Novel Method of Evaluating Dynamic Repeated Measurement Uncertainty

Abstract

Uncertainty evaluation of dynamic repeated measurement is usually a difficult problem. Different from conventional methods, we used a novel method to evaluate the dynamic repeated measurement uncertainty in this paper: (1) When calculating an expectation function, we use a high order Legendre orthogonal polynomial fitting instead of the least squares method, so as to eliminate morbidity phenomenon that sometimes appeared in the application of the conventional method. (2) Using the measurement data and the accurate expectation function, we analyze some statistical characteristics of the dynamic repeated measurement random errors, according to these, we use a random number generator to generate lots of random numbers simulating the random errors in all discrete sampling points and acquire a large size of dynamic repeated measurement simulation data. (3) We get the dynamic repeated measurement uncertainty of all discrete sampling points in terms of calculating standard deviation. Finally, using a special example, we derived some evaluation results from this novel method, the conventional Bessel method and the conventional Grey System method, respectively. This novel evaluation method is proved by the identical results of the three methods.