Reliability Assessment Of Distribution Systems Based On Bayesian Networks

The distribution systems reliability assessment is one of the most important part of power systems reliability assessment. The reliability assessment study to generation systems is of critical theoretic significance and great social economic worth. In general reliability assessment computation, only are the reliability indices computated, little has been done to identify the weak components. However, identifying the weak components is very important.The paper briefly introduces the basal conception of Bayesian networks, then further studys the general inference algorithm, and the Bayesian networks inference program based on bucket elimination is developed. After analysing the usual reliability assessment methods, a new reliability assessment method based on Bayesian networks is presented. The process of this method is as follows: Firstly, build the Bayesian networks models for distribution systems reliability assessment; secondly, anayse the distribution systems reliability by using Bayesian networks inference program. The Bayesian networks automatic generator tool of distribution systems reliability assessment is developed. Using this tool, the building-time of Bayesian networks is remarkably reduced and the reliability assessment model's accuracy is improved.Due to Bayesian networks provide a flexible representing method for uncertainty knowledge. The uncertainty of load forecast is absolved easily by using Bayesian networks method. Utilizing efficient and flexible probabilistic inference of Bayesian networks, not only computing the reliability indices of a distribution system, but also the weak components of the systems be identified.The software of the distribution systems reliability assessment is developed by Microsoft Visual C++6.0 development tool. The reliability index is accounted and the wesk components are analyzed by the software that using the advanced object-oriented means and using the Bayesian networks. The significant condusions are educed passing the analyzed datas. Finally, the test results in the IEEE RBTS bus 2-distribution systerm have verified the effectiveness of this new method and the software. The results show that the new method is valid and available.