Research On Fault Tree Analysis Method For Distribution System

As a part of the power system which directly connected with the users, thedistribution system is a concentrated expression of power system reliability. With thedevelopment of the socio-economic and people’s living standards, the demands on thepower supply reliability have greatly increased. Transmission system failure will causeserious consequences, thus, compared with transmission system, distribution systemreceives less concern. Furthermore, the investment ratio on distribution system is farbelow that of development country. Therefore, it is of great importance to pay moreattention and improve the distribution system reliability.Within the existing distribution system reliability analysis method, analyticalmethod and simulation are universally adopted. However, some features such as thedynamic characteristics resulted from dynamic or standby unit are neglected in thesecommon methods. In this thesis, dynamic fault tree is used to analyze the distributionsystem. Markov method is combined in the traditional dynamic fault tree analysis;however, the calculation process of Markov method is with exponential growth. That is,even if for a simple system, explosion may occur and result can’t be figured out. On thisbasis, Bayesian network is adopted to the system modeling.As one of the graph models, Bayesian network possesses the unique advantage toabundantly consider the system uncertainty and recognize the system weaknesses. Theforward causal reasoning of Bayesian network helps to analyze the system uncertaintyand obtain system conditional probability distribution. The reverse diagnostic reasoningof Bayesian network diagnoses the system weaknesses taking advantage of systemconditional probability, helps to improve system reliability.Since distribution system is connected with the customer, the fault may be causedby users. The other factors such as the shortage of statistical information and thestatistic error, making the distribution system has a certain degree of uncertainty. In theprocess of calculating the reliability index should fully consider these uncertainties. Thispaper used the algorithm of interval which treated the original reliability parameterswith interval number. The uncertainty of the parameters can be calculated in the whole process of the reliability evaluation and we can get the interval indicator through oncecalculation.In this thesis, IEEE RBTS bus2typical distribution system reliability is analyzed.Dynamic fault tree is established and then transformed to Bayesian network, thusreliability indices can be figured out through Bayesian network inference. As a result,feasibility of the proposed method is verified. Moreover, system weaknesses can berecognized making use of Bayesian reverse diagnostic reasoning, thus system structurecan be refined to improve the reliability indices.