Research On Filtering And Evaluation Methods In Precision Measurement Of Form Error

With the fast development of precision machining technology in our country, higher and higher accuracy of measuring comprehensive parameters of rotary parts is required. In order to establish the system of evaluating diameter and form error, to ensure the accuracy and reliability of quantity values of diameter and form measurement nationwide and to participate in international comparisons to ensure international equivalence, an ultra-precise standard system of measuring diameter and form error is developed in National Institute of Metrology.This subject mainly focuses on filtering and error evaluation methods which are used in the standard system of measuring diameter and form error. The major research topics include the following aspects:1. Robust Gaussian regression filtering method is introduced to eliminate edge effect and outlier-induced distortion in the process of applying Gaussian filter for sampled data of measuring form error.2. The filtering characteristics of Gaussian filter are compared with Spline filter on the basis of introducing Spline filter equations. To overcome the impact of surface outlier on Spline filtering and enhance robustness, robust Spline filtering method which uses Tukey's Biweight method as the weight function is applied in processing of sampled data.3. An improved algorithm of area hunting is put forward to evaluate straightness error, roundness error and cylindricity error using minimum zone method. And the accuracy of this algorithm is verified by experiment.4. Software which integrates the filtering and error evaluation methods proposed in this paper is developed, and it is used in the standard system of measuring diameter and form error.New methods of surface profile filtering such as robust Gaussian Regression filter and Robust Spline filter are described and applied in practical measurement to reduce filtering distortion and improve accuracy. Area hunting algorithm is improved to realize form error evaluation using minimum zone method accurately and efficiently.