothed by a filtering algorithm on the data after the abnormal value is rejected, and the filtering algorithm includes a sliding average filter, and the sliding window length of the sliding average filter is set to 5-20; The normalization processing maps the smoothed data to a preset numerical interval, and the preset numerical interval includes the [0, 1] interval.

[0008] Further, in step S2: The cutting state diagnosis model is a model based on deep learning, and the model at least includes one or more of a feature extraction layer, a wear grade classification layer, and a risk assessment layer; The feature extraction layer adopts a time series feature extraction network to perform time series feature extraction on the preprocessed cutting state data, and outputs a high-dimensional feature vector, and the time series feature extraction network includes a bidirectional long short-term memory network; The wear grade classification layer classifies the high-dimensional feature vector through a classification network, and outputs the current wear grade of the tool, and the classification network includes a fully connected network, and the wear grade at least includes one or more of no wear, slight wear, moderate wear, and severe wear; The risk assessment layer outputs a machining risk value based on the wear grade and real-time cutting parameters through a regression model, and the regression model includes a risk regression model, and the numerical interval of the machining risk value includes [0, 1], and the larger the value, the higher the risk.

[0009] Further, the cutting state diagnosis model is trained using a historical cutting data set, and the training process takes the weighted sum of multiple loss functions as the objective function, and the multiple loss functions at least include one or more of cross-entropy loss and mean square error loss, and the cross-entropy loss corresponds to wear grade classification, and the mean square error loss corresponds to risk assessment.

[0010] Further, in step S3, the preset tool life optimization target is set based on the current wear grade of the tool: When the current wear level of the tool is no wear or slight wear, the machining efficiency is prioritized, and a tool life extension rate target is set; When the current wear level of the tool is moderate wear or severe wear, the wear rate is prioritized to be reduced, and a tool remaining life extension rate target is set.

[0011] Further, when constructing an adaptive control target function, taking cutting parameters as optimization variables, the adaptive control target function is a weighted sum of a machining efficiency index and a tool life extension index, the machining efficiency index is calculated based on cutting parameters and normalized to a preset interval, and the tool life extension index is calculated based on a historical wear law model; According to the current wear level of the tool, the weight coefficients of the machining efficiency index and the tool life extension index are dynamically adjusted, and a cutting parameter constraint condition is set, the cutting parameter constraint condition is set based on physical limit parameters of the numerical control machine tool and the tool.

[0012] Further, an optimization algorithm is used to solve the adaptive control target function, and optimal cutting parameters are output to form a cutting parameter regulation strategy, the optimization algorithm includes a particle swarm optimization algorithm.

[0013] Further, an iteration regulation period is set in step S5, cutting state data after regulation is collected in real time in each period, and the operation of step S2 is repeated to obtain a new wear level and a risk value; A regulation effect evaluation index is constructed, the regulation effect evaluation index includes one or more of a wear improvement rate and a risk reduction rate; Based on the regulation effect evaluation index, it is determined whether the current regulation strategy is effective, if not, secondary optimization is triggered, the weight coefficients are adjusted again by executing the adaptive control algorithm, and a new cutting parameter regulation strategy is generated; The iteration process is repeated until the machining task is completed or the wear level of the tool reaches a preset severe wear threshold, triggering a tool replacement reminder.

[0014] A numerical control machine tool tool regulation system based on cutting state diagnosis is applied to execute the numerical control machine tool tool regulation method based on cutting state diagnosis, the system includes the following modules: A data acquisition module, a data preprocessing module, a cutting state diagnosis module, a regulation strategy generation module, a regulation execution module and an iteration optimization module; The data acquisition module is used to acquire cutting state original data and regulated cutting state feedback data in the machining process of the numerical control machine tool; The data preprocessing module is used to preprocess the cutting state original data and the cutting state feedback data; The cutting state diagnosis module is configured to input the preprocessed cutting state data into the trained cutting state diagnosis model, and output a current tool wear level and a machining risk assessment result. The control strategy generation module is configured to generate a cutting parameter control strategy based on the current tool wear level, the machining risk assessment result, and a preset tool life optimization target through an adaptive control algorithm. The control execution module is configured to send the cutting parameter control strategy to a numerical control machine tool actuator to complete dynamic control of the tool cutting process. The iterative optimization module is configured to iteratively optimize the cutting parameter control strategy based on cutting state feedback data until the tool life optimization target is met or the machining task is completed.

[0015] The present application has the following advantages: 1. The cutting state data of the numerical control machine tool is obtained in real time through the multi-sensor array, and the data is accurately analyzed through the deep learning model, which can monitor the wear level and machining risk of the tool in real time, and provide strong support for accurate diagnosis and dynamic control of the tool state.

[0016] 2. According to the current wear level and machining risk assessment of the tool, the system can dynamically generate a cutting parameter control strategy and adjust the machining parameters such as feed speed, cutting depth and spindle speed in real time, thereby effectively prolonging the tool life or reducing the wear rate.

[0017] 3. The adaptive control algorithm is used in combination with the cutting efficiency and tool life extension optimization targets to ensure that the optimized machining parameters can maximize the machining efficiency and tool life in different wear states, and the adaptive adjustment method can continuously optimize according to the actual machining state, thereby improving the machining precision and production efficiency.

[0018] 4. By collecting the cutting state feedback data after the control in real time, the system can further optimize the cutting parameters according to the feedback information to form a closed-loop control, so that the tool control can not only respond to real-time changes, but also continuously iteratively optimize in the entire machining process to ensure the smooth completion of the machining task.

[0019] 5. The machining risk value is calculated through the risk assessment layer, and the system can timely adjust the machining strategy to avoid machining quality problems caused by excessive tool wear or improper machining conditions, and further reduce the production risk. DETAILED DESCRIPTION

[0020] The drawings described herein are used to provide further understanding of the present application, and form a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application, and do not constitute an improper limitation on the present application. In the drawings: Figure 1Flow chart of the method of the present application; Figure 2 Flow chart of the adaptive control algorithm of the present application; Figure 3 Block diagram of the system of the present application; Figure 4 Flow chart of the iterative optimization closed loop of the present application. DETAILED DESCRIPTION

[0021] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application; obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative work are within the protection scope of the present application.

[0022] Embodiment 1: Please refer to Figure 1 、 2 , 4, the present application provides a technical solution: a numerical control machine tool cutter control method based on cutting state diagnosis, the method comprising: S1, acquiring cutting state original data in the machining process of the numerical control machine tool and preprocessing; Wherein, numerical control machine tool is the abbreviation of digital control machine tool, which is an automatic machine tool equipped with a program control system. The control system can logically process programs with control codes or other symbolic instructions and translate them into coded numbers. The numbers are input into the numerical control device through the information carrier, and various control signals are sent out by the numerical control device after operation processing, to control the action of the machine tool. The machine tool automatically processes parts according to the shape and size required by the drawing; the cutting state original data is the unprocessed various data related to the cutting process directly collected in the machining process of the numerical control machine tool, such as cutting force signal, vibration signal, acoustic emission signal, motor current signal, etc. These data reflect the real-time situation of the interaction between the tool and the workpiece in the cutting process; preprocessing is a series of processing operations on the cutting state original data to improve data quality and extract useful features for subsequent analysis; S2, inputting the preprocessed cutting state data into the trained cutting state diagnosis model, and outputting the current wear grade of the tool and the processing risk assessment result; The pre-processed cutting state data is cutting state data obtained after the pre-processing operation in S1, the quality of which is improved, the characteristics are more obvious, and it is more suitable for input to the subsequent model for analysis; the trained cutting state diagnosis model is a model obtained by training a machine learning or deep learning model on a large number of sample data of known cutting states; the current tool wear level is a classification of the current tool wear degree according to a certain wear standard and classification method, and different wear levels correspond to different performance states and service lives of the tool; the processing risk assessment result is a conclusion obtained by assessing the risks that may occur in the machining process of the numerical control machine tool, usually in the form of risk level or risk probability, helping the operator to understand the safety and stability of the current machining process; S3, based on the current tool wear level, the processing risk assessment result and the preset tool life optimization target, a cutting parameter control strategy is generated through an adaptive control algorithm, the cutting parameters including feed speed, cutting depth and spindle speed; The tool life optimization target is a target preset according to production demand and cost consideration; the adaptive control algorithm is a control algorithm that can automatically adjust the control parameters according to the real-time changes of system input and output to make the system achieve optimal performance, in this method, the adaptive control algorithm dynamically generates a cutting parameter control strategy according to the current tool wear level, the processing risk assessment result and the preset tool life optimization target, to adapt to the changing cutting state; the cutting parameters are adjustable parameters that affect the cutting process and machining quality in the cutting process, mainly including: feed speed, cutting depth, spindle speed; the cutting parameter control strategy is generated according to the adaptive control algorithm, and is a specific scheme for adjusting the cutting parameters such as feed speed, cutting depth and spindle speed; S4, the cutting parameter control strategy is sent to the numerical control machine tool actuator to complete the dynamic control of the tool cutting process; The cutting parameter control strategy is the same as defined in S3, which is applied to actual machining here; the numerical control machine tool actuator is a component responsible for executing specific actions in the numerical control machine tool, including feed servo system, spindle drive system, etc.; the dynamic control of the tool cutting process is to change the tool cutting behavior in real time by adjusting the cutting parameters according to the real-time cutting state and the preset target during the machining process, to adapt to different machining conditions and requirements, and to ensure the stability of the machining process and the machining quality; S5, real-time acquisition of cutting state feedback data of the regulated numerical control machine tool, iterative optimization of the cutting parameter control strategy, until the tool life optimization target is met or the machining task is completed; The cutting state feedback data of the regulated numerical control machine tool is the cutting state data of the numerical control machine tool collected again after the cutting parameter regulation strategy is executed, which reflects the actual changes of the cutting process after the regulation strategy is implemented. By comparing and analyzing the data before and after the regulation, the effect of the regulation strategy can be evaluated. The iterative optimization is a process of repeatedly adjusting and improving the cutting parameter regulation strategy according to the real-time collected cutting state feedback data. By continuously analyzing the gap between the feedback data and the tool life optimization target, the parameters or strategies in the adaptive control algorithm are adjusted, so that the cutting parameter regulation strategy is more reasonable and effective, and gradually approaches the optimal machining state. The tool life optimization target is the same as the definition in S3. The final purpose of the iterative optimization is to meet the target or complete the machining task. The machining task is considered to be completed when all workpieces are machined according to the predetermined machining requirements, or the machining is stopped due to other reasons. Before the machining task is completed, the cutting parameter regulation strategy is iteratively optimized to optimize the tool life as much as possible.

[0023] It should be noted that during operation, the original data is acquired and preprocessed to remove noise and other interference and improve data quality to lay a foundation for accurate analysis. The trained diagnostic model is used to output the tool wear grade and processing risk assessment to timely grasp the tool state and prevent potential problems in advance. According to the information from multiple aspects, the adaptive algorithm is used to generate a cutting parameter regulation strategy, and the feed rate and other parameters are dynamically adjusted according to the actual situation to realize personalized and accurate processing, which not only guarantees the processing quality but also prolongs the tool life. The strategy is issued to the actuator to complete dynamic regulation, which can respond to changes in cutting state in real time, collect feedback data to iteratively optimize the strategy, form a closed-loop control, and continuously approach the optimal machining state until the target is met or the task is completed, effectively improving processing efficiency, reducing cost, and enhancing production stability and reliability.

[0024] In an embodiment, the cutting state original data in the machining process of the numerical control machine tool is acquired and preprocessed, including: The cutting state original data is collected by a multi-sensor array, and the multi-sensor array includes a cutting force sensor, a vibration sensor, a temperature sensor, and an acoustic emission sensor. The collected original data constitutes a multi-dimensional time series data set , , wherein represents the dimension of the sensor data, and , represents the data collection time step; The multi-dimensional time series data set is preprocessed, and the preprocessing steps include: The abnormal values are removed; Data smoothing: the data after removing the abnormal values is smoothed by sliding average filtering, and the sliding window length is set to , and The calculation formula is: in Indicates the first 3D sensor in the first Smoothed data at each time step; Normalization process: Map the smoothed data to The normalization formula for the interval is: in , They represent the first Minimum and maximum values ​​of the data after smoothing by the dimensional sensor. This is the normalized cutting state data.

[0025] This design uses a multi-sensor array to collect raw cutting data, forming a multi-dimensional time-series dataset that comprehensively captures cutting process information. During preprocessing, outliers are removed to avoid interference; moving average filtering smooths the data, reducing random fluctuations; and normalization maps the data to the [0,1] interval, eliminating dimensional influences, improving data quality, and providing accurate and unified input for subsequent models. This makes diagnostic results more reliable, lays the foundation for precise control of tool cutting parameters, and helps improve machining efficiency and tool life.

[0026] In one embodiment, the preprocessed cutting state data is input into the trained cutting state diagnostic model, and the current tool wear level and machining risk assessment results are output, including: A deep learning-based cutting condition diagnostic model is constructed, which includes a feature extraction layer, a wear level classification layer, and a risk assessment layer. Feature extraction layer: A bidirectional long short-term memory network (Bi-LSTM) is used to extract temporal features from the preprocessed cutting state data, outputting a high-dimensional feature vector. , The feature dimension is calculated using the following formula: ; Wear level classification layer: High-dimensional feature vectors are processed through a fully connected network. Classify and output the current wear level of the tool. ,in These correspond to no wear, slight wear, moderate wear, and severe wear, respectively. The classification function is: in , For classification layer parameters; Risk assessment layer: based on wear level and real-time cutting parameters, the processing risk value is output by a risk regression model , The larger the value is, the higher the risk is, and the regression formula is: wherein is a current cutting parameter vector, represents a splicing of a feature vector and a parameter vector, , is a regression layer parameter, is a Sigmoid activation function; The cutting state diagnosis model is trained by using a historical cutting data set, and a weighted sum of a cross-entropy loss (wear grade classification) and a mean square error loss (risk assessment) is taken as an objective function, and the weight coefficients are respectively set as = 0.7, = 0.3, and the objective function formula is: wherein , are a predicted wear grade and an actual wear grade respectively, , are a predicted risk value and an actual risk value respectively; The preprocessed cutting state data is input into the trained model, and a current wear grade of the tool and a processing risk assessment result are output.

[0027] In this way, the diagnosis model based on deep learning is designed, the Bi-LSTM feature extraction layer can capture time sequence features, the wear grade classification layer can accurately classify the tool wear grade through a full connection network and a Softmax function, the risk assessment layer can output a risk value by using a risk regression model in combination with the wear grade and the cutting parameters, the model is trained by taking a weighted sum of a cross-entropy and a mean square error loss as an objective function, the classification accuracy and the risk assessment accuracy are considered, the tool state can be diagnosed in real time and accurately, and the processing risk can be assessed, which provides a key basis for subsequent adaptive control, and guarantees the stability and safety of the processing process.

[0028] In an embodiment, based on the current wear grade of the tool, the processing risk assessment result and a preset tool life optimization target, a cutting parameter control strategy is generated by an adaptive control algorithm, including: The tool life optimization target is set, and the target includes: When (no wear / slight wear), the processing efficiency is preferentially ensured, and the tool life extension rate target ; When When the tool is in the state of moderate / serious wear, the priority is to reduce the wear rate, and the tool life extension rate target ; An adaptive control objective function is constructed to cut parameters , wherein is the spindle speed, is the feed speed, is the cutting depth, and the optimization variable is , wherein is the machining efficiency index, and the calculation formula is , which is normalized to ; is the tool life extension index, which is calculated based on the historical wear law model, and the formula is , is the wear rate under the parameter , and is the current cutting parameter; , is the weight coefficient, which is dynamically adjusted according to the wear grade: When , =0.6, =0.4; When , =0.3, =0.7; Set the cutting parameter constraint condition: , , , wherein , , etc. are physical limit parameters of numerical control machine tools and tools; The particle swarm optimization algorithm is used to solve the adaptive control objective function, and the optimal cutting parameter is output, forming a cutting parameter regulation strategy.

[0029] In this way, the tool life optimization target under different wear grades is designed to ensure machining efficiency or reduce wear rate, the adaptive control objective function is constructed, the machining efficiency and tool life extension index are considered comprehensively, the weight coefficient is dynamically adjusted according to the wear grade, the cutting parameter constraint condition is set, the particle swarm optimization algorithm is used to solve, the optimal cutting parameter is output, and the cutting parameter is automatically adjusted according to the real-time state of the tool. While ensuring the machining quality, the tool life is maximally extended, the production efficiency and economic benefit are improved, and the adaptability and robustness of the system are enhanced.

[0030] In an embodiment, the cutting parameter regulation strategy is issued to the numerical control machine tool actuator to complete the dynamic regulation of the tool cutting process, including: A cutting parameter conversion interface is constructed to convert the optimal cutting parameters obtained by optimization into G code instructions recognizable by the numerical control machine tool, and the conversion rule follows the ISO6983 standard. The G code instructions are issued to the numerical control system of the numerical control machine tool through industrial Ethernet. After receiving the instructions, the numerical control system drives the servo motor, spindle driver and other actuators to adjust the feed axis speed, spindle speed and tool cutting depth to realize real-time update of the cutting parameters. A regulation response threshold is set: When the parameter adjustment amplitude , the regulation is delayed until the start of the next machining process; When , the regulation is immediately executed and the machining is temporarily paused for 0.5-1s to avoid machining errors caused by parameter mutations.

[0031] In this way, the optimal parameters are converted into G code instructions, which are easy to recognize by the machine tool and are issued through industrial Ethernet. The numerical control system drives the actuators to update the cutting parameters in real time. The regulation response threshold is set to avoid frequent regulation by delaying the execution of small amplitude adjustments. Large amplitude adjustments are immediately executed and the machining is temporarily paused to prevent errors caused by parameter mutations, ensuring accurate and stable adjustment of the cutting parameters and smooth dynamic regulation of the tool cutting process, improving machining precision and product quality, and reducing machining problems caused by parameter mutations.

[0032] In an embodiment, the cutting state feedback data of the numerical control machine tool after regulation is collected in real time, and the cutting parameter regulation strategy is iteratively optimized, including: An iterative regulation period is set , and In each period, the cutting state data after regulation is collected in real time, and the steps S2-S3 are repeated to obtain new wear grades and risk values ; An evaluation index of the regulation effect is constructed: Wear improvement rate: If , it is positive improvement; Risk reduction rate: If , it is positive reduction; When and , it is determined that the current regulation strategy is effective, and ​Continue to execute; When or , trigger secondary optimization, re-execute the adaptive control algorithm of S3 step, adjust the weight coefficient 、 , specifically, down-regulate , up-regulate , generate a new cutting parameter regulation strategy; Repeat the above iteration process until the machining task is completed or the tool wear level reaches the severe wear threshold , trigger the tool replacement reminder.

[0033] In this way, the iteration control period is set, the feedback data is collected in real time, the diagnosis and strategy generation steps are repeated, the new wear level and risk value are obtained, the control effect evaluation index is constructed, the strategy effectiveness is judged through the wear improvement rate and the risk reduction rate, and the strategy is maintained if it is effective, otherwise the weight coefficient is adjusted to generate a new strategy, and the iteration is repeated until the task is completed or the tool is severely worn. According to the real-time feedback, the cutting parameters are continuously optimized, so that the regulation strategy always adapts to the change of cutting state, improves the stability and controllability of the machining process, and realizes the optimal balance of tool life and machining efficiency.

[0034] Example 2: Please refer to Figure 3 , a numerical control machine tool tool regulation system based on cutting state diagnosis, applied to the numerical control machine tool tool regulation method based on cutting state diagnosis as described above, the system comprises the following modules: Data acquisition module, data preprocessing module, cutting state diagnosis module, regulation strategy generation module, regulation execution module and iteration optimization module; The data acquisition module is used to acquire the cutting state original data and the cutting state feedback data after regulation in the machining process of the numerical control machine tool; The data preprocessing module is used to preprocess the cutting state original data and the cutting state feedback data; The cutting state diagnosis module is used to input the preprocessed cutting state data into the trained cutting state diagnosis model, and output the current tool wear level and the machining risk evaluation result; The regulation strategy generation module is used to generate a cutting parameter regulation strategy through an adaptive control algorithm based on the current tool wear level, the machining risk evaluation result and the preset tool life optimization target; The regulation execution module is used to issue the cutting parameter regulation strategy to the numerical control machine tool execution mechanism to complete the dynamic regulation of the tool cutting process; The iteration optimization module is used to iteratively optimize the cutting parameter regulation strategy based on the cutting state feedback data until the tool life optimization target is met or the machining task is completed.

[0035] Those skilled in the art can understand that all or part of the steps in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through programs, and therefore the present application can adopt a completely hardware embodiment, a completely software embodiment, or an embodiment in the form of a combination of software and hardware aspects. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer usable program codes.

[0036] The above embodiments have been described in detail, and the principles and embodiments of the present application have been described by applying specific examples. The above embodiment descriptions are only used to help understand the method of the present application and its core idea; meanwhile, for those skilled in the art, according to the idea of the present application, the specific embodiments and application scope will have changes, and the above description should not be understood as a limitation on the present application.

Claims

1. A CNC machine tool tool control method based on cutting condition diagnosis, characterized in that, Includes the following steps: S1. Acquire raw data of the cutting state during the CNC machine tool machining process and perform preprocessing; S2. Input the preprocessed cutting state data into the trained cutting state diagnostic model, and output the current tool wear level and machining risk assessment results; S3. Based on the current wear level of the tool, the machining risk assessment results, and the preset tool life optimization target, an adaptive control algorithm is used to generate a cutting parameter control strategy. The cutting parameters include at least one or more of the following: feed rate, depth of cut, and spindle speed. S4. The cutting parameter control strategy is sent to the CNC machine tool actuator to complete the dynamic control of the tool cutting process; S5. Real-time acquisition of cutting status feedback data of the CNC machine tool after adjustment, and iterative optimization of the cutting parameter adjustment strategy until the tool life optimization target is met or the machining task is completed.

2. The CNC machine tool control method based on cutting condition diagnosis according to claim 1, characterized in that, Step S1 involves acquiring and preprocessing the raw cutting state data during the CNC machine tool machining process, including: Raw cutting state data is collected by a multi-sensor array, which includes at least one or more of cutting force sensors, vibration sensors, temperature sensors, and acoustic emission sensors. The collected raw data constitutes a multi-dimensional time-series dataset. The multi-dimensional time-series dataset is preprocessed, and the preprocessing steps include at least one or more of outlier removal, data smoothing, and normalization.

3. The CNC machine tool control method based on cutting condition diagnosis according to claim 2, characterized in that: The data smoothing process uses a filtering algorithm to smooth the data after removing outliers. The filtering algorithm includes a moving average filter, and the moving average filter has a sliding window length of 5-20. The normalization process maps the smoothed data to a preset numerical range, which includes the [0,1] range.

4. The CNC machine tool control method based on cutting condition diagnosis according to claim 1, characterized in that, In step S2: The cutting condition diagnostic model is a deep learning-based model, which includes at least one or more of the following: a feature extraction layer, a wear level classification layer, and a risk assessment layer. The feature extraction layer uses a temporal feature extraction network to extract temporal features from the preprocessed cutting state data and outputs a high-dimensional feature vector. The temporal feature extraction network includes a bidirectional long short-term memory network. The wear level classification layer classifies high-dimensional feature vectors through a classification network and outputs the current wear level of the tool. The classification network includes a fully connected network, and the wear level includes at least one or more of the following: no wear, slight wear, moderate wear, and severe wear. The risk assessment layer outputs machining risk values ​​based on wear level and real-time cutting parameters through a regression model. The regression model includes a risk regression model, and the numerical range of the machining risk value includes [0,1]. The larger the value, the higher the risk.

5. The CNC machine tool control method based on cutting condition diagnosis according to claim 4, characterized in that: The cutting condition diagnostic model is trained using a historical cutting dataset. The training process uses a weighted sum of multiple loss functions as the objective function. The multiple loss functions include at least one or more of cross-entropy loss and mean square error loss. The cross-entropy loss corresponds to wear level classification, and the mean square error loss corresponds to risk assessment.

6. The CNC machine tool control method based on cutting condition diagnosis according to claim 1, characterized in that, The preset tool life optimization target in step S3 is set based on the current tool wear level: When the current wear level of the tool is no wear or slight wear, prioritize machining efficiency and set a tool life extension rate target. When the current wear level of the tool is moderate or severe, prioritize reducing the wear rate and set a target for the remaining tool life extension rate.

7. The CNC machine tool control method based on cutting condition diagnosis according to claim 6, characterized in that: When constructing the adaptive control objective function, the cutting parameters are used as optimization variables. The adaptive control objective function is a weighted sum of the machining efficiency index and the tool life extension index. The machining efficiency index is calculated based on the cutting parameters and normalized to a preset range. The tool life extension index is calculated based on a historical wear pattern model. The weighting coefficients of machining efficiency index and tool life extension index are dynamically adjusted according to the current wear level of the tool, and cutting parameter constraints are set based on the physical limit parameters of CNC machine tool and tool.

8. The CNC machine tool control method based on cutting condition diagnosis according to claim 7, characterized in that: The adaptive control objective function is solved using an optimization algorithm, and the optimal cutting parameters are output to form a cutting parameter control strategy. The optimization algorithm includes particle swarm optimization.

9. The CNC machine tool control method based on cutting condition diagnosis according to claim 1, characterized in that: In step S5, an iterative control cycle is set, and cutting state data after control is collected in real time in each cycle. The operation of step S2 is repeated to obtain a new wear level and risk value. Construct an evaluation index for the control effect, wherein the evaluation index for the control effect includes at least one or more of wear improvement rate and risk reduction rate; The effectiveness of the current control strategy is determined based on the control effect evaluation index. If it is ineffective, a secondary optimization is triggered, and the adaptive control algorithm is re-executed to adjust the weight coefficients and generate a new cutting parameter control strategy. The process repeats until the machining task is completed or the tool wear level reaches the preset severe wear threshold, at which point a tool replacement reminder is triggered.

10. A CNC machine tool tool control system based on cutting condition diagnosis, characterized in that, The system, applied to the CNC machine tool tool control method based on cutting condition diagnosis as described in any one of claims 1-9, comprises the following modules: The system includes a data acquisition module, a data preprocessing module, a cutting condition diagnosis module, a control strategy generation module, a control execution module, and an iterative optimization module. The data acquisition module is used to acquire raw data of the cutting state and feedback data of the adjusted cutting state during the CNC machine tool machining process; The data preprocessing module is used to preprocess the raw cutting state data and the cutting state feedback data; The cutting condition diagnosis module is used to input the preprocessed cutting condition data into the trained cutting condition diagnosis model and output the current tool wear level and machining risk assessment results. The control strategy generation module is used to generate a cutting parameter control strategy based on the current tool wear level, machining risk assessment results and preset tool life optimization targets, through an adaptive control algorithm. The control and execution module is used to send the cutting parameter control strategy to the CNC machine tool execution mechanism to complete the dynamic control of the tool cutting process; The iterative optimization module is used to iteratively optimize the cutting parameter control strategy based on cutting state feedback data until the tool life optimization target is met or the machining task is completed.

Images