Abstract:

As insufficient original reliability parameters greatly reduce the authenticity and effectiveness of the reliability assessment results of power systems, an optimized GM （ 1,1 ） model is used to predict original reliability parameters for the purpose of exploiting some small sample systems. In the optimized GM （ 1,1 ） model, the fitting series of the original parameter series within different time is obtained based on the optimization of the initial value using the least square method, and the fuzzy nearnesses between the original series and the fitting series are used as weight coefficients to treat the prediction values with weighted optimization. This model reflects not only the latest changes but also the overall development trend of data. Therefore, it fully exploits the information contained in original parameters and solves the problem existing the conventional GM （ 1,1 ） model, namely the rapid decline of prediction precision with the prediction point. The proposed model is applicable to the multi-point prediction of original reliability parameters of new electricity components.

Key words: power system, original reliability parameter, fuzzy nearness, optimized GM （ 1,1 ） prediction