Research On Surface Quality Evaluation And Prediction Model Based On CCOS Small Tools Polishing Of Quartz Glass

With the development of the economy and society, optical elements of highprecision and high-quality have attracted more and more attention, and they are applied more and more extensively in the field of aerospace, national defense, space strategy and information industry. To obtain optical elements of high-precision and high-quality needs advanced and scientific processing methods. Compared with the traditional processing methods, computer-controlled small tools polishing technology in this paper can get more accurate analysis of the surface shape for the workpiece. Its polishing process control is more reliable. In addition, good surface quality measurement and evaluation methods are also of great significance to guide the polishing in order to obtain optical elements of high-precision and high-quality, as well as the mapping prediction model between polishing parameters and surface quality.This paper explores a new characterization method to evaluate the surface quality of quartz glass. Firstly, extract the valley damage of quartz glass surface contour using wavelet analysis method, and then define a new parameter to evaluate the surface damage degree, at last, draw the damage degree curve that can evaluate the valley damage effectively and intuitively. This new parameter complements the traditional roughness characterization methods to make the surface quality evaluation for quartz glass more comprehensive and accurate. The orthogonal experiments in three factors and four levels are in form of planetary motion. The main affecting factors include the polishing pressure, tool speed and eccentric in the paper. During experimental data processing, Taguchi method and variance analysis are used to evaluate the effect of various factors on the surface roughness and get the optimal combination of parameters preliminarily within certain range of the computer controlled small tools polishing process. And then, according to the experimental data, BP neural network prediction model and a nonlinear regression prediction mode are established to research the relationship between polishing parameters and surface roughness of quartz glass. By comparing generalization error and relative error of the two predictive models, BP neural network model is better, and it can predict the mapping between polishing parameters and surface roughness of quartz glass to a certain extent. It has certain value to guide and control the polishing process parameters for the purpose of getting the desired surface quality of the polished workpiece.