Research On Uncertainty Evaluation Method Of Coaxiality Error Measurement

In the design and manufacture of precision machinery,the measurement and evaluation of geometric error is a very important content.The normal operation of the equipment and the normal coordination of the shaft parts are closely related to the coaxiality error.The new generation of GPS system and uncertainty theory show that the measurement result of geometric error including the measurement uncertainty is more reasonable.At present,the evaluation method given in the internationally certified "guide to the expression of measurement uncertainty" is based on the classical statistical theory.It needs a large number of measurement data samples when it is used.In the actual measurement process,this will lead to high cost and low efficiency,so it is usually unable to meet the above conditions and suppose it obeys the prior probability density.In order to solve this problem,it is of great theoretical significance and practical value to study the evaluation theory of measurement uncertainty under small sample size and unknown distribution.In this paper,based on the evaluation of measurement uncertainty by non statistical method,in the process of studying coaxiality error and measurement uncertainty,the regeneration weight least square method is first discussed to preprocess the measured data,and then compared with the least square method,genetic algorithm and BP neural network method to fit the center coordinates and spatial linear equation,and finally get coaxiality error knot Then,based on the support vector machine algorithm,the probability density of the error results is estimated,and the measurement uncertainty is obtained by numerical integration of the probability density function.In addition,the model and formula of uncertainty evaluation of coaxiality error measurement are derived in detail according to the gum method proposed in the guide to the expression of measurement uncertainty.Finally,the core shaft of the classical part is selected,and the method is applied by the measured data,and the results are compared with the results of gum method,Monte Carlo method and Measurement Institute,respectively,to verify the accuracy and rationality of the test results.