A constant stability control system for a numerical control machining center

Abstract

The application provides a constant stability control system for a numerical control machining center, which comprises a constant stability control unit, a multi-dimensional sensor group, a data storage module and a compensation execution module, the system takes the constant stability control unit as an execution main body, and the control method of the constant stability control unit comprises the following steps: step one, the constant stability control unit collects tool wear data, load fluctuation data and material hardness data in a machining process of the numerical control machining center in real time through the preset multi-dimensional sensor group, and the collection frequency is 99 Hz. Through high-frequency collection of working condition data by the multi-dimensional sensor group, a multi-dimensional perception model is constructed in combination with a fuzzy control algorithm, comprehensive compensation under the multi-factor coupling of tool wear, load fluctuation and material hardness is realized, the limitation of existing fixed parameter compensation is broken, the compensation range is expanded from plus or minus 5% to plus or minus 20%, and the adaptability is significantly improved.

Description

Technical Field

[0001] This invention relates to the field of precision machining technology, and in particular to a constant stability control system for CNC machining centers. Background Technology

[0002] As the core equipment for precision machining, the compensation accuracy of the constant stability control system of CNC machining center directly determines the dimensional accuracy and surface quality of the machined parts. Especially in high-end manufacturing fields such as aerospace, the requirements for machining stability are even more stringent.

[0003] In existing constant-stability control systems of CNC machining centers, adaptive compensation functions mostly employ fixed-parameter compensation modes. Their core drawback lies in their inability to adjust compensation parameters in real-time according to dynamic operating conditions during machining. During machining, dynamic factors such as tool wear, load fluctuations, and material hardness changes are constantly present. Fixed-parameter compensation cannot adapt to these changing conditions, leading to a continuous accumulation of machining errors. Statistics show that some aerospace parts manufacturers experience scrap rates as high as 8% due to this problem, resulting in significant production cost waste. While current technologies involve tool management and related techniques, none have achieved dynamic self-tuning of compensation parameters, nor do they cover the comprehensive compensation scenarios involving the coupling of multiple factors such as tool wear, load fluctuations, and material hardness. The compensation range is narrow, only ±5%, with poor adaptability, failing to meet the machining requirements of high-precision parts. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to overcome the defects of the existing technology. The present invention proposes a constant and stable control system for CNC machining centers.

[0005] To address the issue that existing constant-stability control systems in CNC machining centers often employ fixed-parameter compensation, a core flaw lies in their inability to adjust compensation parameters in real-time based on dynamic operating conditions during machining. During machining, dynamic factors such as tool wear, load fluctuations, and material hardness changes are ever-present. Fixed-parameter compensation cannot adapt to these changing conditions, leading to a continuous accumulation of machining errors. Statistics show that some aerospace parts manufacturers experience scrap rates as high as 8% due to this problem, resulting in significant production cost waste. While current technologies involve tool management and related techniques, none have achieved dynamic self-tuning of compensation parameters, nor do they cover the comprehensive compensation scenarios involving multiple coupled factors such as tool wear, load fluctuations, and material hardness. Furthermore, their compensation range is narrow, only ±5%, exhibiting poor adaptability and failing to meet the machining requirements of high-precision parts. The technical solution adopted in this invention is: A constant-stability control system for a CNC machining center includes: a constant-stability control unit, a multi-dimensional sensor group, a data storage module, and a compensation execution module. The system uses the constant-stability control unit as the execution entity, and the control method of the constant-stability control unit includes the following steps: Step 1: The constant stability control unit collects tool wear data, load fluctuation data, and material hardness data in real time during the CNC machining process through a preset multi-dimensional sensor group, with a collection frequency of 99Hz; Step 2: The constant stability control unit preprocesses the collected tool wear data, load fluctuation data, and material hardness data, and removes abnormal data; it constructs a multi-dimensional perception model based on the fuzzy control algorithm, and performs feature extraction and fusion on the preprocessed multi-dimensional working condition data; it establishes the correlation modeling logic between tool wear and machining error, and realizes the mapping between working condition data and machining error. Step 3: Based on the processed operating data and the associated model, the constant stability control unit determines whether the current processing conditions have changed and whether the processing error exceeds the preset threshold; at the same time, it determines whether the accuracy of the compensation model has decreased. Step 4: Based on the judgment results of Step 3, the constant stability control unit automatically adjusts the compensation parameters to achieve a compensation range of ±20% and a compensation accuracy of 95%. Online meta-learning technology is introduced to update the parameters of the compensation model, realize the self-evolution of the compensation model, and control the accuracy fluctuation during long-term operation within ±1.5 micrometers.

[0006] Preferably, the data preprocessing includes data denoising and data standardization, wherein the data standardization uses the Z-score standardization algorithm, and the calculation formula is: in, For standardized data, This is the original collected data. The mean of the original data. This represents the standard deviation of the original data.

[0007] Preferably, the multi-dimensional perception model constructed by the fuzzy control algorithm includes three stages: fuzzification, fuzzy inference, and defuzzification. The fuzzification employs a triangular membership function, and the tool wear amount... Load fluctuation value Material hardness The membership functions are as follows: Tool wear membership function: Membership function for load fluctuation value: Material hardness membership function: in, These are three threshold values ​​for tool wear. These are three thresholds for load fluctuation values. These are three threshold values ​​for material hardness.

[0008] Preferably, the correlation modeling logic between tool wear and machining error adopts a linear regression model, and the calculation formula is as follows: in, Due to machining error, These are the error influence coefficients for tool wear, load fluctuation, and material hardness, respectively. This is the error correction term, with a value range of ±0.1 micrometers; This refers to the amount of tool wear. is the load fluctuation value, and h is the material hardness.

[0009] Preferably, the online meta-learning technique enables the self-evolution of the compensation model, and the model parameters are updated using the gradient descent method. The parameter update formula is as follows: in, For the updated compensation model parameters, These are the parameters of the current compensation model. The learning rate is 0.01 to 0.05. The gradient of the loss function under the current parameters, the loss function , This represents the actual machining error. To predict processing errors in the model, This represents the number of samples.

[0010] Preferably, the preset threshold for processing error is ±0.01mm. When the processing error exceeds this threshold, the constant stability control unit immediately initiates the compensation parameter adjustment process.

[0011] Preferably, the multi-dimensional sensor group includes a tool wear sensor, a load sensor, and a hardness sensor, which respectively collect the tool wear amount, spindle load fluctuation value, and hardness value of the processed material.

[0012] Preferably, the constant stability control unit is used to execute the entire process of data acquisition, data processing, data judgment, compensation parameter adjustment and model self-evolution, and the constant stability control unit has a built-in fuzzy control algorithm and an online meta-learning model; The multi-dimensional sensor group includes a tool wear sensor, a load sensor, and a hardness sensor. The tool wear sensor, load sensor, and hardness sensor are respectively installed at the tool chuck, spindle, and worktable of the CNC machining center to collect tool wear amount, spindle load fluctuation value, and aluminum alloy material hardness value. The data storage module is used to store the collected working condition data, processed data, compensation parameters, model parameters and processing error data, which facilitates subsequent data traceability and model optimization. The compensation execution module is linked with the servo system of the CNC machining center, receives the compensation parameters output by the constant stability control unit, and drives the servo motor to adjust the feed speed and spindle speed to achieve real-time compensation of machining errors.

[0013] Compared with the prior art, the beneficial effects of the present invention are: This invention uses a multi-dimensional sensor array to collect working condition data at high frequency and combines it with a fuzzy control algorithm to construct a multi-dimensional perception model. This enables comprehensive compensation for multiple factors such as tool wear, load fluctuation, and material hardness. It breaks through the limitations of existing fixed parameter compensation, expands the compensation range from ±5% to ±20%, and significantly improves adaptability.

[0014] This invention achieves dynamic self-tuning of compensation parameters. By establishing a correlation model between tool wear and machining error, it captures changes in working conditions in real time and automatically adjusts compensation parameters, achieving a compensation accuracy of 95%. This effectively reduces the accumulation of machining errors and can reduce the scrap rate of aerospace parts from 8% to below 1%, significantly reducing production costs.

[0015] This invention introduces online meta-learning technology to achieve self-evolution of the compensation model, which solves the problem of compensation accuracy decay caused by mechanical wear during long-term operation, controls accuracy fluctuation within ±1.5 micrometers, ensures the long-term stability of CNC machining center operation, and meets the processing requirements of high-end precision parts. Attached Figure Description

[0016] The disclosure of this invention is illustrated with reference to the accompanying drawings. It should be understood that the drawings are for illustrative purposes only and are not intended to limit the scope of protection of this invention. In the drawings, the same reference numerals are used to refer to the same parts. Wherein: Figure 1 This is a block diagram illustrating the construction of the constant-stability control system of the present invention; Figure 2 This is a flowchart illustrating the control process of the constant stability control system of the present invention. Detailed Implementation

[0017] It is readily understood that, based on the technical solution of this invention, those skilled in the art can propose various interchangeable structural methods and implementations without altering the essential spirit of the invention. Therefore, the following detailed embodiments and accompanying drawings are merely illustrative examples of the technical solution of this invention and should not be considered as the entirety of the invention or as limitations or restrictions on the technical solution of this invention.

[0018] Specific embodiments of the present invention are described below with reference to the accompanying drawings.

[0019] Please see Figures 1-2 This embodiment proposes a constant-stability control system for CNC machining centers, including a constant-stability control unit, a multi-dimensional sensor group, a data storage module, and a compensation execution module, wherein: Stability Control Unit: As the sole execution entity, it adopts an NVIDIA Jetson AGX Orin computing unit with a main frequency of 4.5GHz and 32GB of memory. It is used to execute all processes such as data acquisition, data processing, data judgment, compensation parameter adjustment and model self-evolution. It has built-in fuzzy control algorithm and online meta-learning model. Multi-dimensional sensor group: including tool wear sensor (model: Kistler9257B), load sensor (model: HBMC16), hardness sensor (model: WilsonVH1150), which are installed at the tool chuck, spindle, and worktable of the CNC machining center to collect tool wear, spindle load fluctuation value, and aluminum alloy material hardness value. Data storage module: Uses SSD solid-state drive with a capacity of 1TB, used to store collected working condition data, processed data, compensation parameters, model parameters and processing error data, to facilitate subsequent data traceability and model optimization; Compensation execution module: It is linked with the servo system of the CNC machining center, receives the compensation parameters output by the constant stability control unit, drives the servo motor to adjust the feed speed and spindle speed, and realizes real-time compensation of machining errors.

[0020] The specific steps of the control flow of this system are as follows: The constant-stability control unit controls the multi-dimensional sensor group to start data acquisition, with the acquisition frequency set to 99Hz, meaning data is acquired approximately every 10.1ms. 1. Tool wear sensor collects tool wear data. The acquisition range is 0–0.1 mm, and the accuracy is 0.001 mm. 2. The load sensor collects the load fluctuation value of the spindle. The acquisition range is 0–1000N, and the accuracy is 1N. 3. Hardness sensor collects hardness values ​​of aluminum alloy materials. The acquisition range is 50–150 HB, and the accuracy is 1 HB. The collected raw data is transmitted to the constant stability control unit in real time and stored in the data storage module.

[0021] The constant-stability control unit preprocesses the collected raw data and then constructs a multi-dimensional perception model using a fuzzy control algorithm. The specific process is as follows: 1. Data Preprocessing: First, the 3σ criterion is used to remove outliers, i.e., when the original data exceeds... When the range is reached, the data is identified as outlier and removed; subsequently, the data is standardized using the Z-score standardization algorithm, calculated as follows: in, For standardized data, This is the original collected data. The mean of the original data. This represents the standard deviation of the original data. In this embodiment, the mean of the original data for tool wear is... Standard deviation The mean of the raw data for load fluctuation values Standard deviation Mean value of raw material hardness data Standard deviation .

[0022] 2. Multi-dimensional perception model construction: Based on fuzzy control algorithms, a multi-dimensional perception model for tool wear, load fluctuation, and material hardness is constructed, consisting of three stages: fuzzification, fuzzy inference, and defuzzification. (1) Fuzzification: Using the triangular membership function, the standardized working condition data is mapped to fuzzy quantities, and the thresholds for each parameter are set as follows: tool wear amount , , Load fluctuation value , , Material hardness , , The corresponding membership functions are as follows: Tool wear membership function: Membership function for load fluctuation value: Material hardness membership function: (2) Fuzzy Reasoning: The Mamdani reasoning method is adopted, and fuzzy reasoning rules are set. For example, if the tool wear is small, the load fluctuation is small, and the material hardness is moderate, the compensation parameter takes the minimum value; if the tool wear is large, the load fluctuation is large, and the material hardness is high, the compensation parameter takes the maximum value. A total of 27 fuzzy reasoning rules are set, covering all working condition combinations. The reasoning formula is as follows: in, For the membership degree of a fuzzy relationship, , These represent the membership degrees of the input fuzzy quantities. This is for the smaller value operation.

[0023] (3) Defuzzification: Using the centroid method, the fuzzy quantities obtained from fuzzy inference are converted into precise compensation parameter adjustment quantities. The calculation formula is as follows: in, To accurately compensate for parameter adjustments, For the first Membership degree of a fuzzy quantity For the first The precise value corresponding to each fuzzy quantity The number of fuzzy quantities.

[0024] 3. Correlation Modeling: Establish a correlation model between tool wear and machining error, using a linear regression model. The calculation formula is as follows: in, For machining error (unit: micrometers) Tool wear (unit: mm). Load fluctuation value (unit: N). Material hardness (unit: HB), error correction term The coefficients were obtained through multiple experimental calibrations (the tool wear amount is in mm, which needs to be multiplied by 1000 to be converted to micrometers; combined with the experimental calibration coefficients, the coefficients were corrected to 800 to ensure accurate error calculation).

[0025] The constant-stability control unit makes two judgments based on the processed operating condition data and the correlation model: 1. Operating condition changes and error judgment: The constant stability control unit calculates the current machining error. Compare it with a preset threshold of ±0.01 mm (i.e., ±10 micrometers). If If the error is measured in micrometers, it is determined that the processing error exceeds the threshold, and the compensation parameters need to be adjusted. At the same time, the current working condition data is compared with the working condition data at the previous moment. If the change in any data exceeds 10%, it is determined that the working condition has changed, and the compensation parameter adjustment process is initiated.

[0026] 2. Model accuracy decay assessment: The steady-state control unit calculates and compensates for the prediction error of the model. ,in To predict the processing error using the model, if the prediction error is collected in 10 consecutive samples... If the accuracy of the compensation model decreases to micrometers, the online meta-learning model evolution process is initiated.

[0027] Based on the above judgment results, the constant stability control unit executes the corresponding output operation: 1. Compensation Parameter Adjustment: If it is determined that the compensation parameters need to be adjusted, the constant stability control unit automatically updates the compensation parameters based on the precise adjustment amount obtained from the defuzzification, controlling the compensation range within ±20% to ensure that the compensation accuracy reaches 95%, i.e., the compensation reduces the machining error. Micrometers; the compensation parameters are transmitted to the servo system through the compensation execution module, which drives the servo motor to adjust the feed speed and spindle speed to achieve error compensation.

[0028] 2. Model Self-Evolution: If it is determined that the accuracy of the compensation model has decayed, the steady-state control unit initiates the online meta-learning process, using gradient descent to update the model parameters. The parameter update formula is: Among them, learning rate loss function , This represents the actual processing error from the last 10 data collections. The corresponding model predicts the processing error; through parameter updates, the compensation model achieves self-evolution, and the accuracy fluctuation is controlled back within ±1.5 micrometers.

[0029] The constant stability control unit periodically verifies and calibrates the compensation model. Every 24 hours of operation, it collects 100 sets of processing data, calculates the compensation accuracy and accuracy fluctuation. If the compensation accuracy is lower than 95% or the accuracy fluctuation exceeds ±1.5 micrometers, it automatically adjusts the membership function threshold of the fuzzy control algorithm and the learning rate of online meta-learning to ensure that the model is always in the optimal state. At the same time, the verification data is stored in the data storage module for subsequent model optimization.

[0030] Example Example 1 (Aerospace Aluminum Alloy Parts Machining Scenario) This embodiment uses a VMC850 CNC machining center to process aerospace aluminum alloy parts (material hardness 100±10HB). The machining process is milling, with a spindle speed of 10000 r / min, a feed rate of 500 mm / min, and a machining accuracy requirement of ±5 micrometers. The constant stability control system of this invention is installed on the CNC machining center. The sensor group is installed according to the requirements of Section 5.1, and the parameters of the constant stability control unit are set according to Section 5.2.

[0031] Implementation process 1. Start the CNC machining center and the constant stability control system. The multi-dimensional sensor group begins to collect data at a frequency of 99Hz, and collects tool wear, load fluctuation value and material hardness value in real time. 2. The constant stability control unit preprocesses the collected raw data, removes abnormal data, performs standardization, and then extracts features and performs fuzzy inference through a multi-dimensional perception model constructed by fuzzy control algorithm. After defuzzification, the compensation parameter adjustment amount is obtained. 3. The constant stability control unit calculates the current processing error through the correlation model and determines whether the processing error exceeds the preset threshold, whether the working condition has changed, and whether the model accuracy has decreased. 4. When the tool wear increases from 0.02mm to 0.06mm (80% change), the load fluctuation increases from 450N to 600N (33.3% change), and the material hardness increases from 95HB to 115HB (21.1% change) during machining, the working conditions change significantly, the machining error reaches 12 micrometers, exceeding the preset threshold, and the constant stability control unit automatically adjusts the compensation parameters, adjusting the compensation amount from +3% to +12%. 5. After 48 hours of continuous processing, the accuracy of the compensation model decayed, and the prediction error reached 1.8 micrometers for 10 consecutive times. The constant stability control unit started the online meta-learning model evolution, updated the model parameters, and after evolution, the prediction error dropped to 1.2 micrometers, and the accuracy fluctuation was controlled within ±1.5 micrometers. 6. After processing, 100 aluminum alloy parts were subjected to precision testing, and processing errors and scrap rates were statistically analyzed.

[0032] The implementation results of this embodiment are as follows: 1. Compensation accuracy: The machining error is controlled within ±0.5 micrometers, and the compensation accuracy reaches 97%, exceeding the preset 95%, which meets the machining accuracy requirements of aerospace parts; 2. Precision fluctuation: During long-term (48-hour) operation, the precision fluctuation is controlled within ±1.2 micrometers, which is lower than the preset ±1.5 micrometers, thus achieving long-term processing stability; 3. Scrap rate: Out of 100 processed parts, the number of scraps is 1, reducing the scrap rate to 1%, which is significantly lower than the 8% of the existing technology. 4. Adaptability: The compensation range is stable at ±20%, which can effectively adapt to the dynamic changes of tool wear, load fluctuation and material hardness. No manual adjustment of compensation parameters is required, and the degree of automation is high.

[0033] The implementation results show that the constant stability control system for CNC machining centers of the present invention can effectively solve the shortcomings of the prior art, realize the dynamic self-tuning of compensation parameters and model self-evolution, improve machining stability and accuracy, reduce scrap rate, adapt to the machining needs of high-end precision parts, and has good practicality and promotion value.

[0034] Example 2 (Automotive Parts Processing Scenario) This embodiment uses a VMC1060 CNC machining center to process an automobile engine cylinder block (cast iron material, material hardness 180±15HB). The machining process is boring, with a spindle speed of 8000 r / min, a feed rate of 300 mm / min, and a machining accuracy requirement of ±8 micrometers. The constant stability control system of this invention is installed on the CNC machining center. The sensor group follows the model in Section 5.1, but the installation positions are adjusted as follows: the tool wear sensor is installed at the boring bar, the load sensor is installed at the end of the boring spindle, and the hardness sensor is installed next to the table feed guide. The parameters of the constant stability control unit are adjusted as follows: learning rate α=0.04, and machining error preset threshold ±0.015 mm (±15 micrometers).

[0035] Implementation process 1. Start the CNC machining center and the constant stability control system. The multi-dimensional sensor group starts data acquisition at a frequency of 99Hz to collect the boring tool wear, boring spindle load fluctuation value, and cast iron material hardness value in real time. 2. The constant stability control unit preprocesses the collected raw data, removes outliers using the 3σ criterion, and then standardizes it using the Z-score standardization algorithm (in this embodiment, the average boring tool wear is used). Standard deviation Average load fluctuation value Standard deviation Average material hardness Standard deviation ); 3. Construct a multi-dimensional perception model based on fuzzy control algorithm and adjust the membership function threshold: tool wear amount Load fluctuation value Material hardness The process involves fuzzification, fuzzy inference (using 27 inference rules adapted to the processing characteristics of cast iron materials), and defuzzification to obtain the adjustment amount of the compensation parameters. 4. Establish a correlation model adapted to cast iron boring, and adjust the calculation formula as follows: (in, The measurement is for machining error, in micrometers. This refers to the wear of the boring tool, in mm. This represents load fluctuation values, in N. Material hardness, unit HB; (micrometers, calibrated through cast iron machining experiments). 5. The constant stability control unit calculates the current machining error through an association model and judges the changes in working conditions and error thresholds: when the boring tool wear increases from 0.03mm to 0.07mm (change of 133.3%), the load fluctuation increases from 550N to 720N (change of 30.9%), the material hardness increases from 170HB to 195HB (change of 14.7%), and the machining error reaches 17 micrometers, exceeding the preset threshold of ±15 micrometers, the constant stability control unit automatically adjusts the compensation parameters, adjusting the compensation amount from +4% to +14%. 6. After 72 hours of continuous processing, the accuracy of the compensation model began to decay, with the prediction error reaching 1.7 micrometers for 10 consecutive times. The constant stability control unit then initiated online meta-learning model evolution, updating the formula according to the parameters. After updating the model parameters, the prediction error was reduced to 1.3 micrometers, and the accuracy fluctuation was controlled within ±1.5 micrometers. 7. After machining, the precision of the boring parts of 150 automobile engine cylinder blocks was inspected, and the machining error and scrap rate were statistically analyzed.

[0036] The implementation results of this embodiment are as follows: 1. Compensation accuracy: The boring machining error is controlled within ±0.6 micrometers, and the compensation accuracy reaches 96.7%, exceeding the preset 95%, which meets the machining accuracy requirements of automobile engine cylinder blocks; 2. Precision fluctuation: During long-cycle (72-hour) operation, the precision fluctuation is controlled within ±1.3 micrometers, which is lower than the preset ±1.5 micrometers, achieving stability during long-term continuous processing; 3. Scrap rate: Out of 150 processed parts, the number of scrapped parts is 2, and the scrap rate is reduced to 1.3%. Compared with the existing technology of 8%, the scrap rate is significantly reduced, which reduces the production cost of automotive parts. 4. Adaptability: The compensation range is stable at ±20%, which can effectively adapt to the machining conditions of high hardness of cast iron materials, fast wear of boring tools, and large load fluctuations. No manual intervention is required to adjust the compensation parameters, which can meet the batch processing needs of automotive parts.

[0037] Example 3 (Precision mold parts machining scenario) This embodiment uses a VMC1270 CNC machining center to process a precision plastic mold cavity (mold steel material, material hardness 280±20HB). The processing step is grinding, with a spindle speed of 12000 r / min, a feed rate of 200 mm / min, and a machining accuracy requirement of ±3 micrometers (higher accuracy is required). The constant stability control system of this invention is installed on the CNC machining center. The sensor group is upgraded to: tool wear sensor (model: Kistler9258B, accuracy 0.0005mm), load sensor (model: HBMC18, accuracy 0.5N), and hardness sensor (model: WilsonVH1200, accuracy 0.5HB). The installation positions correspond to the grinding tool, grinding spindle, and mold steel blank surface. The parameters of the constant stability control unit are adjusted as follows: learning rate α=0.02, and machining error preset threshold ±0.005mm (±5 micrometers).

[0038] Implementation process 1. Start the CNC machining center and the constant stability control system. The multi-dimensional sensor group starts high-precision data acquisition at a frequency of 99Hz to collect the wear of grinding tools, the load fluctuation of the grinding spindle, and the hardness of the mold steel material in real time. 2. The constant stability control unit preprocesses the collected high-precision raw data, removes outlier data using the 3σ criterion, and then standardizes it using the Z-score standardization algorithm (in this embodiment, the average wear of the grinding tool is used). Standard deviation Average load fluctuation value Standard deviation Average material hardness Standard deviation ); 3. Construct a high-precision, multi-dimensional perception model based on fuzzy control algorithm, and adjust the membership function threshold: tool wear amount. Load fluctuation value Material hardness The fuzzy inference rules were optimized, and five new high-precision processing-specific rules were added to complete fuzzification, fuzzy inference, and defuzzification, thereby obtaining the adjustment amount of high-precision compensation parameters. 4. Establish a correlation model adapted to die steel grinding, and adjust the calculation formula as follows: (in, The measurement is for machining error, in micrometers. This refers to the wear of the grinding tool, in mm. This represents load fluctuation values, in N. Material hardness, unit HB; (micrometers, calibrated through high-precision grinding experiments on mold steel). 5. The constant stability control unit calculates the current machining error through an association model and judges the changes in working conditions and error thresholds: when the wear of the grinding tool increases from 0.02mm to 0.05mm (change of 150%), the load fluctuation increases from 680N to 750N (change of 10.3%), the material hardness increases from 270HB to 300HB (change of 11.1%), and the machining error reaches 6 micrometers, exceeding the preset threshold of ±5 micrometers, the constant stability control unit automatically adjusts the compensation parameters, adjusting the compensation amount from +5% to +15%. 6. After 60 hours of continuous processing, the accuracy of the compensation model began to decay, with the prediction error reaching 1.6 micrometers for 10 consecutive times. The constant stability control unit then initiated online meta-learning model evolution, updating the formula according to the parameters. After updating the model parameters, the prediction error was reduced to 1.1 micrometers, and the accuracy fluctuation was controlled within ±1.5 micrometers. 7. After processing, the cavities of 80 precision plastic molds were inspected for accuracy, with a focus on surface roughness and dimensional accuracy. Processing errors and scrap rates were also statistically analyzed.

[0039] The implementation results of this embodiment are as follows: 1. Compensation accuracy: The grinding process error is controlled within ±0.4 micrometers, and the compensation accuracy reaches 98.8%, exceeding the preset 95%, which meets the high-precision machining requirements of ±3 micrometers for precision mold cavities; 2. Precision fluctuation: During long-cycle (60 hours) operation, the precision fluctuation is controlled within ±1.1 micrometers, which is lower than the preset ±1.5 micrometers, ensuring the dimensional consistency of precision mold processing; 3. Scrap rate: Out of 80 processed parts, the number of scraps was 0, and the scrap rate was reduced to 0%, which completely solved the scrap problem caused by the accumulation of precision mold processing errors in the existing technology and improved the pass rate of mold processing; 4. Adaptability: The compensation range is stable at ±20%, which can effectively adapt to the harsh working conditions of high mold steel hardness, fast grinding tool wear, and high-precision machining. The automatic adjustment of compensation parameters reduces the manual dependence of precision mold machining and improves machining efficiency.

[0040] Examples 1, 2, and 3 further demonstrate that the constant stability control system for CNC machining centers of the present invention can be adapted to machining scenarios with different materials (aluminum alloys, cast iron, mold steel), different machining processes (milling, boring, grinding), and different precision requirements. It can achieve dynamic self-tuning of compensation parameters and self-evolution of the model, significantly improving machining stability and precision, reducing scrap rate, and has broad practicality and promotional value.

[0041] The technical scope of this invention is not limited to the content described above. Those skilled in the art can make various modifications and variations to the above embodiments without departing from the technical concept of this invention, and all such modifications and variations should fall within the protection scope of this invention.

Claims

1. A constant stability control system for a CNC machining center, characterized in that, include: The system comprises a constant-stability control unit, a multi-dimensional sensor group, a data storage module, and a compensation execution module. The constant-stability control unit serves as the primary execution unit, and its control method includes the following steps: Step 1: The constant stability control unit collects tool wear data, load fluctuation data, and material hardness data in real time during the CNC machining process through a preset multi-dimensional sensor group, with a collection frequency of 99Hz; Step 2: The constant stability control unit preprocesses the collected tool wear data, load fluctuation data, and material hardness data, and removes abnormal data; A multi-dimensional perception model is constructed based on fuzzy control algorithm, and features are extracted and fused from the preprocessed multi-dimensional working condition data; a correlation modeling logic between tool wear and machining error is established to realize the mapping between working condition data and machining error; Step 3: Based on the processed operating data and the associated model, the constant stability control unit determines whether the current processing conditions have changed and whether the processing error exceeds the preset threshold; at the same time, it determines whether the accuracy of the compensation model has decreased. Step 4: Based on the judgment results of Step 3, the constant stability control unit automatically adjusts the compensation parameters to achieve a compensation range of ±20% and a compensation accuracy of 95%. Online meta-learning technology is introduced to update the parameters of the compensation model, realize the self-evolution of the compensation model, and control the accuracy fluctuation during long-term operation within ±1.5 micrometers.

2. The constant stability control system for CNC machining centers according to claim 1, characterized in that, The data preprocessing includes data noise reduction and data standardization, wherein the data standardization uses the Z-score standardization algorithm, and the calculation formula is as follows: in, For standardized data, This is the original collected data. The mean of the original data. This represents the standard deviation of the original data.

3. The constant stability control system for CNC machining centers according to claim 1, characterized in that, The multi-dimensional perception model constructed by the fuzzy control algorithm includes three stages: fuzzification, fuzzy inference, and defuzzification. Fuzzification employs a triangular membership function and tool wear measurement. Load fluctuation value Material hardness The membership functions are as follows: Tool wear membership function: Membership function for load fluctuation value: Material hardness membership function: in, These are three threshold values ​​for tool wear. These are three thresholds for load fluctuation values. These are three threshold values ​​for material hardness.

4. The constant stability control system for CNC machining centers according to claim 1, characterized in that, The correlation modeling logic between tool wear and machining error adopts a linear regression model, and the calculation formula is as follows: in, Due to machining error, These are the error influence coefficients for tool wear, load fluctuation, and material hardness, respectively. This is the error correction term, with a value range of ±0.1 micrometers; This refers to the amount of tool wear. is the load fluctuation value, and h is the material hardness.

5. The constant stability control system for CNC machining centers according to claim 1, characterized in that, The online meta-learning technique enables the self-evolution of the compensation model, and uses gradient descent to update the model parameters. The parameter update formula is as follows: in, For the updated compensation model parameters, These are the parameters of the current compensation model. The learning rate is 0.01 to 0.

05. The gradient of the loss function under the current parameters, the loss function , This represents the actual machining error. To predict processing errors in the model, This represents the number of samples.

6. The constant stability control system for CNC machining centers according to claim 1, characterized in that, The preset threshold for processing error is ±0.01mm. When the processing error exceeds this threshold, the constant stability control unit immediately initiates the compensation parameter adjustment process.

7. The constant stability control system for CNC machining centers according to claim 1, characterized in that, The multi-dimensional sensor group includes a tool wear sensor, a load sensor, and a hardness sensor, which respectively collect the tool wear amount, spindle load fluctuation value, and hardness value of the processed material.

8. The constant stability control system for CNC machining centers according to claim 1, characterized in that, The constant stability control unit is used to execute the entire process of data acquisition, data processing, data judgment, compensation parameter adjustment and model self-evolution. The constant stability control unit has a built-in fuzzy control algorithm and an online meta-learning model. The multi-dimensional sensor group includes a tool wear sensor, a load sensor, and a hardness sensor. The tool wear sensor, load sensor, and hardness sensor are respectively installed at the tool chuck, spindle, and worktable of the CNC machining center to collect tool wear amount, spindle load fluctuation value, and aluminum alloy material hardness value. The data storage module is used to store the collected working condition data, processed data, compensation parameters, model parameters and processing error data, which facilitates subsequent data traceability and model optimization. The compensation execution module is linked with the servo system of the CNC machining center, receives the compensation parameters output by the constant stability control unit, and drives the servo motor to adjust the feed speed and spindle speed to achieve real-time compensation of machining errors.

Images