Error Prediction And Compensation Of On-line Inspection System For NC Machining And Inspection System Integration

The application of complex surface parts are used more and more widely in the areas of shipbuilding, aerospace, mold, automotive and other industrials. Such complex parts need a high machining precision and corresponding inspection system, current system is difficult to meet this requirements of fast, high-precision, on-line testing, and becomes a barrier for the development of high precision equipment manufacturing industry. Inspection of complex surface parts is usually implemented off-line methods, such as CMM. But because of secondary clamping errors, expensive equipment, high operating environment requirements and higher operating difficulty problem, it is difficult to popularize in small and medium-sized enterprises. On-line inspection can directly measure machining precision without disassembling the workpiece, besides by modifying the NC code to compensate processing error and form a "machining-measurement-compensation" closed loop system, so it has been a popular approach in study abroad in recent years. In order to tackle the difficult problems of an on-line inspection system, this paper focuses on the error prediction and compensation method of on-line inspection system, especially the residual errors after the compensation of machine geometry errors, probe radius errors and pre-travel errors. An error regression model has. been established through the algorithm of Partial Least-Squares Regression and use the built regression model to compensate errors again. At the same time, proposed a method based on Bootstrap to evaluate test results uncertainty of on-line inspection system. The main contents of this dissertation are listed as follows.(1) Investigate and overview the development status on the topics of error analysis and compensation methods of on-line inspection system all over the world, especially the compensation of machine geometry errors, probe radius errors and pre-travel errors, and point out the difficult issues to be solved in this area.(2) Inspection accuracy influencing factors of an on-line inspection system are discussed and error compensation is performed in detail. The main errors of the inspection system, i.e. geometric errors of machine tools, pre-travel errors and probe radius errors, are analyzed. The geometric error model of three-axis numerical control machine is established and the individual error is identified with a laser interferometer. The pre-travel error can be compensated by put the measuring points map to a standard ball in the normal vector direction.The probe radius errors can alse be compensated in the normal vector directions.(3) Based on the results of on-line testing system after the basic error compensation, the study investigated the residual error and created the GM(1,1) model of the gray theory and the ARIMA model of time series analysis to describe and predict the error.(4) Because the results predicted by the GM (1,1) and ARIMA model are not ideal, the partial least squares regression algorithm is proposed to predict and compensate the residual error along the normal vector direction.(5)This study proposed an analysis of standard uncertainty of on-line inspection system error analysis based on Bootstrap methods to provide a complete evaluation methods and indicators for the on-line inspection system.(6) On the basis of these studies, on-line inspection system "uncertainty analysis module" and "on-line inspection system error prediction and compensation module" have been developed to achieve the forecast, compensation and with uncertainty analysis of inspection error functions.