Curve fitting with weight assignment under evidence theory combination rule

In engineering practices, curve fitting is a common method to evaluate the performance of machines or equipments using the collected data. If the collected data works as a whole and can not be divided into groups, conventional curve fitting methods can be used to perform the task. Researchers have proposed some improved methods with higher precision and computational efficiency, in order to handle some special situations. Another special situation, however, involves the collected data that consist of more than one sample set, which show obvious differences in collection methods, collection districts and other aspects one could meet. If we treat this collected data as one set, the information reflecting the differences may be ignored and lost in the curve fitting procedure. D-S evidence theory is a widely used method to solve multiple-source uncertain and imprecise information fusion. In this paper, the evidence theory combination rule is introduced to curve fitting preprocessing for weight factor assignment according to the fusion result, so as to address this special situation.