Research On Cutting Tool-wear State Detection Based On Machine Vision

With the development of intelligent manufacturing,more stringent requirements are put forward for advanced manufacturing and processing technology in the field of industrial production.As one of the most widely used technologies in the processing process,it is necessary to improve the production efficiency of turning processing to promote the continuous development of advanced manufacturing processing technology.The cutting tool has a very important influence on the machining precision and surface machining quality of the workpiece in the process of turning.In the process of on-site processing,in order to ensure the machining precision of the workpiece,it is necessary to ensure that the state of the tool in the process of processing.How to grasp the real-time status of a tool efficiently and ensure the replacement of a failed tool in a timely manner on the premise of maximizing the service life of the tool is a problem that needs to be solved in the current tool condition detection.The research on tool wear detection based on machine vision has been carried out in this thesis which using machine vision to monitor tool wear status.The traditional tool detection process that requires the operator to observe the machining surface of the workpiece to determine the tool wear status would be avoided,however,the traditional method requires higher operator experience and has a certain hysteresis.Some workpiece will be scrapped if the failed tool is used to process the surface of the workpiece.These shortcomings can be avoided effectively with visual inspection and it allows you to make a decision when the tool is about to fail and prompt you to replace it in advance.Combining image processing technology with tool processing state detection technology,tool wear detection based on machine vision has the advantages of high precision,no-contact,easy operation and so on.In order to achieve the intelligent tool wear detection target,the hardware architecture of the tool wear detection system was established,the corresponding image acquisition and processing algorithm software is compiled,the man-machine interaction interface is developed,and the corresponding turning experiment is carried out.In the hardware architecture,the CMOS industrial camera,telecentric lens and ring light source were selected to ensure the efficient acquisition of tool wear images.According to the accuracy requirement of the system,the industrial camera with specific resolution was selected and matched with the telecentric lens.According to the lighting requirements of the imaging environment and tool wear detection,the 30° ring light source and the related lighting scheme were selected and the effect was tested.In the part of image acquisition and processing,the processing algorithm of tool wear image was developed according to the selected hardware characteristics.According to the characteristics of the collected images,the effect of median filter and gaussian filter were compared by the experiment.The threshold processing effect of the adaptive OTSU method on the tool wear image was tested and the processing effect of different binary morphological operations on the tool wear image was compared and an improved algorithm scheme according to the deficiency of processing effect was designed.An improved algorithm for extracting the maximum contour of the connected domain based on suzuki algorithm was proposed and the contour of the maximum wear region of the tool was extracted by intelligent filtering.Finally,the feasibility of the algorithm was verified by images.According to the characteristics of the tool wear profile,the numerical parameters of the tool maximum wear were extracted by using the algorithm of marking maximum wear with the external rectangle.Based on Windows Form and open source visual library OpenCV,the man-machine interface of the system is developed on the Windows operating platform.The feasibility of the algorithm of the tool wear detection system is verified by onmachine test in the field of turning.The comparison of the tool wear value measured by the algorithm with the tool wear amount measured by the laboratory microscope shows that the system wear detection error is within 6%,which can fully meet the needs of tool wear status detection.