Reliability evaluation in electrical power generation with uncertainty modeling by fuzzy number

This paper presents an alternative approach in modeling and dealing with uncertainties in electrical power generation reliability evaluation. The proposed method, different from conventional methods which are mainly based on a probabilistic approach, employs fuzzy set concepts together with probabilistic theory. Firstly, in generation modeling, probabilistic concepts are still used to represent the failure-repair cycle of generators. However, fuzzy numbers, instead of an expected value, will be used to represent failure and repair rates. Secondly, in load modeling, the cumulative load duration/frequency curves will also be described by fuzzy number so that the uncertainty on the forecasted load can be taken into account. With the proposed method, the uncertainties embedded in both generation and demand sides can be handled more appropriately and flexibly. Then, the fuzzy reliability indices (LOLE, EENS, F&D) can be obtained using fuzzy arithmetic operations. The proposed method has been tested on IEEE-RTS and compared with a conventional probabilistic approach. The obtained results show that fuzzy reliability indices will provide a better explanation than crisp indices.