Research On Techniques Of Cutting Tool Wear Condition Monitoring Based On Geometrical Shape And Texture Analysis Of Tool Wear Images

Principally tool wear monitoring system can be classified into two groups, whether if they measure tool wear directly at the cutting edge of the worn tool (direct measurement techniques) or if the measured parameters of signals of the cutting process allow to draw conclusion upon the degree of tool wear (indirect measurement techniques). Now many efforts are made in the field of indirect techniques and made few in the direct field. But the direct techniques based on the image processing and computer vision has better study value and foreground. This paper discusses our work on image geometric shape and texture generated from the flank face and investigates the relation between the flank wear and the image' geometric shape and texture. The paper gives the detecting principle and method, also provides some new algorithms. The main ideas in every chapter is followingThe first chapter expounds the significance of cutting tool detection and its actuality firstly and then compares the detection methods of tool condition. After these, the research aims and contents are proposed. Finally, the main innovations of the dissertation are put forward.The second chapter analyzes the characters of tool's wearing surface. It is the gist that we select the flank face as the detecting object.The third chapter analyzes the geometry shape of flank face that is weared or not weared firstly, and then the detection method of tool condition based on Average Distance or Mahalanobis distance is put forward. The validity of these methods is tested also.The fourth chapter analyzes the texture characteristic of flank face that is weared or not weared firstly, then the gray co-occurrence matrix, auto-correlation function characteristic and Fourier characteristic are adopted to analyze the tool condition and some characteristics are selected from them to detect the tool condition.The fifth chapter applies the Dynamic Programming Chain Method to optimize the algorithm serials. The examination prove the result of optimize isn't sensitivity to the decline of image quality caused by the distortion of optics system. This improves the adaptability of the detecting system. Synthesis opinion of analysis is given at last.The sixth chapter summarizes the main fruits and conclusion in the dissertation and prospects the future of the research.