Numerical control machining optimization control system and method based on industrial big data processing

Abstract

This invention discloses a CNC machining optimization control system and method based on industrial big data processing, belonging to the field of CNC machining and intelligent control technology. The system includes units for multi-source data acquisition from the industrial site, industrial big data preprocessing and feature extraction, a CNC machining full-process database, intelligent identification and anomaly warning of machining status, intelligent optimization of process parameters, CNC equipment instruction generation and issuance, and closed-loop verification and model iteration of machining effects. These units collaborate to complete the intelligent monitoring and optimization control of the entire CNC machining process. This invention achieves comprehensive monitoring and anomaly warning of the machining process, optimizes process parameters to improve machining quality, realizes multi-equipment collaborative scheduling and full lifecycle management of tools, possesses closed-loop iterative optimization capabilities, and improves production efficiency and system stability.

Description

Technical Field

[0001] This invention relates to the field of CNC machining and intelligent control technology, and in particular to a CNC machining optimization control system and method based on industrial big data processing. Background Technology

[0002] CNC machining is a core technology in modern manufacturing, and its machining accuracy and production efficiency directly affect product quality and manufacturing cycle. Traditional CNC machining systems mostly use fixed process parameters for production. The setting of these parameters relies heavily on the operator's experience and is difficult to dynamically adjust based on real-time conditions during machining. When tools wear out, equipment conditions change, or the environment fluctuates, fixed parameters can easily lead to problems such as decreased machining accuracy, substandard surface quality, or even tool breakage. Furthermore, traditional systems lack comprehensive monitoring of the machining process, and the detection of abnormal events mainly relies on manual inspection, which has a significant lag and cannot promptly address sudden anomalies during machining, easily causing batch scrap and production interruptions.

[0003] With the development of Industrial Internet technology, some CNC machining systems have begun to incorporate industrial big data technology. However, the data acquisition capabilities of existing systems are significantly insufficient. Most systems only collect basic operating parameters of the CNC system, lacking the collection of key status parameters such as equipment vibration, temperature, and hydraulic pressure, as well as station-level environmental parameters. Furthermore, the time synchronization of data from different sources is poor, making it difficult to comprehensively reflect the true state of the machining process. Existing systems have limited data preprocessing and feature extraction capabilities, failing to effectively eliminate noise interference in the industrial environment and struggling to extract effective features related to machining quality from massive amounts of data, resulting in low accuracy in subsequent status identification and parameter optimization. Simultaneously, most systems employ offline optimization methods, failing to achieve online adaptive adjustment of process parameters and making it difficult to adapt to dynamic changes during the machining process.

[0004] Most existing CNC machining systems operate in a stand-alone mode, lacking the ability to coordinate and schedule multiple devices. They cannot dynamically allocate tasks based on the overall production status of the workshop, easily leading to situations where some machines are overloaded while others are idle, impacting overall workshop efficiency. Regarding tool management, existing systems often use fixed tool change cycles, failing to intelligently recommend tool change times based on actual tool wear. This easily results in tool overuse or premature replacement, affecting machining quality and tool utilization. Furthermore, existing systems lack a complete closed-loop optimization mechanism, unable to continuously optimize the system model based on machining feedback. As equipment ages and production conditions change, system performance gradually declines. Summary of the Invention

[0005] The present invention proposes a numerical control machining optimization control system and method based on industrial big data processing to solve the problems mentioned in the prior art.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a CNC machining optimization control system based on industrial big data processing, comprising: The system includes: an industrial site multi-source data acquisition unit; an industrial big data preprocessing and feature extraction unit that communicates bidirectionally with the industrial site multi-source data acquisition unit; a CNC machining full-process database that communicates bidirectionally with the industrial big data preprocessing and feature extraction unit; a machining status intelligent identification and anomaly early warning unit that communicates bidirectionally with the CNC machining full-process database; a process parameter intelligent optimization unit that communicates bidirectionally with the CNC machining full-process database and the machining status intelligent identification and anomaly early warning unit; a CNC equipment instruction generation and issuance unit that communicates bidirectionally with the process parameter intelligent optimization unit; and a machining effect closed-loop verification and model iteration unit that communicates bidirectionally with the CNC equipment instruction generation and issuance unit. Industrial site multi-source data acquisition unit: completes real-time acquisition of equipment operation data, process parameter data, processing quality data, and environmental status data during CNC machining; Industrial Big Data Preprocessing and Feature Extraction Unit: Completes the cleaning, standardization and multi-dimensional feature extraction of collected data; CNC machining full-process database completes the time-series storage and unified management of all machining data. Intelligent processing status recognition and anomaly early warning unit: Completes processing status recognition and abnormal event detection and early warning; Intelligent process parameter optimization unit: generates the optimal combination of process parameters based on historical data and real-time status; The CNC equipment instruction generation and issuance unit completes the conversion and issuance of CNC machining instructions. The machining effect closed-loop verification and model iteration unit completes the machining effect verification and continuous iterative optimization of the system model.

[0007] Furthermore, it also includes a multi-device collaborative scheduling unit. This unit has bidirectional communication connections with the CNC machining full-process database, the process parameter intelligent optimization unit, and the CNC equipment instruction generation and issuance unit. It constructs a digital twin model of workshop production to map the physical production status of the workshop in real time. Based on the model, it completes the simulation pre-scheduling of processing tasks, identifies production bottlenecks and equipment conflicts in advance, dynamically adjusts task priorities according to order delivery date, processing difficulty, and urgency, and completes the dynamic allocation of processing tasks by combining equipment health status and processing capacity. This achieves load balancing and cross-process collaborative scheduling among multiple devices.

[0008] Furthermore, it also includes a tool lifecycle management unit. This unit has bidirectional communication connections with the industrial site multi-source data acquisition unit, the CNC machining full-process database, and the process parameter intelligent optimization unit. It integrates spindle vibration, cutting power, motor current, and visual inspection data to complete multi-dimensional monitoring of tool wear status, establishes a tool wear prediction model to predict the remaining tool life, automatically generates a tool replacement reminder and schedules idle operators when the remaining tool life reaches a preset threshold, establishes a tool lifecycle health record to record the entire process information of tool procurement, use, wear, and tool replacement, and automatically fine-tunes the cutting parameters of the corresponding machining task based on the real-time wear degree of the tool.

[0009] Furthermore, the multi-source data acquisition unit in the industrial field includes an equipment operation data acquisition subunit, a process parameter acquisition subunit, a processing quality inspection subunit, and an environmental status acquisition subunit. The equipment operation data acquisition subunit simultaneously collects equipment operating parameters such as spindle temperature, guide rail displacement, and hydraulic pressure through a combination of the CNC system's native interface and edge-side sensors. The process parameter acquisition subunit collects the CNC system's command parameters and actual execution parameters in real time, comparing and analyzing the deviation between the command and execution. The processing quality inspection subunit integrates data from online visual inspection equipment, laser displacement sensors, and coordinate measuring machines. The environmental status acquisition subunit adopts a distributed deployment approach, deploying environmental sensors at each processing station to collect station-level temperature, humidity, and vibration parameters. All acquisition units achieve time synchronization through a global time synchronization protocol.

[0010] Furthermore, the industrial big data preprocessing and feature extraction unit incorporates a data preprocessing submodule and a multi-scale feature extraction submodule. The data preprocessing submodule uses a sliding window filtering algorithm to eliminate high-frequency noise for time-series data, an isolated forest algorithm to identify and remove outlier data, and linear interpolation to fill in missing data, while also normalizing all data. The multi-scale feature extraction submodule uses wavelet packet transform to extract energy features from different frequency bands, employs statistical analysis methods to extract mean, variance, peak value, and kurtosis time-domain features, uses mutual information to calculate the correlation between each feature and processing quality, ranks features based on correlation, and uses principal component analysis to reduce feature dimensionality, generating a low-dimensional, highly discriminative feature subset. For different processing stages—roughing, semi-finishing, and finishing—an adaptive feature extraction strategy is used to extract the most sensitive features for the corresponding stage.

[0011] Furthermore, the intelligent machining status recognition and anomaly early warning unit is pre-trained based on labeled CNC machining historical datasets. It includes a built-in machining status classification submodule and an anomaly event detection submodule. The machining status classification submodule classifies and identifies different machining states, including normal machining, tool wear, tool breakage, chatter, and overload. The anomaly event detection submodule detects and locates various anomalies by comparing real-time collected data with a normal machining status benchmark. It then generates an anomaly severity score using a machining anomaly severity quantification model. The calculation formula is as follows: ;in, To score the severity of processing abnormalities, This represents the total number of anomaly detection features, with a dimension of 1. For the first The weight coefficients of each anomaly detection feature. For the first Real-time measurement values ​​of each feature For the first Normal processing baseline values ​​for each feature For the first The difference in the normal processing range of each feature.

[0012] Furthermore, the intelligent optimization unit for process parameters incorporates a process parameter knowledge base, a multi-objective optimization submodule, and a parameter verification submodule. The process parameter knowledge base stores historically optimal combinations of process parameters for different materials, equipment, and processing requirements, and uses a dynamic update mechanism to automatically store effective parameter combinations after each processing task is completed. The multi-objective optimization submodule employs an improved non-dominated sorting genetic algorithm combined with deep reinforcement learning, using processing accuracy, surface quality, processing efficiency, and tool life as optimization objectives to generate a Pareto optimal solution set, and selects the optimal combination of process parameters based on user needs. The parameter verification submodule, based on a CNC machining digital twin model, performs virtual machining simulation of the optimized parameters, simulating cutting forces, cutting temperatures, and workpiece deformation during the machining process, predicting machining quality and tool wear, and automatically re-optimizing if the simulation results do not meet the requirements. An online adaptive adjustment mechanism for process parameters is also added, dynamically fine-tuning parameters based on real-time data during the machining process.

[0013] Furthermore, this includes the following steps: The system uses a multi-source data acquisition unit in the industrial field to collect real-time data on equipment operation, process parameters, machining quality, and environmental conditions during the CNC machining process. The industrial big data preprocessing and feature extraction unit completes the preprocessing and multi-dimensional feature extraction of the collected data, and stores the processed data into the CNC machining full-process database. The intelligent processing status identification and anomaly early warning unit completes the processing status identification and abnormal event detection and early warning. When an abnormal event is detected, the corresponding level of early warning information is issued and the emergency response procedure is initiated. Based on historical processing data and real-time processing status, the intelligent process parameter optimization unit generates the optimal combination of process parameters that meets multiple objective constraints. The optimized process parameters are converted into machining instructions that can be recognized by the CNC equipment through the CNC equipment instruction generation and issuance unit, and then issued to the corresponding CNC equipment for execution; The processing effect is verified in real time through closed-loop verification of processing effect and model iteration unit. Based on the verification results, the incremental training and parameter update of the system model are completed.

[0014] Furthermore, a digital twin model of workshop production is constructed through a multi-device collaborative scheduling unit. Based on the model, simulation pre-scheduling of processing tasks is completed, production bottlenecks and equipment conflicts are identified in advance, and task priorities are dynamically adjusted according to order delivery date, processing difficulty, and urgency. The dynamic allocation of processing tasks is completed by combining equipment health status and processing capacity.

[0015] Furthermore, by integrating multi-source sensor data through the tool lifecycle management unit, real-time monitoring of tool wear status is achieved, a tool wear prediction model is established to predict the remaining tool life, and when the remaining tool life reaches a preset threshold, a tool replacement reminder is automatically generated and idle operators are scheduled. At the same time, non-urgent machining tasks on the corresponding equipment are suspended. After the tool replacement is completed, tool setting and parameter calibration are automatically performed, and the tool lifecycle health record is updated.

[0016] Compared with existing technologies, the beneficial effects of this invention are: This invention utilizes a multi-source data acquisition unit in an industrial setting to comprehensively collect data on equipment operation, process parameters, machining quality, and environmental conditions during CNC machining. A global time synchronization protocol ensures time consistency across different data sources, providing a high-quality data foundation for comprehensive analysis of the machining process. The industrial big data preprocessing and feature extraction unit effectively eliminates noise interference in the industrial environment, extracts effective features at multiple scales, and employs adaptive feature extraction strategies based on different machining stages to improve feature recognition and relevance.

[0017] This invention utilizes an intelligent processing status identification and anomaly early warning unit to accurately identify various processing states and abnormal events. Through multi-feature weighted fusion, it achieves precise quantification of anomaly severity, generates handling suggestions based on an anomaly event knowledge base, and disseminates tiered early warning information through multiple channels, improving the response speed and handling efficiency of anomaly events. The intelligent process parameter optimization unit combines multi-objective optimization algorithms and digital twin simulation technology to generate optimal process parameter combinations that satisfy multi-objective constraints. It also dynamically fine-tunes parameters through an online adaptive adjustment mechanism to maintain the stability of the processing.

[0018] This invention utilizes a multi-device collaborative scheduling unit to construct a digital twin model of workshop production, achieving real-time mapping of production status. Through simulation-based pre-scheduling, production bottlenecks are identified in advance. Dynamic task allocation is achieved by combining task priority and equipment status, realizing load balancing across multiple devices and cross-process collaborative scheduling, thereby improving overall workshop production efficiency. The tool lifecycle management unit integrates multi-source data to achieve real-time monitoring of tool wear status and prediction of remaining life. Based on the actual wear level, it intelligently recommends tool replacement timing and compensates for machining errors by fine-tuning cutting parameters, improving tool utilization and machining quality.

[0019] This invention establishes a complete closed-loop verification and model iteration mechanism for processing effects. It can incrementally train and update the system model based on real-time feedback of processing effects, enabling the system to continuously adapt to equipment aging and changes in production conditions, maintaining stable operating performance over the long term. The CNC machining full-process database enables time-series storage and unified management of all processing data, providing data support for continuous system optimization and complete data evidence for production process traceability and analysis. Attached Figure Description

[0020] Figure 1 This is a schematic block diagram illustrating the overall architecture and business process of the CNC machining optimization control proposed in this invention; Figure 2 This is a schematic block diagram of the industrial field multi-source data acquisition and feature extraction flowchart proposed in this invention; Figure 3 This is the intelligent identification and hierarchical early warning decision diagram of the processing status proposed in this invention; Figure 4 This is a flowchart of the intelligent multi-objective optimization and virtual processing verification of process parameters proposed in this invention; Figure 5 This is a logic diagram of cross-process multi-equipment collaborative scheduling and tool lifecycle management proposed in this invention. Detailed Implementation

[0021] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] Reference Figures 1 to 5 A numerical control machining optimization control system based on industrial big data processing, comprising: The system includes: an industrial site multi-source data acquisition unit; an industrial big data preprocessing and feature extraction unit that communicates bidirectionally with the industrial site multi-source data acquisition unit; a CNC machining full-process database that communicates bidirectionally with the industrial big data preprocessing and feature extraction unit; a machining status intelligent identification and anomaly early warning unit that communicates bidirectionally with the CNC machining full-process database; a process parameter intelligent optimization unit that communicates bidirectionally with the CNC machining full-process database and the machining status intelligent identification and anomaly early warning unit; a CNC equipment instruction generation and issuance unit that communicates bidirectionally with the process parameter intelligent optimization unit; and a machining effect closed-loop verification and model iteration unit that communicates bidirectionally with the CNC equipment instruction generation and issuance unit. Industrial site multi-source data acquisition unit: completes real-time acquisition of equipment operation data, process parameter data, processing quality data, and environmental status data during CNC machining; Industrial Big Data Preprocessing and Feature Extraction Unit: Completes the cleaning, standardization and multi-dimensional feature extraction of collected data; CNC machining full-process database completes the time-series storage and unified management of all machining data. Intelligent processing status recognition and anomaly early warning unit: Completes processing status recognition and abnormal event detection and early warning; Intelligent process parameter optimization unit: generates the optimal combination of process parameters based on historical data and real-time status; The CNC equipment instruction generation and issuance unit completes the conversion and issuance of CNC machining instructions. The machining effect closed-loop verification and model iteration unit completes the machining effect verification and continuous iterative optimization of the system model.

[0023] This invention also includes a multi-device collaborative scheduling unit, which is bidirectionally connected to the CNC machining full-process database, the process parameter intelligent optimization unit, and the CNC equipment instruction generation and issuance unit. This unit constructs a digital twin model of workshop production, mapping the physical production status of the workshop in real time. Based on the model, it performs simulation pre-scheduling of processing tasks, identifies production bottlenecks and equipment conflicts in advance, dynamically adjusts task priorities according to order delivery dates, processing difficulty, and urgency, and dynamically allocates processing tasks based on equipment health status and processing capacity. This achieves load balancing and cross-process collaborative scheduling among multiple devices. When a device malfunctions or processing is delayed, unfinished tasks are automatically transferred and subsequent production plans are adjusted. The collaborative material delivery system ensures timely material delivery, achieving seamless integration of the overall production process.

[0024] This invention also includes a tool lifecycle management unit. This unit has bidirectional communication with a multi-source data acquisition unit in the industrial field, a CNC machining full-process database, and a process parameter intelligent optimization unit. It integrates spindle vibration, cutting power, motor current, and visual inspection data to achieve multi-dimensional monitoring of tool wear status, establishes a tool wear prediction model to predict the remaining tool life, automatically generates a tool replacement reminder and schedules idle operators when the remaining tool life reaches a preset threshold, establishes a tool lifecycle health record to record the entire process information of tool procurement, use, wear, and tool replacement, and automatically fine-tunes the cutting parameters of the corresponding machining task according to the real-time wear degree of the tool to compensate for machining errors, and realizes intelligent recommendation of tool replacement timing and full-process control of tool use.

[0025] In this invention, the industrial site multi-source data acquisition unit includes an equipment operation data acquisition subunit, a process parameter acquisition subunit, a processing quality inspection subunit, and an environmental status acquisition subunit. The equipment operation data acquisition subunit achieves millisecond-level high-frequency data acquisition by combining the native interface of the CNC system with edge-side sensors, simultaneously acquiring equipment operating parameters such as spindle temperature, guide rail displacement, and hydraulic pressure. The process parameter acquisition subunit acquires the command parameters and actual execution parameters of the CNC system in real time, comparing and analyzing the deviation between the command and execution. The processing quality inspection subunit integrates data from online visual inspection equipment, laser displacement sensors, and coordinate measuring machines to achieve multi-dimensional detection of processing quality. The environmental status acquisition subunit adopts a distributed deployment method, deploying environmental sensors at each processing station to acquire station-level temperature, humidity, and vibration parameters. All acquisition units achieve time synchronization through a global time synchronization protocol, with the time consistency error of data from different sources not exceeding 1 millisecond.

[0026] In this invention, the industrial big data preprocessing and feature extraction unit incorporates a data preprocessing submodule and a multi-scale feature extraction submodule. The data preprocessing submodule uses a sliding window filtering algorithm to eliminate high-frequency noise for time-series data, an isolated forest algorithm to identify and remove outlier data, and linear interpolation to fill in missing data. Simultaneously, it performs min-max normalization on all data to eliminate differences in the dimensions of different features. The multi-scale feature extraction submodule uses wavelet packet transform to extract energy features from different frequency bands, employs statistical analysis methods to extract mean, variance, peak value, and kurtosis time-domain features, uses mutual information to calculate the correlation between each feature and processing quality, ranks features based on correlation, and uses principal component analysis to reduce feature dimensionality, generating a low-dimensional, highly discriminative feature subset. For different processing stages—roughing, semi-finishing, and finishing—an adaptive feature extraction strategy is used to extract the most sensitive features for the corresponding stage.

[0027] In this invention, the intelligent machining status recognition and anomaly early warning unit is pre-trained based on a labeled CNC machining historical dataset. It incorporates a machining status classification submodule and an anomaly event detection submodule. The machining status classification submodule classifies and identifies different machining states, including normal machining, tool wear, tool breakage, chatter, and overload. The anomaly event detection submodule detects and locates various anomalies by comparing real-time collected data with a normal machining status benchmark. A severity score for each anomaly is generated using a machining anomaly severity quantification model. Based on the score, graded early warning information is issued to the corresponding personnel. The calculation formula is as follows: ;in, The severity of the processing anomaly is scored, with a dimension of 1. The higher the score, the more severe the anomaly. This represents the total number of anomaly detection features, with a dimension of 1. For the first The weight coefficient of each anomaly detection feature, with a dimension of 1, is determined by the degree of influence of that feature on processing quality, and all The sum is 1. For the first Real-time measurements of each feature, with dimensions consistent with the corresponding feature. For the first Normal processing reference value for each feature, dimensions and Maintain consistency. For the first The difference in the normal processing range of each feature, with units of... To maintain consistency, the severity of processing anomalies is accurately quantified through multi-feature weighted fusion, providing a quantitative basis for graded early warning and emergency response. At the same time, an anomaly knowledge base is established to store the handling methods for various anomalies. Root cause analysis is completed based on anomaly characteristics to generate handling suggestions. Early warning information is released through multiple channels, including on-site audible and visual alarms and terminal message push.

[0028] In this invention, the intelligent optimization unit for process parameters incorporates a process parameter knowledge base, a multi-objective optimization submodule, and a parameter verification submodule. The process parameter knowledge base stores historically optimal combinations of process parameters for different materials, equipment, and processing requirements. A dynamic update mechanism automatically stores effective parameter combinations after each processing task is completed, and outdated data is periodically cleaned up. The multi-objective optimization submodule uses an improved non-dominated sorting genetic algorithm combined with deep reinforcement learning to generate Pareto optimal solution sets with processing accuracy, surface quality, processing efficiency, and tool life as optimization objectives. It then selects the optimal combination of process parameters based on user needs. The parameter verification submodule, based on a CNC machining digital twin model, performs virtual machining simulation of the optimized parameters, simulating cutting force, cutting temperature, and workpiece deformation during processing, predicting processing quality and tool wear. If the simulation results do not meet requirements, it automatically re-optimizes. An online adaptive adjustment mechanism for process parameters is also included, dynamically fine-tuning parameters based on real-time data during processing to maintain process stability.

[0029] This invention includes the following steps: The system uses a multi-source data acquisition unit in the industrial field to collect real-time data on equipment operation, process parameters, machining quality, and environmental conditions during the CNC machining process. The industrial big data preprocessing and feature extraction unit completes the preprocessing and multi-dimensional feature extraction of the collected data, and stores the processed data into the CNC machining full-process database. The intelligent processing status identification and anomaly early warning unit completes the processing status identification and abnormal event detection and early warning. When an abnormal event is detected, the corresponding level of early warning information is issued and the emergency response procedure is initiated. Based on historical processing data and real-time processing status, the intelligent process parameter optimization unit generates the optimal combination of process parameters that meets multiple objective constraints. The optimized process parameters are converted into machining instructions that can be recognized by the CNC equipment through the CNC equipment instruction generation and issuance unit, and then issued to the corresponding CNC equipment for execution; The processing effect is verified in real time through closed-loop verification of processing effect and model iteration unit. Based on the verification results, the incremental training and parameter update of the system model are completed.

[0030] In this invention, a digital twin model of workshop production is constructed through a multi-device collaborative scheduling unit. Based on the model, simulation pre-scheduling of processing tasks is completed, production bottlenecks and equipment conflicts are identified in advance, and task priorities are dynamically adjusted according to order delivery date, processing difficulty, and urgency. The dynamic allocation of processing tasks is completed by combining equipment health status and processing capacity. When a piece of equipment malfunctions or processing is delayed, the unfinished processing tasks are automatically transferred to other idle equipment. At the same time, the execution order of subsequent tasks is adjusted, and the material distribution system is coordinated to complete the timely delivery of materials, maintaining the smooth execution of the overall production plan.

[0031] In this invention, a tool lifecycle management unit integrates multi-source sensor data to achieve real-time monitoring of tool wear status, establishes a tool wear prediction model to predict the remaining tool life, automatically generates a tool change reminder and schedules idle operators when the remaining tool life reaches a preset threshold, and simultaneously suspends non-urgent machining tasks on the corresponding equipment. After tool change, tool setting and parameter calibration are automatically performed, the tool lifecycle health record is updated, and cutting parameters are automatically fine-tuned according to the real-time wear level of the tool to compensate for machining errors and maintain stable operation of the machining process.

[0032] The processing effect closed-loop verification and model iteration unit incorporates a multi-dimensional factorization module for processing effects, a model update triggering submodule, a model verification and rollback submodule, and a full-link feedback submodule. The multi-dimensional factorization module correlates real-time process parameters, equipment operating status, tool wear data, and final quality inspection data for the current processing, using correlation analysis to pinpoint whether deviations arise from process parameter settings, equipment status fluctuations, tool wear, or environmental interference. The model update triggering submodule supports both fixed batch update and abnormal trigger update modes. In addition to the preset fixed batch update cycle, when three or more consecutive non-conforming products occur or the abnormal event detection accuracy falls below 95%, an emergency model iteration is automatically triggered. The process flow is streamlined; the model verification and rollback submodule stipulates that all updated models must first complete accuracy verification on an offline test set containing no less than 1,000 historical labeled data. Only after successful verification can the model be deployed online in a gray-scale manner. If the pass rate of the new model in 20 consecutive processing batches decreases by more than 2% compared to the previous stable version, the system will automatically roll back to the previous verified stable model version. The end-to-end feedback submodule will synchronize the verification results back to the process parameter knowledge base, the intelligent identification and anomaly warning unit for processing status, and the tool life cycle management unit, updating the effect score of the corresponding parameter combination, the weight threshold of the anomaly detection feature, and the parameters of the tool wear prediction model, so as to achieve collaborative closed-loop optimization of all units in the entire system.

[0033] The weight coefficients in the quantitative model for severe processing anomalies are determined by combining the entropy weight method with the analytic hierarchy process (AHP). First, the objective weights of each feature are calculated using the entropy weight method to reflect the dispersion and information content of the feature data itself. Then, the subjective weights are determined by introducing the experience of process experts through the AHP. Finally, the objective weights and subjective weights are weighted and integrated in a 7:3 ratio to obtain the comprehensive weight coefficient. The sum of the comprehensive weight coefficients of all features is 1.

[0034] The task transfer of the multi-device collaborative scheduling unit must meet the process connection constraints and equipment compatibility constraints. The task transfer can only be executed when the current process completion rate of the task to be transferred is not less than 90% and the target equipment has the processing capability and tool configuration of the corresponding process. The task transfer priority is executed in the order of urgent order tasks > regular order tasks > trial production tasks. The transferred task automatically inherits the priority and delivery requirements of the original task. At the same time, the system automatically updates the material distribution plan and the production takt time of subsequent processes.

[0035] The fine-tuning of cutting parameters in the tool lifecycle management unit must comply with safety constraints and range limits. The single fine-tuning range of spindle speed shall not exceed ±5% of the current set value, the single fine-tuning range of feed rate shall not exceed ±3% of the current set value, and the single fine-tuning range of depth of cut shall not exceed ±2% of the current set value. All parameter fine-tuning must not exceed the rated safety parameter range of the corresponding CNC equipment and tool. Spindle vibration and cutting power shall be monitored in real time during the fine-tuning process. If abnormal fluctuations occur, the fine-tuning shall be stopped immediately and the original parameters shall be restored.

[0036] The following two examples further illustrate the specific implementation of this system: Example 1: Implementation of Optimized Control Method for CNC Machining of Aerospace Titanium Alloy Blades This embodiment applies to the precision machining of titanium alloy blades for aerospace engines. The material being machined is TC4 titanium alloy, with a machining accuracy requirement of IT5 and a surface roughness requirement of no more than Ra0.8μm. The machining process includes three stages: roughing, semi-finishing, and finishing, involving various machining techniques such as milling, drilling, and grinding. Titanium alloy has poor thermal conductivity and high hardness, resulting in rapid tool wear during machining and a tendency to generate chatter and machining deformation. Therefore, extremely high requirements are placed on monitoring the machining status and controlling process parameters.

[0037] A multi-source data acquisition unit is deployed at the industrial site to achieve full-dimensional data acquisition of the machining process. The equipment operation data acquisition subunit combines the native interface of the CNC system with edge-side sensors to acquire parameters such as spindle speed, feed rate, cutting power, motor current, spindle vibration, spindle temperature, guideway displacement, and hydraulic pressure at a frequency of 1 millisecond. The process parameter acquisition subunit acquires the command parameters and actual execution parameters of the CNC system in real time, comparing and analyzing the deviation between the command and execution. The machining quality inspection subunit integrates data from online vision inspection equipment, laser displacement sensors, and coordinate measuring machines to achieve multi-dimensional inspection of blade profile dimensions, surface roughness, and geometric tolerances. The environmental condition acquisition subunit deploys environmental sensors at each machining station to acquire station-level temperature, humidity, and vibration parameters. All acquisition units achieve time synchronization through a global time synchronization protocol, ensuring that the time consistency error of data from different sources does not exceed 1 millisecond.

[0038] An industrial big data preprocessing and feature extraction unit is deployed to complete the preprocessing and feature extraction of the collected data. The data preprocessing submodule employs a sliding window filtering algorithm to eliminate high-frequency noise for time-series data, an isolated forest algorithm to identify and remove outlier data, and linear interpolation to fill in missing data. Simultaneously, min-max normalization is performed on all data to eliminate differences in the dimensions of different features. The multi-scale feature extraction submodule uses wavelet packet transform to extract energy features of different frequency bands, employs statistical analysis methods to extract mean, variance, peak value, and kurtosis time-domain features, uses mutual information to calculate the correlation between each feature and processing quality, ranks features based on correlation, and uses principal component analysis to reduce feature dimensionality, generating a low-dimensional, highly discriminative feature subset. For different processing stages—roughing, semi-finishing, and finishing—an adaptive feature extraction strategy is adopted to extract the most sensitive features for the corresponding stage.

[0039] A comprehensive CNC machining database is deployed to achieve time-series storage and unified management of all machining data. The database employs an architecture combining time-series and relational databases. The time-series database stores frequently collected equipment operation data and process parameter data, while the relational database stores machining quality data, equipment information, tool information, and production task information. The database establishes data indexes according to machining batches, part numbers, and equipment numbers, supporting multi-dimensional data retrieval and statistical analysis. All data is stored for at least three years.

[0040] A processing status intelligent identification and anomaly early warning unit is deployed to complete the identification of processing status and the detection and early warning of abnormal events. This unit is pre-trained based on a labeled dataset of 100,000-level titanium alloy processing history data. The processing status classification submodule classifies and identifies different processing statuses, including normal processing, tool wear, tool breakage, chatter, and overload. The anomaly event detection submodule detects and locates various anomalies by comparing real-time collected data with a normal processing status benchmark. A severity score for each anomaly is generated using a processing anomaly severity quantification model, and anomalies are classified into three levels: general anomalies, important anomalies, and emergency anomalies. General anomalies are pushed to on-site operators, important anomalies are simultaneously pushed to team leaders and process engineers, and emergency anomalies are simultaneously pushed to workshop management personnel. An anomaly event knowledge base is also established to store handling methods for various anomalies. Root cause analysis is performed based on anomaly characteristics to generate handling suggestions. Early warning information is disseminated through multiple channels, including on-site audible and visual alarms and terminal message pushes.

[0041] A process parameter intelligent optimization unit is deployed to generate the optimal combination of process parameters. A process parameter knowledge base stores historical optimal combinations of process parameters for different materials, equipment, and processing requirements. A dynamic update mechanism automatically saves effective parameter combinations after each processing task is completed, and outdated data is periodically cleaned up. The multi-objective optimization submodule uses an improved non-dominated sorting genetic algorithm combined with deep reinforcement learning, with processing accuracy, surface quality, processing efficiency, and tool life as optimization objectives, generating a Pareto optimal solution set and selecting the optimal process parameter combination based on user needs. The parameter verification submodule, based on a CNC machining digital twin model, performs virtual machining simulation of the optimized parameters, simulating cutting forces, cutting temperatures, and workpiece deformation during processing, predicting processing quality and tool wear. If the simulation results do not meet requirements, automatic re-optimization is performed. An online adaptive adjustment mechanism for process parameters is incorporated, dynamically fine-tuning parameters based on real-time data during processing to maintain process stability.

[0042] A CNC equipment instruction generation and distribution unit is deployed to complete the conversion and distribution of CNC machining instructions. This unit converts optimized process parameters into G-code and M-code recognizable by the corresponding CNC equipment, automatically generates a complete CNC machining program, and distributes the program to the corresponding CNC equipment for execution via industrial Ethernet. Simultaneously, it receives real-time execution feedback information from the CNC equipment to monitor the execution status of the machining program.

[0043] A closed-loop verification and model iteration unit for processing effects is deployed to continuously iterate and optimize the system model. Based on the quality inspection data after processing, this unit compares and analyzes the deviation between the actual and expected processing effects, calculating the system's processing accuracy pass rate, anomaly detection accuracy, and process parameter optimization effectiveness. Based on the verification results and newly added labeled data, incremental training and parameter updates are performed on the processing status recognition model, anomaly detection model, and process parameter optimization model, with a model update conducted every 100 processing batches.

[0044] A multi-device collaborative scheduling unit is deployed to achieve load balancing and cross-process collaborative scheduling among multiple devices. This unit constructs a digital twin model of the workshop production, mapping the physical production status in real time. Based on the model, it performs simulation-based pre-scheduling of processing tasks, identifying production bottlenecks and equipment conflicts in advance. Task priorities are dynamically adjusted according to order delivery dates, processing difficulty, and urgency, and processing tasks are dynamically allocated based on equipment health status and processing capacity. When a device malfunctions or experiences a processing delay, unfinished tasks are automatically transferred and subsequent production plans are adjusted. The collaborative material delivery system ensures timely material delivery, achieving seamless integration of the entire production process.

[0045] Deploy a tool lifecycle management unit to achieve full-process control over tool usage. This unit integrates spindle vibration, cutting power, motor current, and visual inspection data to achieve multi-dimensional monitoring of tool wear status and establishes a tool wear prediction model to predict remaining tool life. When the remaining tool life reaches a preset threshold, it automatically generates a tool change reminder and schedules idle operators. A tool lifecycle health record is established to document the entire process of tool procurement, use, wear, and tool change. Based on the real-time tool wear level, it automatically fine-tunes the cutting parameters for the corresponding machining task, compensates for machining errors, and intelligently recommends tool change timing.

[0046] In this embodiment, the industrial site multi-source data acquisition unit realizes full-dimensional data acquisition of the titanium alloy processing process. The global time synchronization protocol ensures the time consistency of the data, providing a high-quality data foundation for processing status analysis. The industrial big data preprocessing and feature extraction unit effectively eliminates noise interference in the industrial site, and the adaptive feature extraction strategy improves the relevance of features. The intelligent processing status identification and anomaly early warning unit can promptly detect abnormal events in the processing process, and multi-channel hierarchical early warning improves the efficiency of anomaly handling. The intelligent process parameter optimization unit, combined with digital twin simulation technology, generates the optimal combination of process parameters that meets multiple objective constraints, and the online adaptive adjustment mechanism maintains the stability of the processing process. The multi-equipment collaborative scheduling unit realizes global optimization of workshop production, and the tool lifecycle management unit improves tool utilization and processing quality.

[0047] Example 2: Implementation Method for Optimized Control of Batch CNC Machining of Automobile Engine Cylinder Blocks This embodiment applies to the batch CNC machining of automotive engine cylinder blocks. The material being machined is gray cast iron, with an annual processing capacity of 100,000 units. The machining process includes multiple machining techniques such as milling, drilling, boring, and tapping, involving a flexible production line composed of 12 CNC machining centers. Cylinder block machining is characterized by large batch sizes, numerous processes, and strict cycle time requirements. Failure or abnormality of any single piece of equipment can affect the operation of the entire production line, placing extremely high demands on the coordinated scheduling of multiple machines and production stability.

[0048] Multi-source data acquisition units are deployed on-site to achieve full-dimensional data acquisition across the production line. The equipment operation data acquisition subunit combines the CNC system's native interface with edge-side sensors to collect spindle speed, feed rate, cutting power, motor current, spindle vibration, spindle temperature, guideway displacement, and hydraulic pressure parameters for each machining center at a 1-millisecond acquisition frequency. The process parameter acquisition subunit collects the CNC system's command parameters and actual execution parameters in real time, comparing and analyzing the deviation between commands and execution. The machining quality inspection subunit integrates data from online vision inspection equipment, laser displacement sensors, and online coordinate measuring machines to achieve online detection of key cylinder dimensions, hole accuracy, and surface roughness. The environmental status acquisition subunit deploys environmental sensors at each machining station to collect station-level temperature, humidity, and vibration parameters. All acquisition units achieve time synchronization through a global time synchronization protocol, ensuring that the time consistency error of data from different sources does not exceed 1 millisecond.

[0049] An industrial big data preprocessing and feature extraction unit is deployed to complete the preprocessing and feature extraction of the collected data. The data preprocessing submodule employs a sliding window filtering algorithm to eliminate high-frequency noise for time-series data, an isolated forest algorithm to identify and remove outlier data, and linear interpolation to fill in missing data. Simultaneously, min-max normalization is performed on all data to eliminate differences in the dimensions of different features. The multi-scale feature extraction submodule uses wavelet packet transform to extract energy features of different frequency bands, employs statistical analysis methods to extract mean, variance, peak value, and kurtosis time-domain features, uses mutual information to calculate the correlation between each feature and processing quality, ranks features based on correlation, and uses principal component analysis to reduce feature dimensionality, generating a low-dimensional, highly discriminative feature subset. For different processing stages—roughing, semi-finishing, and finishing—an adaptive feature extraction strategy is adopted to extract the most sensitive features for the corresponding stage.

[0050] A full-process CNC machining database is deployed to achieve time-series storage and unified management of all machining data. The database adopts a distributed time-series database architecture, supporting high-speed writing and rapid retrieval of massive amounts of data. It stores operating data, process parameter data, machining quality data, and production task data for all equipment on the production line. The database establishes multi-level data indexes according to production batch, part number, equipment number, and process number, supporting real-time data monitoring and historical data traceability. The storage period for all data is no less than 5 years.

[0051] A processing status intelligent identification and anomaly early warning unit is deployed to complete the identification of processing status and the detection and early warning of abnormal events. This unit is pre-trained based on a labeled dataset of millions of cylinder block processing historical data. The processing status classification submodule classifies and identifies different processing states, including normal processing, tool wear, tool breakage, chatter, and overload. The anomaly event detection submodule compares real-time collected data with a normal processing status benchmark to detect and locate various anomalies. A severity score for each anomaly is generated using a processing anomaly severity quantification model, and anomalies are classified into three levels: general anomalies, important anomalies, and emergency anomalies. General anomalies are pushed to on-site operators, important anomalies are simultaneously pushed to the production line team leader, and emergency anomalies are simultaneously pushed to workshop management personnel, automatically triggering a production line stoppage procedure. An anomaly event knowledge base is also established to store handling methods for various anomalies. Root cause analysis is performed based on anomaly characteristics to generate handling suggestions. Early warning information is disseminated through multiple channels, including on-site audible and visual alarms and terminal message pushes.

[0052] A process parameter intelligent optimization unit is deployed to generate the optimal combination of process parameters. A process parameter knowledge base stores historical optimal combinations of process parameters for different materials, equipment, and processing requirements. A dynamic update mechanism automatically saves effective parameter combinations after each processing task is completed, and outdated data is periodically cleaned up. The multi-objective optimization submodule uses an improved non-dominated sorting genetic algorithm combined with deep reinforcement learning, taking machining accuracy, surface quality, machining efficiency, and tool life as optimization objectives, to generate a Pareto optimal solution set, and selects the optimal combination of process parameters based on production line cycle time requirements. The parameter verification submodule, based on a CNC machining digital twin model, performs virtual machining simulation of the optimized parameters, simulating cutting forces, cutting temperatures, and tool wear during the machining process. If the simulation results do not meet the requirements, automatic re-optimization is performed. An online adaptive adjustment mechanism for process parameters is incorporated, dynamically fine-tuning parameters based on real-time data during machining to maintain process stability.

[0053] A CNC equipment instruction generation and distribution unit is deployed to complete the conversion and distribution of CNC machining instructions. This unit converts optimized process parameters into G-code and M-code recognizable by the corresponding CNC equipment, automatically generates a complete CNC machining program, and distributes the program to the corresponding CNC equipment for execution via industrial Ethernet. Simultaneously, it receives real-time execution feedback information from the CNC equipment, monitoring the execution status of the machining program and the production line's operating cycle time.

[0054] A closed-loop verification and model iteration unit for processing effects is deployed to continuously iterate and optimize the system model. Based on online quality inspection data, this unit compares and analyzes the deviation between the actual and expected processing effects, calculating the system's processing accuracy pass rate, anomaly detection accuracy, and process parameter optimization effectiveness. Based on the verification results and newly added labeled data, incremental training and parameter updates are performed on the processing status recognition model, anomaly detection model, and process parameter optimization model, with a model update occurring every 1000 processing batches.

[0055] A multi-device collaborative scheduling unit is deployed to achieve collaborative scheduling and load balancing of multiple devices on the production line. This unit constructs a digital twin model of the production line, mapping its physical production status in real time. Based on the model, it performs simulation-based pre-scheduling of processing tasks, proactively identifying production bottlenecks and equipment conflicts. Task priorities are dynamically adjusted according to order delivery dates, processing difficulty, and urgency, and processing tasks are dynamically allocated based on equipment health status and processing capacity. When a device malfunctions or experiences a processing delay, unfinished processing tasks are automatically transferred to other idle devices. Simultaneously, the execution order of subsequent tasks is adjusted, and the material delivery system coordinates to ensure timely material delivery, maintaining the continuous and stable operation of the production line.

[0056] Deploy a tool lifecycle management unit to achieve full lifecycle control of all tools on the production line. This unit integrates spindle vibration, cutting power, motor current, and visual inspection data to achieve multi-dimensional monitoring of the wear status of each tool, and establishes a tool wear prediction model to predict the remaining tool life. When the remaining tool life reaches a preset threshold, it automatically generates a tool change reminder and schedules idle operators to complete the tool change operation within the production line's tool change window, avoiding disruption to production cycle time. A tool lifecycle health record is established to record the entire process of each tool's procurement, use, wear, and tool change. Based on the real-time wear level of the tool, it automatically fine-tunes the cutting parameters for the corresponding machining task to compensate for machining errors.

[0057] In this embodiment, the industrial site multi-source data acquisition unit realizes full-dimensional data acquisition of the batch processing of automotive engine cylinder blocks, providing a data foundation for comprehensive monitoring of the production line. The industrial big data preprocessing and feature extraction unit effectively processes massive amounts of industrial data, improving the quality of features. The intelligent processing status identification and anomaly early warning unit can promptly detect processing anomalies, and emergency anomalies automatically trigger the production line pause procedure, avoiding the generation of batch scrap. The intelligent process parameter optimization unit generates the optimal combination of process parameters that meets the production line cycle time requirements, and the online adaptive adjustment mechanism maintains the stability of batch processing. The multi-equipment collaborative scheduling unit realizes global optimization of the production line, improving overall production efficiency. The tool lifecycle management unit realizes unified management of tools from multiple machines, improving tool utilization.

[0058] Reference Figure 2 This diagram details the evolution from low-level multi-source sensing to high-dimensional feature reduction. The acquisition end achieves global time synchronization of four-dimensional data (equipment, process, quality, and environment) via a millisecond-level high-frequency interface. After the data enters the preprocessing stage, sliding window filtering, isolated forest, and linear interpolation are used to denoise, remove, and complete data, and normalization eliminates dimensional differences. Finally, the multi-scale feature extraction module adaptively extracts time-frequency domain features for different stages such as coarse and fine processing, and uses mutual information and principal component analysis (PCA) to perform feature ranking and dimensionality reduction, outputting a low-dimensional, highly discriminative feature subset.

[0059] Reference Figure 3 This diagram illustrates the system's intelligent classification and multi-channel hierarchical response mechanism for processing anomalies. The system loads a pre-trained recognition model and inputs high-discrimination features. The processing status classification module precisely identifies specific abnormal states such as wear, breakage, and chatter. The anomaly detection module then compares the deviation with historical normal benchmarks to calculate a processing anomaly severity score. Based on the score, the system triggers corresponding low, medium, and high-level early warning branches, automatically retrieves the anomaly event knowledge base for root cause analysis, and issues handling suggestions through differentiated channels such as terminal push notifications and on-site audible and visual alarms.

[0060] Reference Figure 4 This diagram illustrates the control loop of process parameters from optimization calculation to digital twin virtual verification. The process parameter knowledge base is used as the initial optimal combination of inputs. A multi-objective optimization submodule integrates an improved non-dominated sorting genetic algorithm with deep reinforcement learning to balance four core indicators: machining accuracy, surface quality, machining efficiency, and tool life, generating a Pareto optimal solution set. Before parameters are issued, virtual machining simulation must be performed in the CNC machining digital twin model to dynamically simulate cutting forces, temperature, and deformation. If quality constraints are not met, a forced return to re-optimization is initiated; if verification is successful, an online adaptive fine-tuning mechanism is activated.

[0061] Reference Figure 5 This diagram illustrates the advanced collaborative mechanism of the system at the levels of overall workshop scheduling and key tooling (tool) management. On the scheduling side, the production digital twin model maps the physical workshop in real time, achieving load balancing and dynamic allocation of processing tasks based on delivery dates and equipment health, and coordinating with warehousing and distribution. On the tool management side, the system integrates vibration, power, and current data to predict tool life. The two units are vertically integrated: when the remaining tool life reaches a critical threshold or a sudden equipment failure occurs, the system automatically triggers a replacement reminder, fine-tunes processing parameters to compensate for errors, and automatically transfers remaining tasks to idle equipment, ensuring seamless operation of the entire production line.

[0062] The above are merely preferred embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A numerical control machining optimization control system based on industrial big data processing, characterized in that, include: The system includes: an industrial site multi-source data acquisition unit; an industrial big data preprocessing and feature extraction unit that communicates bidirectionally with the industrial site multi-source data acquisition unit; a CNC machining full-process database that communicates bidirectionally with the industrial big data preprocessing and feature extraction unit; a machining status intelligent identification and anomaly early warning unit that communicates bidirectionally with the CNC machining full-process database; a process parameter intelligent optimization unit that communicates bidirectionally with the CNC machining full-process database and the machining status intelligent identification and anomaly early warning unit; a CNC equipment instruction generation and issuance unit that communicates bidirectionally with the process parameter intelligent optimization unit; and a machining effect closed-loop verification and model iteration unit that communicates bidirectionally with the CNC equipment instruction generation and issuance unit. Industrial site multi-source data acquisition unit: completes real-time acquisition of equipment operation data, process parameter data, processing quality data, and environmental status data during CNC machining; Industrial Big Data Preprocessing and Feature Extraction Unit: Completes the cleaning, standardization and multi-dimensional feature extraction of collected data; CNC machining full-process database completes the time-series storage and unified management of all machining data. Intelligent processing status recognition and anomaly early warning unit: Completes processing status recognition and abnormal event detection and early warning; Intelligent process parameter optimization unit: generates the optimal combination of process parameters based on historical data and real-time status; The CNC equipment instruction generation and issuance unit completes the conversion and issuance of CNC machining instructions. The machining effect closed-loop verification and model iteration unit completes the machining effect verification and continuous iterative optimization of the system model.

2. The industrial big data processing based NC machining optimization control system according to claim 1, characterized in that, It also includes a multi-device collaborative scheduling unit, which has bidirectional communication connections with the CNC machining full-process database, the process parameter intelligent optimization unit, and the CNC equipment instruction generation and issuance unit. It constructs a digital twin model of workshop production to map the physical production status of the workshop in real time. Based on the model, it completes the simulation pre-scheduling of processing tasks, identifies production bottlenecks and equipment conflicts in advance, dynamically adjusts task priorities according to order delivery date, processing difficulty, and urgency, and completes the dynamic allocation of processing tasks by combining equipment health status and processing capacity, thus achieving load balancing and cross-process collaborative scheduling among multiple devices. 3.The industrial big data processing based NC machining optimization control system according to claim 1, wherein, It also includes a tool lifecycle management unit, which has bidirectional communication connections with the industrial site multi-source data acquisition unit, the CNC machining full-process database, and the process parameter intelligent optimization unit. It integrates spindle vibration, cutting power, motor current, and visual inspection data to complete multi-dimensional monitoring of tool wear status, establishes a tool wear prediction model to predict the remaining tool life, automatically generates a tool replacement reminder and schedules idle operators when the remaining tool life reaches a preset threshold, establishes a tool lifecycle health record to record the entire process information of tool procurement, use, wear, and tool replacement, and automatically fine-tunes the cutting parameters of the corresponding machining task according to the real-time wear degree of the tool.

4. The industrial big data processing based NC machining optimization control system according to claim 1, characterized in that, The industrial site multi-source data acquisition unit includes an equipment operation data acquisition subunit, a process parameter acquisition subunit, a machining quality inspection subunit, and an environmental status acquisition subunit. The equipment operation data acquisition subunit simultaneously collects equipment operating parameters such as spindle temperature, guide rail displacement, and hydraulic pressure through a combination of the CNC system's native interface and edge-side sensors. The process parameter acquisition subunit collects the CNC system's command parameters and actual execution parameters in real time, comparing and analyzing the deviation between the command and execution. The machining quality inspection subunit integrates data from online visual inspection equipment, laser displacement sensors, and coordinate measuring machines. The environmental status acquisition subunit adopts a distributed deployment approach, deploying environmental sensors at each machining station to collect station-level temperature, humidity, and vibration parameters. All acquisition units achieve time synchronization through a global time synchronization protocol.

5. The industrial big data processing based NC machining optimization control system according to claim 1, characterized in that, The industrial big data preprocessing and feature extraction unit has a built-in data preprocessing submodule and a multi-scale feature extraction submodule. The data preprocessing submodule uses a sliding window filtering algorithm to eliminate high-frequency noise for time series data, uses an isolated forest algorithm to identify and remove abnormal data, uses linear interpolation to fill missing data, and performs normalization processing on all data. The multi-scale feature extraction submodule uses wavelet packet transform to extract energy features of different frequency bands, employs statistical analysis methods to extract mean, variance, peak value, and kurtosis time-domain features, uses mutual information to calculate the correlation between each feature and processing quality, ranks features based on correlation, and uses principal component analysis to reduce feature dimensionality, generating a low-dimensional, highly discriminative feature subset. For different processing stages such as roughing, semi-finishing, and finishing, an adaptive feature extraction strategy is adopted to extract the most sensitive features for the corresponding stage.

6. The industrial big data processing based NC machining optimization control system according to claim 1, wherein, The intelligent machining status recognition and anomaly early warning unit is pre-trained based on an annotated CNC machining historical dataset. It includes a machining status classification submodule and an anomaly event detection submodule. The machining status classification submodule classifies and identifies different machining states, including normal machining, tool wear, tool breakage, chatter, and overload. The anomaly event detection submodule detects and locates various anomalies by comparing real-time collected data with a normal machining status benchmark. A severity score for each anomaly is generated using a machining anomaly severity quantification model. The calculation formula is as follows: ;in, To score the severity of processing abnormalities, This represents the total number of anomaly detection features, with a dimension of 1. For the first The weight coefficients of each anomaly detection feature. For the first Real-time measurement values ​​of each feature For the first Normal processing baseline values ​​for each feature For the first The difference in the normal processing range of each feature.

7. The CNC machining optimization control system based on industrial big data processing according to claim 1, characterized in that, The intelligent optimization unit for process parameters incorporates a process parameter knowledge base, a multi-objective optimization submodule, and a parameter verification submodule. The process parameter knowledge base stores historically optimal process parameter combinations for different materials, equipment, and processing requirements, and uses a dynamic update mechanism to automatically store effective parameter combinations after each processing task is completed. The multi-objective optimization submodule employs an improved non-dominated sorting genetic algorithm combined with deep reinforcement learning, using processing accuracy, surface quality, processing efficiency, and tool life as optimization objectives to generate a Pareto optimal solution set, selecting the optimal process parameter combination based on user needs. The parameter verification submodule, based on a CNC machining digital twin model, performs virtual machining simulation of the optimized parameters, simulating cutting forces, cutting temperatures, and workpiece deformation during processing, predicting processing quality and tool wear. If the simulation results do not meet requirements, it automatically re-optimizes. An online adaptive adjustment mechanism for process parameters is also included, dynamically fine-tuning parameters based on real-time data during processing.

8. The numerical control machining optimization control method based on industrial big data processing, applicable to the numerical control machining optimization control system based on industrial big data processing according to any one of claims 1 to 7, characterized in that, Includes the following steps: The system uses a multi-source data acquisition unit in the industrial field to collect real-time data on equipment operation, process parameters, machining quality, and environmental conditions during the CNC machining process. The industrial big data preprocessing and feature extraction unit completes the preprocessing and multi-dimensional feature extraction of the collected data, and stores the processed data into the CNC machining full-process database. The intelligent processing status identification and anomaly early warning unit completes the processing status identification and abnormal event detection and early warning. When an abnormal event is detected, the corresponding level of early warning information is issued and the emergency response procedure is initiated. Based on historical processing data and real-time processing status, the intelligent process parameter optimization unit generates the optimal combination of process parameters that meets multiple objective constraints. The optimized process parameters are converted into machining instructions that can be recognized by the CNC equipment through the CNC equipment instruction generation and issuance unit, and then issued to the corresponding CNC equipment for execution; The processing effect is verified in real time through closed-loop verification of processing effect and model iteration unit. Based on the verification results, the incremental training and parameter update of the system model are completed. 9.The NC machining optimization control method based on industrial big data processing according to claim 8, wherein, A digital twin model of workshop production is constructed by a multi-device collaborative scheduling unit. Based on the model, the simulation pre-scheduling of processing tasks is completed, production bottlenecks and equipment conflicts are identified in advance, and task priorities are dynamically adjusted according to order delivery date, processing difficulty and urgency. The dynamic allocation of processing tasks is completed by combining equipment health status and processing capacity. 10.The NC machining optimization control method based on industrial big data processing according to claim 8, wherein, By integrating multi-source sensor data through the tool lifecycle management unit, real-time monitoring of tool wear status is achieved. A tool wear prediction model is established to predict the remaining tool life. When the remaining tool life reaches a preset threshold, a tool change reminder is automatically generated and idle operators are dispatched. At the same time, non-urgent machining tasks on the corresponding equipment are suspended. After the tool change is completed, tool setting and parameter calibration are automatically performed, and the tool lifecycle health record is updated.

Images