Research on FSV in membership function credibility verification — for system performance evaluation

Abstract

In order to overcome the application shortcomings of the membership function in traditional AHP-fuzzy comprehensive evaluation method under multi-levels evaluation, this paper presents a new assessment method which is based on the curve characteristics of equipment performance degradation. Depending on the actual descent of the device performance when disturbed, the level of assess and the transition range and the membership function between the different levels are determined, then we can set up the membership function model that is applicable to the evaluation of specific equipment. The Feature Selective Validation (FSV) method is the core of the electromagnetic simulation validity assessment system established by the IEEE standard 1597.1. FSV method is based on the data characteristics to carry out the evaluation of the credibility of the results, it can also evaluate the differences between the simulation data and the credible data quantitatively, which will be introduced in. First, the mathematical model of membership function based on the performance of equipment is established. Second, comparing the traditional membership function with the member function which is based on equipment performance, and applying the FSV to evaluate the results of the two methods. Last, it is concluded that the membership function model based on device performance is more effective. The differences caused by the inaccuracy of the traditional membership function and the linearization of the equipment performance degradation curve are overcome. The more accurate membership function of different grades can be gotten.