| All surfaces, either natural or artificial, exhibit roughness at some scale. Study on the surface texture is called texture analysis and has important application in the field of computer vision. In Mechanical Engineering, conducting research on the roughness of surfaces of the machine element also is profoundly practically significant. Characterization and evaluation of machined surface topography has always being one of the most vital and appealing issues in the area of surface metrology and tribology. It is well known the geometric structure of the rough surface influences numerous physical properties of the machine part, as friction, wear, lubrication, contact stiffness, seal, contact resistance and so forth. On the other hand, the surface quality reflects the fabrication capability of the processing apparatus. It seems the machine part surface is the carrier of functional information for both machine element and processing apparatus. Presently, with the precision test and control technology, the online measurement of surface topography has come true, which makes the surface quality a feedback reference point of the close-loop control of modern manufacturing system, not merely an index in the quality control of production system. Hence, characterization and analysis of the machine element surface stands out as an ever-increasingly important traditional as well as burgeoning issue in mechanical engineering.In the traditional founded surface metrology, the irregular roughness of surface topography falls into three scales, which were form, waviness and surface roughness respectively, according to their corresponding wavelength or frequency. The mainly adopted tool was spectral analysis and mathematical statistics, on which corresponding metrological criteria were established. A great amount of experiments since 1970s show machine surface is multi-scale and non-stationary random - a fact that fluctuates the theoretical foundation of the tradition surface metrology, which was on the basis of stationary random process.Multi-scale Estimation Theory proposed by A. S. Willsky at MIT, French mathematician A. Benveniste and R. Nikoukhah in 1990, combines the dynamic system estimation theory, recognition method, wavelets analysis, random process and mathematical statistics, and proves an effective tool for analyzing the phenomena and processes which possess multi-scale property. This paper introduces MSET into surface metrology and the main studies lie in two aspects.First, the two most common used methods adopted in machined surface assessment are summarized. Among filtering methods, wavelet-based filtering method is proposed to be applied in establishment of the surface evaluation reference. Wavelets analysis, key content of MSET, is systematically studied. From the perspective of engineering application, the theory and practice of CWT and DWT are presented. The wavelet decomposition is compared with the traditional analogous and digital filter theory. The linearly non- aberrant property of wavelet filtering isdemonstrated through the use of Shannon wavelet in the process of decomposition and reconstruction. Two numerical examples are exploited to demonstrate the three scale surface error separating process. The correlation between the evaluation parameters and the instrument resolution is illustrated to explain the significance of establishing new criteria within limited band, which may be called band-limited criteria. In addition, the redundancy of evaluation parameter set is discussed using former experiment data.Second, a I/f process model is proposed to describe the machined surface profile, which constitutes another new research object in surface characterization. The standard deviation of the local fractal dimensions can quantitivcly assesses the homogeneity of engineered surfaces due to their relationship discussed. Furthermore, because of the quantitive connection between fractal dimension and the spectra parameter of a \lf process, standard deviation of the latter is equivalent to that of t... |
PDF Full Text Request