POWER-SYSTEM RELIABILITY EVALUATION USING FUZZY-LOGIC AND NEURAL NETWORKS

Problems related to Power System Reliability calculations are very com plex since it involves modeling the stochastic nature of the power sys tem. Till now most of the reliability calculations were performed usin g a probalilistic model which aims at foreseeing only the average outa ge performance of a group of units during a long period of time [2]. M ost models permit only the existance of two states i.e., whether an un it is available or not. Inclusion of additional states increases the c omplexity of the model. Using modern tools like fuzzy logic and neural networks, it is much more easier to incorporate the stochastic nature of the power system. It is also very convenient to model the power sy stem closer to the actual operating indices rather than to use average d indices. This paper presents the generator and load model which use fuzzy logic. Further the models are extended to the system with many u nits which is defined on aggregate basis. Finally, the system reliabil ity is defined in the fuzzified jargon as Linguistic Reserve Capacity States. Artificial Neural Network is used to model the essential param eters of the generating units. The method presented can be easily inco rporated into the existing frequency and duration approach for the eva luation of power system reliability. The proposed method has been appl ied to existing system data and the results are presented and discusse d.