Space error compensation and precision improvement method for combined processing machine tool

Abstract

The invention discloses a space error compensation and precision improvement method for a combined machine tool. The method comprises the steps of 1, system preparation and sensor arrangement; 2, real-time data monitoring and processing; 3, training a dynamic geometric error compensation model; 4, multi-modal data fusion is carried out; 5, generating and executing an intelligent numerical control instruction; and step 6, a post-processing and feedback mechanism. Through real-time dynamic error compensation and multi-modal data fusion, real-time errors of the machine tool in the machining process can be effectively recognized and corrected. The high-precision compensation mechanism ensures that the geometrical shape and the size of a workpiece can strictly meet the design requirements in a complex machining environment, the overall quality of a product is greatly improved, a real-time monitoring system and a dynamic updating mechanism enable a machine tool to rapidly adapt to environment changes, and the machining efficiency is improved. In combination with deep learning and data fusion technologies, the system can optimize a compensation strategy in real time along with continuous changes of processing conditions.

Description

Technical Field

[0001] The present invention relates to the technical field of machining machine tools, and in particular to a method for compensating spatial errors and improving precision of a composite machining machine tool. Background Art

[0002] The spatial geometric error of a machine tool refers to the geometric error caused in three-dimensional coordinates during the machining process of a CNC machine tool. Therefore, error compensation is an effective measure to improve the accuracy of a machine tool, and the correct identification of the spatial motion error of a machine tool is the key to this technology. Common instruments for error identification of CNC machine tools include laser interferometers and ballbars. Laser interferometers can complete the work of error identification with high accuracy, but the measurement time is long. It usually takes several days to identify a machine tool, and the technical requirements for the operator are high. The ballbar instrument can perform error separation, but it can only derive the error within the current test position and test range.

[0003] However, the existing technology still has major deficiencies, such as:

[0004] Existing geometric error compensation methods usually rely on static data training and lack the ability to respond to dynamic environments and machine tool status changes in real time. This approach makes it difficult for compensation models to adapt to rapidly changing machining conditions in practical applications, resulting in unstable machining accuracy. In addition, existing technologies often rely on data from a single sensor and cannot fully and accurately reflect the overall condition of the machine tool. Such limitations greatly reduce the effectiveness of error compensation. Summary of the invention

[0005] The purpose of the present invention is to provide a method for compensating spatial errors and improving precision of a composite machining machine tool, so as to solve the problems raised in the above-mentioned background technology.

[0006] To achieve the above objectives, the present invention provides the following technical solutions:.

[0007] Compared with the prior art, the present invention has the following beneficial effects:

[0008] 1. Through real-time dynamic error compensation and multi-modal data fusion, this solution can effectively identify and correct the real-time errors of machine tools during processing. This high-precision compensation mechanism ensures that the geometric shape and size of the workpiece can strictly meet the design requirements in a complex processing environment, greatly improving the overall quality of the product;

[0009] 2. The real-time monitoring system and dynamic update mechanism enable the machine tool to quickly adapt to environmental changes, reducing downtime, rework and material waste caused by processing errors. This efficient error compensation not only shortens the processing cycle, but also reduces production costs and improves the economic benefits of the enterprise;

[0010] 3. Combining deep learning and data fusion technology, the compensation model has the ability to self-learn and adjust. As the processing conditions continue to change, the system can optimize the compensation strategy in real time. This intelligent feature enables the machine tool to maintain high performance in different working environments, further improving its adaptability and flexibility in various application scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] Figure 1 It is a schematic diagram of the step flow structure of the present invention. DETAILED DESCRIPTION

[0012] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0013] See also Figure 1 , the present invention provides a technical solution:

[0014] Step 1: System preparation and sensor placement

[0015] Select sensor:

[0016] Select high-precision laser knives, industrial cameras, force sensors, temperature sensors, vibration sensors, etc. to ensure the comprehensiveness and accuracy of the data.

[0017] Select a specific sensor model, such as:

[0018] Laser knife: select the laser knife model "RoboLaser 3000", industrial camera: select "HighResCam5000", the resolution can reach 5MP, force sensor: select "ForceSense 200", the range is 0-100N, temperature sensor: select "TempGuard 100", the accuracy is ±0.5°C, vibration sensor: select "VibeSense 400", which can monitor the frequency range of 0-1000Hz. In this embodiment, the selection of a preferred specific model for the corresponding sensor model can be selected according to actual needs and circumstances, and the use of each sensor is well known to those skilled in the art, so it will not be elaborated here.

[0019] Installation and calibration

[0020] Install the laser cutter and industrial camera on the CNC machine tool and calibrate them to ensure their measurement accuracy. The force sensor should be installed in the key parts of the machine tool, and the temperature and vibration sensors should be distributed in different areas of the machine tool.

[0021] The laser cutter is installed on the spindle of the machine tool, the industrial camera is fixed on the side of the machine tool to ensure that the processing area can be clearly captured, the force sensor is installed on the part where the tool contacts the workpiece, the temperature sensor is set outside and inside the machine tool to monitor the environment and machine tool temperature, and the vibration sensor is installed on the base of the machine tool;

[0022] The laser cutter is located at a suitable position on the machine tool spindle, usually at the center of the spindle, to ensure that the laser cutter is consistent with the vertical direction of the workpiece. A special fixture and bolts are used to fix the laser cutter to ensure that it does not move during processing. The power cord and control line of the laser cutter are connected to the control system of the machine tool.

[0023] Industrial camera, installed on the side of the machine tool, make sure the lens is aimed at the processing area, the height should be adjusted to clearly capture the entire processing process, use a shockproof mount to fix the camera, and consider using a dust cover to protect the lens, and connect the camera's data cable to the data acquisition module;

[0024] Place the force sensor where the tool contacts the workpiece, usually at the bottom of the tool or on the clamping device. Use a connector to connect the force sensor to the machine tool control system to ensure that its readings can be transmitted in real time. Record the zero point of the sensor without load, and then calibrate it by applying a known force to ensure the accuracy of the readings.

[0025] Install temperature sensors inside and outside the machine tool. The external one is used to detect the ambient temperature. Use insulating materials to protect the sensor. Use shielding covers made of metal or other materials to reduce the sensor's exposure to radiant heat or electromagnetic interference. Set thresholds, monitor sensor data in real time, and identify and eliminate abnormal readings affected by external factors.

[0026] Mount vibration sensors at strategic locations on the machine tool base or frame to obtain the most accurate vibration data.

[0027] Data acquisition module construction:

[0028] Build a data acquisition module to collect data from each sensor in real time and connect it to the machine tool control system through a high-frequency data transmission protocol.

[0029] “National Instruments” NI-USB-6212 was selected as the data acquisition module to connect all sensors.

[0030] LabVIEW software is used for data acquisition and processing. The software, acquisition modules and equipment involved here are well known to those skilled in the art.

[0031] Step 2: Real-time data monitoring and processing

[0032] Real-time data collection

[0033] Start the data acquisition module to monitor the machine tool's motion status, physical environment, and forces and vibrations during the machining process in real time.

[0034] The acquisition frequency is set to 1000 Hz to ensure real-time acquisition of sensor data.

[0035] Ensure that the selected temperature sensor has suitable sensitivity, response time and measurement range. Reasonably arrange the sensors according to process and environmental conditions to ensure that they can accurately capture the area that needs to be monitored. Pre-process the data and set the sampling frequency (such as every second, every minute, etc.) according to the application requirements to ensure the timeliness of the data. In the case of drastic environmental changes, increase the acquisition frequency to increase sensitivity to temperature fluctuations.

[0036] Through high-frequency real-time data acquisition, it is possible to obtain machine tool status information in a timely manner during the processing, including key parameters such as position, speed, force, temperature and vibration. These data provide the basis for subsequent status evaluation and error compensation.

[0037] Data preprocessing:

[0038] Filter and denoise the data collected in real time to ensure data reliability. Use filters to remove sensor noise.

[0039] The collected data is denoised using a Kalman filter.

[0040] In LabVIEW, filter algorithms are written to process the data of force sensors and vibration sensors in real time, remove instantaneous noise, and calibrate the data of temperature sensors to eliminate deviations caused by environmental changes. The purpose of data preprocessing is to improve data quality and ensure the accuracy of subsequent analysis. Through denoising and interpolation processing, errors caused by environmental interference can be effectively reduced, ensuring the accuracy of model training and dynamic compensation.

[0041] Status Assessment

[0042] Based on the collected data, the machine tool's motion status is evaluated in real time, any abnormalities are detected, and this information is passed on to subsequent steps.

[0043] Evaluate the status of the machine tool based on real-time data, set thresholds (for example, force sensor values ​​exceeding 80N are considered abnormal). If the value read by the force sensor exceeds 80N within 1 second, an alarm is triggered and the abnormal event is recorded.

[0044] Real-time data transmission: Select a suitable communication protocol such as MQTT, HTTP, for data transmission, and format the collected temperature data into a consistent data structure.

[0045] Real-time data monitoring: Use the stream processing platform to receive temperature data streams in real time, correct the temperature data according to environmental factors (such as humidity and air pressure), improve the accuracy of the data, set alarm thresholds, monitor in real time whether the temperature exceeds the upper and lower limits, and trigger an alarm when it exceeds the limit. Status assessment is a key link in real-time monitoring. By judging the current status of the machine tool, potential abnormalities can be discovered in time. This provides a real-time data basis for the dynamic compensation model, ensuring that corresponding compensation and adjustments can be made according to actual conditions during the processing process.

[0046] Step 3: Dynamic geometric error compensation model training

[0047] Generate training dataset:

[0048] Using a laser knife and an industrial camera, marks (such as laser patterns) are drawn on prefabricated workpieces according to preset CNC instructions to generate training workpieces.

[0049] Use a laser knife to generate laser patterns according to preset instructions on the prefabricated workpiece.

[0050] Set the processing trajectory and use the CNC system to issue instructions to make the laser knife draw a specific pattern on the workpiece. During the processing, specific marks such as laser patterns or geometric shapes are drawn on the workpiece according to the preset CNC instructions. These marks will be used for subsequent image analysis to ensure that multiple training samples are generated under different processing conditions such as different speeds, temperatures, materials, etc. to increase the diversity of the data.

[0051] The processing parameters and status information of each sample, such as cutting force, temperature, vibration, etc., are recorded and stored together with the labeled images to form a training data set. Generating a training data set is the basis for building a dynamic geometric error compensation model. Through the data collected under different conditions, the model can more comprehensively understand the types and causes of errors that may occur in the actual processing of machine tools.

[0052] Image capture and correction

[0053] The images of the training workpieces are captured using a camera device and spatially calibrated according to the calibration parameters.

[0054] Use an industrial camera to capture images of training workpieces and perform calibration.

[0055] The image processing module is called in LabVIEW, and the camera calibration parameters are applied to perform distortion correction on the captured image to display the true shape of the laser pattern.

[0056] Use an industrial camera to shoot the processed workpiece and capture the pattern image with the mark. These images should be acquired under the same lighting conditions to ensure consistency. The captured images are processed, including denoising, edge detection and geometric correction, to ensure the accuracy and standardization of the images. The processed images are compared with the theoretical processing trajectory, and the error features are extracted and quantified. The purpose of image capture and correction is to ensure high-quality image data for subsequent analysis. Through accurate image processing, any errors and deviations on the workpiece can be clearly identified, thereby providing key feedback data for the error compensation model.

[0057] Deep learning model training

[0058] The corrected image is paired with the standard image of the theoretical machining trajectory as training data to train the deep learning model and generate a dynamic geometric error compensation model.

[0059] The processed image data and the corresponding theoretical processing trajectory are used as input to build a deep learning model, such as a convolutional neural network (CNN). During the training process, data enhancement techniques such as rotation, scaling, and adding noise are used to increase the diversity of training samples and improve the robustness of the model. Loss functions such as mean square error are set to evaluate the gap between the model output and the actual result, and the model parameters are adjusted through the back propagation algorithm to continuously optimize the performance of the model. Deep learning model training is the core of dynamic geometric error compensation. By learning the mapping relationship from input data to compensation output, the model can automatically identify and compensate for errors in the processing process. This makes error compensation more intelligent and adaptive, and can adapt to different processing conditions and workpiece characteristics.

[0060] Step 4: Multimodal data fusion

[0061] Data fusion algorithm design

[0062] A multimodal fusion algorithm is designed to fuse the data from the laser cutter, camera, force sensor, temperature sensor and vibration sensor to form a comprehensive machine tool status information set.

[0063] Select appropriate data fusion algorithms, such as weighted average method, Kalman filter, deep learning fusion model, etc., determine the fusion method according to actual needs, write the implementation code of the data fusion algorithm, ensure that it can process data from different sensors such as laser knives, industrial cameras, force sensors, temperature sensors, and vibration sensors, define the input and output formats of the fusion algorithm, and ensure that data of different dimensions can be processed uniformly. The purpose of data fusion algorithm design is to effectively integrate data from multiple sensors to obtain more comprehensive and accurate machine tool status information. By fusing data from different sources, the error impact of a single sensor can be reduced, the reliability and accuracy of the data can be improved, and the model can better understand the actual working status of the machine tool.

[0064] Write a data fusion algorithm to merge the data from various sensors.

[0065] The weighted average method is implemented in Python to combine the data from temperature, force and vibration sensors to obtain a set of comprehensive status information.

[0066] Fusion model training

[0067] The fused data is used to train a comprehensive model to improve the responsiveness to machine state changes and the accuracy of error compensation.

[0068] Use the integrated data set containing real-time data from different sensors as training samples, build multimodal fusion models such as multi-input neural networks, design the network structure to ensure that it can process data inputs of different modalities, and define the loss function to evaluate the model performance. Apply data enhancement and training techniques such as dropout and batch normalization to improve the robustness and generalization ability of the model, and perform model training.

[0069] The core of fusion model training is to enable the model to learn how to extract useful information from multiple data sources to achieve more efficient state assessment and error compensation. Through a multimodal perspective, the model can capture complex patterns that a single sensor cannot recognize, providing comprehensive information support for dynamic compensation.

[0070] Use the fused data for model training to ensure a comprehensive assessment of the machine tool status.

[0071] A comprehensive evaluation model is generated based on the data from different sensors using the random forest model, and the training dataset contains 4,000 records.

[0072] Real-time update mechanism

[0073] Whenever the system detects a state change, the compensation model is updated in real time to ensure that the compensation instructions reflect the current processing conditions.

[0074] A real-time data update mechanism is set up. When sensor data changes, such as temperature changes or increased vibration, the system can automatically trigger a model update. An incremental learning mechanism is designed to allow the model to make adaptive adjustments after receiving new data instead of completely retraining, ensuring a quick response. The update history and model performance are recorded for subsequent analysis and optimization.

[0075] The real-time update mechanism ensures that the model can adapt to changes in the machine tool status in a timely manner and maintain efficient error compensation. Through incremental learning, the model can not only continue to improve, but also remain flexible and responsive in practical applications. This step significantly improves the intelligence level of the system, enabling it to cope with complex and changing machining environments, and automatically triggers model updates every time the status changes.

[0076] A trigger mechanism is set in the data acquisition module to automatically start the model retraining process when the force sensor data exceeds the set threshold.

[0077] Step 5: Intelligent CNC instruction generation and execution

[0078] Compensation instruction generation

[0079] Based on the dynamic geometric error compensation model and multimodal data fusion results, real-time compensation instructions are generated.

[0080] According to the processing requirements, workpiece characteristics and current machine tool status, an intelligent NC instruction generation algorithm is designed. The algorithm needs to consider the processing path, cutting parameters such as cutting speed, feed rate and compensation strategy, use optimization techniques such as genetic algorithm and particle swarm optimization to generate the optimal processing instructions, ensure efficient and accurate processing of workpieces, write algorithm code, ensure that the output data from the fusion model can be received in real time, and generate corresponding NC instructions based on these data.

[0081] The purpose of the instruction generation algorithm design is to combine real-time data with processing requirements to generate CNC instructions suitable for the current state. With the support of intelligent algorithms, the optimal instructions can be quickly generated under complex conditions, thereby improving processing efficiency and accuracy while reducing the need for manual intervention.

[0082] If the model output error is 0.02mm, a compensation instruction is generated and the tool path is adjusted in real time.

[0083] CNC instruction optimization

[0084] Based on the dynamic geometric error compensation model and multimodal data fusion results, real-time compensation instructions are generated.

[0085] During the processing, the machine tool status such as temperature, vibration, cutting force, etc. is monitored in real time and compared with the currently generated instructions. If the monitored status parameters exceed the set threshold or an abnormal situation occurs, the system can automatically adjust the current CNC instructions, such as reducing the feed speed or adjusting the cutting path. Through the closed-loop feedback mechanism, it is ensured that the instruction adjustment can be reflected in the processing process in a timely manner, and the corresponding data is recorded.

[0086] Real-time adjustment of instructions ensures the safety and stability of the machining process. When the machine tool status changes, timely adjustment of instructions can effectively prevent potential machining problems, improve the quality of the finished product, and reduce the risk of machine tool damage. This step enhances the system's adaptive ability, enabling it to maintain effective machining in complex environments.

[0087] According to the compensated data, the CNC instructions are optimized.

[0088] Machine tool execution instructions

[0089] The optimized CNC instructions are sent to the machine tool control system to start workpiece processing.

[0090] The generated CNC instructions are sent to the machine tool control system to start executing the processing task, and ensure that the execution process of the instructions meets the set parameters. During the processing, the execution results are recorded in real time, including processing time, finished product size, surface quality and other data. The comparison data between the execution results and the theoretical model predictions are collected, the deviations are analyzed and fed back to the system for subsequent instruction adjustments and model optimization.

[0091] Step 6: Post-processing and feedback mechanism

[0092] Processing result evaluation

[0093] After processing is completed, the workpiece is inspected using a laser cutter or camera to evaluate the accuracy of the processing results.

[0094] After processing is completed, use a laser cutter or camera to inspect the workpiece to evaluate the processing accuracy, use a laser scanner to measure the size of the workpiece, obtain the actual processing results, and ensure that they are consistent with the expected size.

[0095] Feedback adjustment mechanism

[0096] The evaluation results are fed back into the algorithm to identify areas that do not meet the accuracy requirements and guide subsequent model optimization and compensation instruction adjustments.

[0097] The evaluation results are fed back to the compensation model for adjustment. If the actual result deviates from the standard value by more than 0.01 mm, the model parameters are recorded and adjusted for retraining.

[0098] Continuous Improvement

[0099] Establish a regular review mechanism to continuously monitor machine tool performance and machining accuracy to improve data collection and compensation models.

[0100] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for compensating spatial errors and improving precision of composite machining machine tools. It is characterized in that The steps include: Step 1: System preparation and sensor placement; Step 2: Real-time data monitoring and processing; Step 3: Dynamic geometric error compensation model training; Step 4: Multimodal data fusion; Step 5: Intelligent CNC instruction generation and execution; Step 6: Post-processing and feedback mechanism.

2. A method for compensating spatial errors and improving precision of a composite machining machine tool according to claim 1, Features: The step 1: system preparation and sensor arrangement, specifically includes the following steps: Select sensor: Select high-precision laser cutters, industrial cameras, force sensors, temperature sensors, vibration sensors, etc. to ensure the comprehensiveness and accuracy of data; Installation and calibration: Install the laser cutter and industrial camera on the CNC machine tool and calibrate them to ensure their measurement accuracy. The force sensor should be installed in the key parts of the machine tool, and the temperature and vibration sensors should be distributed in different areas of the machine tool; Data acquisition module construction: Build a data acquisition module to collect data from each sensor in real time and connect it to the machine tool control system through a high-frequency data transmission protocol.

3. A method for compensating spatial errors and improving precision of a composite machining tool according to claim 2, Features: The step 2: real-time data monitoring and processing, specifically includes the following steps: Real-time data collection: Start the data acquisition module to monitor the machine tool's motion status, physical environment, and forces and vibrations during processing in real time; Data preprocessing: Filter and denoise the data collected in real time to ensure data reliability. Use filters to remove sensor noise; Status Assessment: Based on the collected data, the machine tool's motion status is evaluated in real time, any anomalies are detected, and this information is passed on to subsequent steps.

4. A method for compensating spatial errors and improving precision of a composite machining machine tool according to claim 3, Features: The step 3: dynamic geometric error compensation model training specifically includes the following steps: Generate training dataset: Using a laser cutter and an industrial camera, marks are drawn on prefabricated workpieces according to preset CNC instructions to generate training workpieces; Image Capture and Correction: Capturing images of the training workpiece using a camera device and performing spatial correction according to calibration parameters; Deep learning model training: The corrected image is paired with the standard image of the theoretical machining trajectory as training data to train the deep learning model and generate a dynamic geometric error compensation model.

5. The method for compensating spatial errors and improving precision of a composite machining tool according to claim 1, Features: The step 4: multimodal data fusion, specifically includes the following steps: Data fusion algorithm design: Design a multimodal fusion algorithm to fuse the data from the laser cutter, camera, force sensor, temperature sensor, and vibration sensor to form a comprehensive machine tool status information set; Fusion model training: Use the fused data to train a comprehensive model to improve the responsiveness to machine tool state changes and the accuracy of error compensation; Real-time update mechanism: Whenever the system detects a state change, the compensation model is updated in real time to ensure that the compensation instructions reflect the current processing conditions.

6. A method for compensating spatial errors and improving precision of a composite machining machine tool according to claim 1, Features: The step 5: intelligent numerical control instruction generation and execution, specifically includes the following steps: Compensation instruction generation: Generate real-time compensation instructions based on the dynamic geometric error compensation model and multi-modal data fusion results; CNC instruction optimization: Generate real-time compensation instructions based on the dynamic geometric error compensation model and multi-modal data fusion results; Machine tool execution instructions: The optimized CNC instructions are sent to the machine tool control system to start workpiece processing.

7. A method for compensating spatial errors and improving precision of a composite machining machine tool according to claim 1, Features: The step 6: post-processing and feedback mechanism specifically includes the following steps: Processing result evaluation: After processing is completed, the workpiece is inspected using a laser cutter or camera to evaluate the accuracy of the processing results; Feedback adjustment mechanism: Feed the evaluation results back to the algorithm to identify areas that do not meet accuracy requirements and guide subsequent model optimization and compensation instruction adjustments; Continuous Improvement: Establish a regular review mechanism to continuously monitor machine tool performance and machining accuracy to improve data collection and compensation models.

Images