Research On Tool Wear Detection And Machining Error Compensation Technology In CNC Turning

The proposal of intelligent manufacturing 2025 shows that mechanical manufacturing is developing in the direction of digitization,networking and intelligence.The rapid development of aerospace,vehicle manufacturing and other industries also put forward the requirements of higher processing quality and efficiency for modern advanced manufacturing technology.Tool wear affects the machining surface quality and dimensional accuracy.Therefore,it is necessary to build an efficient and intelligent CNC lathe tool wear detection system.Therefore,To manage tool life and compense machining error caused by tool tip wear,this paper studies the measurement algorithm of tool surface wear value based on machine vision and deep learning,and the machining error compensation technology caused by tool tip wear.The main research contents are as follows:1.The overall scheme of in-machine image acquisition and machining compensation system for turning tool front and rear surfaces based on machine vision has been designed.The mechanism of tool wear and the cause of machining error caused by tool tip wear has been analyzed.According to the requirements of image acquisition system,the camera system,light source system,in-machine image acquisition device and other important components are selected and designed.The upper computer is developed based on MVC framework and React JS technology integrates into “Nangaoyun” platform,realizing cloud control of the whole processing process.It includes the module of tool information management,3d model analysis,automatic generation of lathe code and so on.2.A two-stage tool flank wear detection method based on deep learning is proposed.The first stage is to classify the wear state of the tool flank to determine whether it is a normal wear state or an abnormal wear state(chipping or breakage);the second stage is to identify the flank wear area and calculate the wear value.Aiming at the classification of tool flank wear state,a method for classifying tool flank wear state based on model migration is studied.Comparing with different fine-tuning methods,the final detection model and results are obtained,and the accuracy of classifying the wear state is 94.44%.3.The identification method of tool surface wear area based on semantic segmentation and BEMD is sudied,respectively.The wear region recognition algorithm based on BEMD uses BEMD to obtain the wear region after image enhancement through OTSU algorithm binarization,morphological expansion and cavity filling.The wear region of different wear images is analyzed and verified.The wear region recognition algorithm based on semantic segmentation adopts Deeplab V3 network architecture,and the average intersection ratio obtained on the test set is 95.58%.Compared with the BEMD method,the experimental results show that the algorithm has higher intersection ratio and better robustness.Compared with the optical microscope,the error value is less than 20μm,which verifies the reliability of the system.4.Automatic generation technology of turning code for linear turning body and compensation method of machining error caused by tool tip wear have been researched.By analyzing the 3D model,the contour of the straight turning body is identified and the turning tool processing track is designed.Combined with the tool information and process information,lathe processing code is automatic generated.After the establishment of the tool wear model,the offset of workpiece surface before and after wear is theoretically analyzed.To determine the error value of each position point of the workpiece,the machining error compensation model caused by tool tip radius wear is establishedand the machining error is offset by adding z-direction compensation.The tool radius measurement method based on machine vision is used to get the average tool radius.The machining error compensation experiment caused by tool tip wear is designed.The material is processed with 45 steel and processed with cemented carbide blade.The experimental results show that the average deviation decreases from0.062 mm to 0.033 mm before and after the compensation,which meets the machining requirement of the upper deviation of 0.05 mm.The correctness of the error model and the validity of the compensation system are verified.