Thermal error intelligent compensation control system for optimizing BP neural network based on genetic algorithm

Abstract

The invention discloses a thermal error intelligent compensation control system for optimizing a BP neural network based on a genetic algorithm. The thermal error intelligent compensation control system comprises seven modules which are connected in sequence. The temperature field monitoring module starts a key part temperature sensor to collect data; the processing state import module imports data such as processing parameters; the thermal error analysis module analyzes a thermal deformation rule, and optimizes a BP neural network by using a genetic algorithm to establish a prediction model; the compensation demand judgment module judges a compensation demand; the compensation strategy generation module generates a strategy to determine the compensation amount; the compensation execution terminal sends the compensation amount to realize real-time position compensation; and the compensation effect evaluation terminal analyzes the effect and feeds back optimization. The system can improve the prediction precision and generalization ability, dynamically adjusts the compensation amount, feeds back and optimizes the compensation amount, achieves high-precision and high-adaptability thermal error compensation, and improves the machining quality and efficiency of a machine tool.

Description

Technical Field

[0001] The present invention belongs to the technical field of intelligent manufacturing and precision control, and specifically is a thermal error intelligent compensation control system based on genetic algorithm optimized BP neural network. Background Art

[0002] In modern manufacturing, machine tools serve as core processing equipment, and their machining accuracy directly determines product quality and performance. However, during the machining process, various error sources can affect the stability of machining accuracy, with thermal error being a particularly prominent issue. Heavy-duty machine tools, for example, suffer from severe component heat generation due to their numerous heat sources, complex structures, and high-power drive systems. These components are significantly affected by ambient temperature and heat generated during the machining process. Studies have shown that thermal errors in heavy-duty machine tools account for 40%-70% of total machining errors, and in ultra-high-precision machining, this proportion can reach as high as 80%. Thermal errors not only reduce machine tool manufacturing accuracy but also severely impact workpiece machining quality and production efficiency. While improvements in machine tool manufacturing technology and structural design have led to better control of traditional error sources such as geometric errors, thermal errors have long been a major obstacle to improving machining accuracy and are difficult to effectively suppress using traditional methods.

[0003] Thermal error compensation technology is considered a key means of improving machine tool machining accuracy. Its core lies in establishing a prediction model that accurately reflects thermal errors caused by thermal deformation. Based on the model's prediction results, real-time compensation data is sent to the numerical control system, thereby achieving dynamic compensation of thermal errors. Therefore, the quality of the thermal error prediction model directly determines the effectiveness of the compensation. Currently, commonly used thermal error modeling methods include multivariate linear regression, neural network, Bayesian network, and grey system modeling. However, these methods all have limitations. For example, multivariate linear regression has limited ability to describe complex nonlinear thermal error relationships. Although simple BP neural network modeling possesses self-learning capabilities, it is prone to falling into local minima during training and suffers from overlearning and overreliance on empirical parameters, resulting in poor generalization. Bayesian network modeling struggles to effectively reflect the nonlinear characteristics of thermal errors, resulting in low prediction accuracy and poor robustness. Grey system modeling, however, suffers from significant model variability due to differences in the selection of raw data sequences, and its robustness and adaptability need to be further improved.

[0004] In summary, the existing thermal error modeling methods still have shortcomings in dealing with complex nonlinear relationships, ensuring prediction accuracy and improving model robustness. A more efficient and accurate modeling method is urgently needed to solve the above problems, thereby providing reliable technical support for the thermal error intelligent compensation control system. Summary of the Invention

[0005] The purpose of this invention is to provide an intelligent thermal error compensation control system based on a genetic algorithm-optimized BP neural network. This system improves prediction accuracy and generalization capabilities. By dynamically adjusting the compensation amount and optimizing the compensation effect through feedback, it achieves high-precision and highly adaptable thermal error compensation, effectively improving machine tool processing quality and production efficiency.

[0006] The technical solution of the present invention is a thermal error intelligent compensation control system based on a genetic algorithm optimized BP neural network. The system includes a temperature field monitoring module, a processing state import module, a thermal error analysis module, a compensation demand judgment module, a compensation strategy generation module, a compensation execution terminal, and a compensation effect evaluation terminal connected in sequence:

[0007] The temperature field monitoring module is used to start the temperature sensors arranged at key parts of the machine tool to collect data and record temperature distribution data;

[0008] The processing status import module is used to import the current machine tool processing parameters, workpiece material characteristics, cutting force information and environmental temperature and humidity data;

[0009] The thermal error analysis module is used to analyze the thermal deformation law of the machine tool, obtain the thermal error distribution information, and establish a thermal error prediction model by optimizing the BP neural network through genetic algorithm;

[0010] The compensation requirement judgment module is used to determine the thermal error compensation requirement based on the thermal error distribution information and processing status data;

[0011] The compensation strategy generation module is used to generate an optimal compensation strategy and determine the compensation amount when the judgment result is that compensation is needed;

[0012] The compensation execution terminal is used to send the compensation amount to the CNC system and drive the servo system to perform real-time position compensation;

[0013] The compensation effect evaluation terminal is used to re-collect temperature distribution data and processing status data after compensation is completed, analyze the degree of compensation effect, and perform feedback optimization.

[0014] In the above-mentioned thermal error intelligent compensation control system based on genetic algorithm optimized BP neural network, in the temperature field monitoring module, temperature sensors are respectively arranged at key positions of the spindle box, feed shaft guide rail and bed. The temperature sensors use highly sensitive thermocouples or thermistors, and their measurement range covers the typical temperature range of the machine tool working environment.

[0015] The aforementioned thermal error intelligent compensation control system based on genetic algorithm optimized BP neural network, the processing state import module transmits processing parameters, workpiece material properties, cutting force information and ambient temperature and humidity data through the interface module of the CNC system, wherein the processing parameters include spindle speed, feed speed and cutting depth, and the workpiece material properties include thermal expansion coefficient, hardness and thermal conductivity.

[0016] In the aforementioned thermal error intelligent compensation control system based on genetic algorithm optimized BP neural network, the thermal error analysis module preprocesses the input data, including denoising, normalization, and feature extraction operations, and uses the BP neural network to construct an initial thermal error prediction model; the number of hidden layer nodes in the initial thermal error prediction model is determined according to the following formula:

[0017]

[0018] Where: N h is the hidden layer node, N i is the number of input layer nodes, N o is the number of nodes in the output layer, and a is an adjustment parameter, ranging from 1 to 10.

[0019] The aforementioned thermal error intelligent compensation control system based on genetic algorithm optimization of BP neural network, the genetic algorithm optimizes the weights and biases of the BP neural network, and the fitness function is defined as:

[0020]

[0021] Where: MSE represents mean square error;

[0022] Finally, a set of optimized weight coefficients w is output i , combined with the activation function of the BP neural network output g(x i ) to form a thermal error prediction model

[0023] In the aforementioned thermal error intelligent compensation control system based on genetic algorithm optimized BP neural network, the compensation demand judgment module calculates the difference between the current thermal error value and the preset threshold. If the difference exceeds the set range, it is determined that compensation is required, and the compensation threshold is dynamically adjusted by comprehensively considering the processing accuracy requirements and the workpiece material characteristics.

[0024] The aforementioned thermal error intelligent compensation control system based on genetic algorithm optimized BP neural network, the compensation strategy generation module calculates the theoretical compensation amount according to the thermal error prediction model, and corrects the compensation amount in combination with the kinematic characteristics of the machine tool, supporting the selection of single-point compensation, regional compensation and global compensation modes.

[0025] The aforementioned thermal error intelligent compensation control system based on genetic algorithm optimized BP neural network, the compensation execution terminal receives the compensation instruction through the interface module of the numerical control system, and converts it into a specific motion control signal to drive the servo motor to complete the position adjustment, and at the same time has a real-time monitoring function to detect abnormal conditions during the compensation process.

[0026] In the aforementioned thermal error intelligent compensation control system based on genetic algorithm optimized BP neural network, the compensation effect evaluation terminal re-collects temperature distribution data and processing status data, calculates the change in thermal error before and after compensation, and if the change is less than the set threshold, it is determined that the compensation effect meets the standard, otherwise it enters the feedback optimization link.

[0027] The aforementioned thermal error intelligent compensation control system based on genetic algorithm optimization of BP neural network, the feedback optimization link includes readjusting the parameters of the thermal error prediction model, optimizing the algorithm logic of the compensation strategy generation module and updating the control parameters of the compensation execution terminal, supporting the storage and analysis functions of historical data, and providing a reference basis for subsequent compensation strategies.

[0028] Compared with the prior art, the present invention has the following beneficial effects:

[0029] 1. The genetic algorithm optimizes the weights and biases of the BP neural network through global search, avoiding the defects of traditional BP neural networks that are prone to local minima, and significantly improving the model's ability to fit complex nonlinear thermal error relationships. The hidden layer nodes of the present invention are dynamically determined, combined with the parameter optimization of the genetic algorithm, to enhance the model's adaptability to different machine tool structures and processing scenarios.

[0030] 2. The compensation demand judgment module of the present invention makes a dynamic decision based on the difference between the thermal error value and the preset threshold value, and adjusts the compensation threshold value in combination with the processing accuracy requirements and the workpiece material characteristics (such as thermal expansion coefficient, hardness, etc.) to avoid over-compensation or under-compensation. The present invention supports single-point, regional, and global multi-mode compensation selection, and can formulate refined compensation schemes for the thermal deformation characteristics of different parts of the machine tool (such as the spindle box, feed shaft guide rails and other key parts) to improve compensation efficiency. The compensation execution terminal of the present invention receives the compensation amount in real time through the CNC system interface, drives the servo system to complete the position adjustment, ensures that the compensation action is synchronized with the processing process, and reduces the lag error. The present invention corrects the theoretical compensation amount in combination with the kinematic characteristics of the machine tool, avoids the compensation deviation caused by mechanical structure limitations, and improves the physical feasibility of the compensation action.

[0031] 3. The compensation effect evaluation terminal of the present invention forms a "monitoring-compensation-evaluation-optimization" closed loop by comparing the change in thermal error before and after compensation (if the change is less than a set threshold, it is judged to be qualified). The feedback optimization process of the present invention covers model parameter adjustment, compensation strategy algorithm optimization, and control parameter updating. Combined with historical data storage and analysis functions, it continuously improves the system's adaptability to different processing environments and working conditions. Through precise modeling and dynamic compensation, the present invention can effectively reduce the impact of thermal deformation on processing accuracy and improve workpiece dimensional accuracy and surface quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 It is the overall structural diagram of the system.

[0033] Figure 2 It is a schematic diagram of the temperature field monitoring module.

[0034] Figure 3 This is a flow chart for building a thermal error prediction model.

[0035] Figure 4 It is a flow chart of compensation strategy generation and execution.

[0036] Figure 5 It is a schematic diagram of compensation effect evaluation and feedback optimization. DETAILED DESCRIPTION

[0037] The present invention will be further described below with reference to the accompanying drawings and examples, but they are not intended to limit the present invention.

[0038] Example: A thermal error intelligent compensation control system based on genetic algorithm optimized BP neural network, the overall structure of the system is as follows Figure 1 As shown, the system includes a temperature field monitoring module 1, a processing status import module 2, a thermal error analysis module 3, a compensation demand judgment module 4, a compensation strategy generation module 5, a compensation execution terminal 6, and a compensation effect evaluation terminal 7, which are connected in sequence. The following describes the specific implementation process of each module step by step according to the system operation process and combines it with actual application scenarios.

[0039] During the implementation process, the temperature field monitoring module 1 is first started, which collects real-time temperature data through temperature sensors installed at key parts of the machine tool. Figure 2As shown, temperature sensors are placed in key locations such as the spindle box, feed axis guides, and bed to ensure comprehensive monitoring of the machine tool's overall temperature distribution. The temperature sensors transmit temperature field information to the central processing unit via a data acquisition device, recording temperature variations over time and space. To improve the accuracy and reliability of data acquisition, the temperature sensors use highly sensitive thermocouples or thermistors, whose measurement range covers the typical temperature range of the machine tool's operating environment. Furthermore, the temperature field monitoring module features a self-calibration function, ensuring the accuracy of collected data through regular comparison and correction with a standard temperature source.

[0040] Machining Status Import Module 2 is responsible for importing the current machine tool's machining parameters, workpiece material properties, cutting force information, and ambient temperature and humidity data. This data is transmitted to the central processing unit via the CNC system's interface module. Machining parameters include spindle speed, feed rate, and depth of cut; workpiece material properties include key indicators such as thermal expansion coefficient, hardness, and thermal conductivity; cutting force information is collected in real time by the machine tool's built-in force sensor; and ambient temperature and humidity data is obtained through environmental monitoring equipment within the workshop. This multi-dimensional data provides comprehensive input conditions for subsequent thermal error analysis, helping to more accurately reflect the actual thermal deformation scenarios of the machine tool.

[0041] The thermal error analysis module is the core part of the system, and its main task is to analyze the thermal deformation law of the machine tool and establish a thermal error prediction model. Figure 3 As shown in Figure 1, this module first preprocesses the data provided by the temperature field monitoring module and the processing status import module, including operations such as denoising, normalization, and feature extraction to improve data quality. Subsequently, an initial thermal error prediction model is constructed using a BP neural network. The number of nodes in the BP neural network's input layer equals the dimension of the input variables, and the number of nodes in the output layer corresponds to the predicted value of the thermal error. In the network structure design, the number of hidden layer nodes is determined according to the empirical formula:

[0042]

[0043] Where: N h is the hidden layer node, N i is the number of input layer nodes, N o is the number of nodes in the output layer, and a is an adjustment parameter, ranging from 1 to 10.

[0044] In order to improve the generalization ability of the model, a genetic algorithm is used to optimize the weights and biases of the BP neural network. The optimization process of the genetic algorithm includes steps such as population initialization, fitness function calculation, selection, crossover, and mutation. Its core lies in searching for the optimal solution through iteration. The fitness function is defined as:

[0045]

[0046] Where: MSE represents the mean square error, which is used to measure the deviation between the predicted value and the actual value.

[0047] After multiple iterations, the genetic algorithm finally outputs a set of optimized weight coefficients w i , combined with the activation function of the BP neural network output g(x i ) to form a thermal error prediction model This model can more accurately describe the nonlinear characteristics of thermal errors, thus providing a scientific basis for the formulation of subsequent compensation strategies.

[0048] The compensation demand judgment module 4 determines whether thermal error compensation is necessary based on the thermal error distribution information and machining status data. Specifically, the module first calculates the difference between the current thermal error value and a preset threshold. If the difference exceeds the set range, compensation is determined to be necessary. For example, during a particular machining process, if the thermal error of the spindle box reaches 0.05 mm, and the system-set compensation threshold is 0.03 mm, a compensation demand is triggered. Furthermore, the compensation demand judgment module dynamically adjusts the compensation threshold, taking into account factors such as machining accuracy requirements and workpiece material properties, to ensure the rationality of the compensation strategy.

[0049] If compensation is determined to be necessary, the compensation strategy generation module 5 generates an optimal compensation strategy and determines the compensation amount. This module first calculates the theoretical compensation amount based on the thermal error prediction model and then modifies it based on the machine tool's kinematic characteristics. For example, for a specific machining task, the theoretical compensation amount might be 0.04 mm, but the actual compensation amount might be adjusted to 0.038 mm to account for the servo system's response delay and mechanical backlash. Furthermore, the compensation strategy generation module supports multiple compensation modes, including single-point compensation, regional compensation, and global compensation, allowing users to flexibly configure the compensation based on their specific needs. Among them, the single-point compensation is a local compensation for the thermal error of a single key part of the machine tool. For example, if the heat of the spindle box causes a unidirectional thermal error of the X-axis, the position compensation amount of the X-axis is adjusted separately through the single-point compensation mode (such as only -0.035mm compensation for the X-axis in the embodiment); the regional compensation is a collaborative compensation for the thermal errors of multiple adjacent parts or the same functional area of the machine tool. It is used when the thermal error impact range involves multiple related components (such as the guide rails, lead screws, bearings, etc. of the same coordinate axis), or when there are multiple heat sources in a certain processing area (such as the spindle box and feed axis linkage area). The global compensation is a systematic compensation for the thermal error of the overall structure of the machine tool. Its application scenario is when the entire body of the machine tool is affected by ambient temperature changes or the combined influence of multiple heat sources, resulting in thermal deformation of each coordinate axis or key parts (such as the overall temperature rise of a heavy machine tool after long-term continuous processing). The generated compensation strategy is sent to the CNC system through the compensation execution terminal to drive the servo system for real-time position compensation.

[0050] like Figure 4 As shown, the compensation execution terminal 6 receives compensation instructions through the interface module of the CNC system and converts them into specific motion control signals, driving the servo motor to complete position adjustment. During this process, the compensation execution terminal also has a real-time monitoring function, which can detect abnormalities during the compensation process and issue timely alarms.

[0051] After the compensation is completed, the compensation effect evaluation terminal 7 re-collects the temperature distribution data and processing status data, analyzes the degree of the compensation effect and performs feedback optimization. Figure 5 As shown in the figure, the compensation effect evaluation terminal first obtains the compensated data through the temperature field monitoring module and the processing status import module, and then calculates the change in thermal error before and after compensation. If the change is less than the set threshold, the compensation effect is determined to be up to standard; otherwise, it enters the feedback optimization link. The specific steps of feedback optimization include readjusting the parameters of the thermal error prediction model, optimizing the algorithm logic of the compensation strategy generation module, and updating the control parameters of the compensation execution terminal. For example, during a certain compensation process, if it is found that the thermal error is still beyond the allowable range after compensation, the crossover probability and mutation probability of the genetic algorithm are adjusted to further optimize the weight coefficient of the BP neural network, thereby improving the accuracy of the prediction model. In addition, the compensation effect evaluation terminal also supports the storage and analysis of historical data, providing a reliable reference for the optimization of subsequent compensation strategies.

[0052] In order to better enable relevant personnel in this technical field to fully understand and implement the present invention, the following is a specific implementation case applied to a heavy-duty CNC machine tool, combining the technical solution in the specification with actual scenarios to illustrate in detail the application process and effects of the present invention:

[0053] This embodiment is applied to the high-precision machining scenario of a heavy-duty CNC machine tool. To address the thermal deformation problem of the spindle box, feed axis guide rails, and bed caused by heat during long-term machining, this system realizes real-time compensation of thermal errors and improves the machining accuracy of the workpiece.

[0054] Implementation steps and system operation process:

[0055] 1. Operation of temperature field monitoring module 1:

[0056] Sensor arrangement: High-sensitivity thermocouples (measuring range: -50°C to 200°C, accuracy ±0.5°C) are installed at key heating locations such as the spindle box motor, feed shaft guide slider, and bed column to collect temperature data in real time (sampling frequency: 10Hz).

[0057] Data recording: The temperature distribution data of each measuring point (such as spindle box temperature 45°C, guide rail temperature 38°C, bed temperature 32°C) is transmitted to the system database through the distributed data acquisition module.

[0058] 2. Data integration of processing status import module 2:

[0059] Processing parameters: The current working condition data is obtained through the CNC system interface, including the spindle speed of 1500 r / min, the feed speed of 800 mm / min, and the cutting depth of 2 mm.

[0060] Workpiece material characteristics: The workpiece to be processed is 45# steel, with a thermal expansion coefficient of 11.5×10 -6 / ℃, hardness HB200, thermal conductivity 45W / (m·K).

[0061] Environmental data: The workshop temperature and humidity sensors provide real-time feedback of a temperature of 25°C and a humidity of 55% RH.

[0062] 3. Modeling process of thermal error analysis module 3

[0063] Data preprocessing: De-noising (sliding average filtering), normalization (0-1 standardization) and feature extraction (selecting features such as temperature gradient and spindle load that are strongly related to thermal deformation) of raw data such as temperature and processing parameters.

[0064] Initial BP neural network construction: number of input layer nodes N i =8 (3 temperature measurement points + 3 processing parameters + 2 material properties), the number of output layer nodes is N o =3 (X / Y / Z axis thermal error), the number of hidden layer nodes is based on the formula Calculate and adjust the parameter a=5 to get N h =8.

[0065] Genetic algorithm optimization:

[0066] Fitness function The weights and biases of the BP neural network are optimized through selection, crossover, and mutation operations, and the optimal weight matrix w is output after 50 generations of iteration. i , combined with the activation function of the BP neural network output g(x i ) to form a thermal error prediction model

[0067] Prediction results: X-axis thermal error +0.03mm, Y-axis -0.02mm, Z-axis +0.015mm.

[0068] 4. Decision of Compensation Demand Judgment Module 4

[0069] Threshold setting: According to the processing accuracy requirements (IT6 level, tolerance ±0.02mm), the preset thermal error compensation threshold is ±0.01mm.

[0070] Difference calculation: The current X-axis error of 0.03mm exceeds the threshold of +0.01mm (difference 0.02mm), and compensation is determined to be required. The Y-axis and Z-axis errors do not exceed the threshold, so compensation is not performed for the time being.

[0071] Dynamic adjustment: Due to the high thermal expansion coefficient of the workpiece material, the compensation threshold is temporarily adjusted to ±0.008mm to enhance the compensation sensitivity.

[0072] 5. Strategy formulation of compensation strategy generation module 5

[0073] Theoretical compensation calculation: According to the prediction model, the X-axis needs to be compensated by -0.03mm (to offset the positive error).

[0074] Kinematic correction: Considering the 0.005mm clearance of the machine tool X-axis screw transmission, the correction compensation amount is -0.035mm.

[0075] Mode selection: Select single-point compensation mode to solve the unidirectional error of X-axis caused by local heating of the spindle box.

[0076] 6. Real-time control of compensation execution terminal 6

[0077] Command transmission: Send the compensation command "X-0.035mm" to the servo drive through the RS-232 interface of the CNC system.

[0078] Servo adjustment: The X-axis servo motor drives the ball screw to move in the opposite direction by 0.035mm to complete position compensation. At the same time, the motor current and displacement feedback values are monitored in real time to ensure that there are no abnormal alarms.

[0079] 7. Closed-loop verification of compensation effect evaluation terminal 7

[0080] Data re-collection: After compensation, the temperature (spindle box 46°C, guideway 38°C) and processing parameters (remain unchanged) are collected again.

[0081] Error comparison: Calculate the change in X-axis thermal error: +0.03mm before compensation → +0.005mm after compensation (change 0.025mm). This is less than the set threshold of 0.008mm, indicating that the compensation meets the standard.

[0082] Feedback optimization: The compensation data (temperature, compensation amount, effect) of this time is stored in the historical database to provide a modeling reference for subsequent similar working conditions.

[0083] Implementation effect:

[0084] Improved precision: X-axis thermal error is reduced from 0.03mm to 0.005mm, compensation efficiency reaches 83%, and workpiece size tolerance is controlled within ±0.01mm, meeting IT6 level precision requirements.

[0085] Enhanced stability: After 2 hours of continuous processing, the thermal error fluctuation range is reduced to ±0.003mm, which is 50% lower than the error fluctuation of traditional compensation methods (such as simple BP neural network).

[0086] Production efficiency: The downtime for adjustment due to thermal errors is reduced by approximately 30%, and the scrap rate is reduced from 1.5% to 0.3%, significantly improving processing continuity and economy.

[0087] It can be seen that this embodiment demonstrates the actual application capability of the system in thermal error control of heavy machine tools through the complete process of "multi-source data acquisition-intelligent modeling-dynamic compensation-closed-loop verification", verifies the effectiveness of genetic algorithm optimization of BP neural network in complex nonlinear scenarios, and the effect of dynamic compensation strategy on improving processing accuracy.

Claims

1. The thermal error intelligent compensation control system based on genetic algorithm optimized BP neural network is characterized by: The system comprises a temperature field monitoring module (1), a processing state importing module (2), a thermal error analysis module (3), a compensation demand judgment module (4), a compensation strategy generating module (5), a compensation execution terminal (6) and a compensation effect evaluation terminal (7) which are connected in sequence; The temperature field monitoring module (1) is used to start the temperature sensors arranged at key parts of the machine tool to collect data and record temperature distribution data; The processing state import module (2) is used to import the processing parameters of the current machine tool, workpiece material characteristics, cutting force information and environmental temperature and humidity data; The thermal error analysis module (3) is used to analyze the thermal deformation law of the machine tool, obtain the thermal error distribution information, and establish a thermal error prediction model by optimizing the BP neural network through genetic algorithm; The compensation requirement judgment module (4) is used to judge the thermal error compensation requirement based on the thermal error distribution information and the processing state data; The compensation strategy generation module (5) is used to generate an optimal compensation strategy and determine the compensation amount when the judgment result is that compensation is required; The compensation execution terminal (6) is used to send the compensation amount to the numerical control system to drive the servo system to perform real-time position compensation; The compensation effect evaluation terminal (7) is used to recollect temperature distribution data and processing status data after the compensation is completed, analyze the degree of the compensation effect, and perform feedback optimization.

2. The thermal error intelligent compensation control system based on genetic algorithm optimized BP neural network according to claim 1 is characterized by: In the temperature field monitoring module (1), temperature sensors are respectively arranged at key positions of the spindle box, feed shaft guide rail and bed, and the temperature sensors adopt highly sensitive thermocouples or thermistors, and their measurement range covers the typical temperature range of the machine tool working environment.

3. The thermal error intelligent compensation control system based on genetic algorithm optimized BP neural network according to claim 1 is characterized in that: The processing state import module (2) transmits processing parameters, workpiece material properties, cutting force information and environmental temperature and humidity data through the interface module of the numerical control system, wherein the processing parameters include spindle speed, feed speed and cutting depth, and the workpiece material properties include thermal expansion coefficient, hardness and thermal conductivity.

4. The thermal error intelligent compensation control system based on genetic algorithm optimized BP neural network according to claim 1 is characterized in that: The thermal error analysis module (3) pre-processes the input data, including denoising, normalization and feature extraction operations, and constructs an initial thermal error prediction model using a BP neural network; the number of hidden layer nodes of the initial thermal error prediction model is determined according to the following formula: Where: N h is the hidden layer node, N i is the number of input layer nodes, N o is the number of nodes in the output layer, and a is an adjustment parameter, ranging from 1 to 10.

5. The thermal error intelligent compensation control system based on genetic algorithm optimized BP neural network according to claim 1 is characterized in that: The genetic algorithm optimizes the weights and biases of the BP neural network, and the fitness function is defined as: Where: MSE represents mean square error; Finally, a set of optimized weight coefficients w is output i , combined with the activation function of the BP neural network output g(x i ) to form a thermal error prediction model 6. The thermal error intelligent compensation control system based on genetic algorithm optimized BP neural network according to claim 1 is characterized in that: The compensation requirement judgment module (4) calculates the difference between the current thermal error value and a preset threshold value. If the difference exceeds a set range, it is determined that compensation is required, and the compensation threshold value is dynamically adjusted by comprehensively considering the machining accuracy requirements and the workpiece material characteristics.

7. The thermal error intelligent compensation control system based on genetic algorithm optimized BP neural network according to claim 1 is characterized in that: The compensation strategy generation module (5) calculates the theoretical compensation amount according to the thermal error prediction model, and corrects the compensation amount in combination with the kinematic characteristics of the machine tool, supporting the selection of single-point compensation, regional compensation and global compensation modes.

8. The thermal error intelligent compensation control system based on genetic algorithm optimized BP neural network according to claim 1 is characterized in that: The compensation execution terminal (6) receives the compensation instruction through the interface module of the numerical control system, converts it into a specific motion control signal, drives the servo motor to complete the position adjustment, and has a real-time monitoring function to detect abnormal conditions during the compensation process.

9. The thermal error intelligent compensation control system based on genetic algorithm optimized BP neural network according to claim 1, characterized in that: The compensation effect evaluation terminal (7) recollects temperature distribution data and processing status data, calculates the change in thermal error before and after compensation, and if the change is less than a set threshold, determines that the compensation effect meets the standard, otherwise enters the feedback optimization link.

10. The thermal error intelligent compensation control system based on genetic algorithm optimized BP neural network according to claim 1, characterized in that: The feedback optimization link includes readjusting the parameters of the thermal error prediction model, optimizing the algorithm logic of the compensation strategy generation module, and updating the control parameters of the compensation execution terminal. It supports the storage and analysis functions of historical data and provides a reference basis for subsequent compensation strategies.

Images