Research On The Method And Application Of CNC Machine Tool Condition Visual Monitoring

Under the background of "Made in China 2025" and "Industry 4.0" strategy,CNC machine should develop in the direction of the intelligent integration.CNC machine integrated tool condition monitoring function can greatly reduce the workpiece scrap rate and failure time,effectively improve the production efficiency and tool utilization,and ensure safety in production.Monitoring technology based on the machine vision can realize intelligent,real-time,accurate and rapid monitoring.Therefore,this paper studies the method of on-line extracting tool wear characteristic value and workpiece texture feature for in-process detecting tool wear state based on visual features.The paper first analyzes the mechanism of tool wear and workpiece texture formation.And take the max bandwidth of the tool flank wear area as the tool wear characteristic value,and use the workpiece surface texture feature as the characteristic vector to determine the tool wear state.Considering that the industrial scene images are collected with noise and uneven illumination,the preprocessing methods of tool wear image and workpiece texture image are studied which includs adaptive median filtering and top cap operation.Secondly,the automatic wear characteristic value extraction algorithm is studied for the characteristics of tool flank wear image.We designed automatic selection of seed points and growth threshold for region growing algorithm to segment the flank wear,and extract the characteristic value of the tool wear by minimum bounding rectangle.Compared with other tool wear characteristic value extraction algorithm based on threshold segmentation,the automatic tool wear characteristic value extraction algorithm proposed in this paper is more simple and practical for the actual application.The automatic texture feature extraction algorithm is also studied for the surface texture image characteristics.The amplitude and phase information obtained by the quaternion wavelet transform of the texture image together with the cutting parameters constitute the workpiece texture feature vector.And build a texture image classification model based on BP neural network to determine tool wear condition.Compared with the method of analyzing the workpiece texture feature based on the gray level co-occurrence matrix,the texture feature extracted by the method proposed in this paper is more reliable and accurate.At last,this paper construct the tool wear condition visual monitoring system for the CNC machine,analyzes the feasibility and effectiveness of the tool wear characteristic extraction algorithm and the workpiece surface texture feature analysis algorithm,and present the specific implementation process of the monitoring system.The image processing experiments show that the system is suitable and applicable to non-stop automatic detection of the CNC machine tools wear state for the industrial practice.