Profile Error Analysis of Following Ground Crankshaft Using Robust Gaussian Regression Filter

Abstract

In order to control profile error, many precision grinder manufacturers use profile error pre-compensation method, however, when there are outliers in the original profile error measured data, the compensation accuracy of profile error will be greatly influenced and even the work piece will be scrapped. In this paper, in order to overcome this problem, a robust Gaussian regression filter is researched, at the same time Gaussian filter and Rk filter are used. Three filtration methods are applied based on the same set of profile error data with the same outlier. Comparing the analyzed results, it is obvious that the robust Gaussian regression filter has the strongest anti-outlier capability. The application of this kind of filter is of great significance to improve the crankshaft following grinding machine reliability and performance.