Research On CNC Tool Wear Condition On-line Visual Diagnosis Method

Driven by "Made in China 2025" and "Industry 4.0",intelligent,flexible and Informatization have become an important trend in the development of CNC machine tool.The tool condition visual diagnosis system and intelligent monitoring function are integrated into the CNC machine tool,which can realize on-line monitoring of tool wear condition and life management of tool,reduce downtime and reject rate,and ensure production safety.Due to the mapping relationship between the degree of tool wear and the surface texture of the workpiece,in order to realize the rapid detection and prediction of tool wear in the CNC machining cycle,the tool CCD online monitoring technology and system based on the surface texture of the workpiece were studied.The main research content is as follows:1.A tool wear condition visual diagnosis method based on the surface texture of the workpiece was proposed.The texture feature set of the workpiece surface visual image was input into the support vector machine to establish a classification model for the tool visual diagnosis.The penalty factor and radial basis kernel function parameter of the model were optimized by particle swarm optimization algorithm,which improved the classification accuracy of the model.The initial,normal,and sharp wear condition of the tool were quickly and accurately distinguished by the model.2.Aiming at the characteristics of confused texture direction on the surface of the milled workpiece,a method based on Gabor-GLCM to extract the surface texture feature of workpiece was proposed drew from the method of processing visual information in primary visual cortex in HMAX model.The imaginary part of the multi-directional Gabor filter was constructed to filter the surface image of the workpiece.Since the imaginary part of the filter responds significantly at the jump of texture and defect edge,the texture area of the tool mark was highlighted.Since the filtering has multi-directional selectivity,the main direction of the image texture was distinguished according to the sensitivity of the filtering response to the direction.Gabor texture features and GLCM second-order statistical texture features were extracted from the filtered image.The redundant information in the texture features was removed by the principal component analysis method,and the surface texture features of the workpiece were obtained.3.The CNC embedded control technology and its software and hardware system for online visual diagnosis of the tool were developed based on the interchangeable control technology of the probe,and the CCD inspecting process was integrated into the CNC machining equipment.Compared with the touch trigger probe,the CCD visual diagnosis probe was not only easy and reliable to be operated,but also required no additional path planning.The probe was more efficient and the accuracy could reach 1.875?m.Therefore,the probe was especially suitable for online tool monitoring and life management system in complex multi-process machining.The image experiments were carried out using the surface texture image of the milled workpiece.The rotation invariance of texture features which extracted based on Gabor-GLCM method was verified.The discriminant time of the tool visual diagnosis classification model was shortened after reducing the feature dimension.The accuracy of discriminating the tool wear condition reached to 98.67% after optimizing the model parameters.The controllable function of CNC code,the interchangeable function of CCD probe,the visual diagnosis function of tool wear condition and trace display function of HMI interface were verified through the practical engineering application of CNC machining key parts.