Research On The Technology Of Fault Diagnosis Of Power System Based On Probability Petri Net

With the market development of the electric power industry, requirements of users of electric energy for the reliability of power supply, especially the continuity, are getting higher and higher. But because of the impact of man-made, climate and other factors, the generating of power system fault is inevitable, resulting in serious consequences, such as the interrupting of power supply of users, the declining in the quality of power supply and even damages of electrical equipments. At the same time, with the continuous expansion of the grid scale and the acceleration of the electrification process of the whole society, more and more automation devices are applied to the power system. When the transmission network faults occur, a lot of warming information which is generated by the automatic devices of each level will influx into the control center in a short time, which exceeds the processing capacity of the operating personnel. Therefore, it is necessary to study an effective diagnosis method of the power system, which can determine the faulty components quickly and accurately.Petri net is a graphical modeling tool which is applicable to the analysis of the discrete event dynamic system. In recent years, scholars have applied it to the power grid fault diagnosis, however, the proposed methods have problems as following:it is difficult for the models to adapt to the changes of the network structures, meaning that the versatility and maintenance are weak; when the fault information is lost, diagnostic error will occur easily, which means that the fault tolerance is poor. Against to the above problems, the goal to improve the versatility and fault tolerance of the diagnostic model is set, based on the domestic and international relevant research results, the probability petri nets is applied to the power system fault diagnosis. The work content is mainly divided into the following aspects:(1) An extension study on the traditional probability petri nets is made and the evaluation function of the transition and the weights of input and out put is fixed, which improves the performance of the probability petri net diagnostic. According to the statistics of the relay protection and circuit breakers, with the method of probability the value of the initial marking probability is calculated in detail, which makes the initial marking to describe the uncertainty of the actions of the relay protection and the circuit breakers.(2) Based on the new probability petri net theory, a step-by-step model of power system fault diagnosis containing space-time factors is proposed. Based on the fault information system, lead models of the line, the bus and the transformer components are established, which accurately simulates the logic relationship of protections. The model adapted to different system operating modes by the analysis of the structure of one topology is made by the fault tree of suspicious components.(3) Against to the problems of matrix dimension being too large, weak adaptability and combinatorial explosion of the petri net models in complex power grid fault diagnosis, based on the above modeling method and combined with petri nets simplification techniques, hierarchical subnet probability petri net model of the fault diagnosis is constructed. Namely, the bottom subnet models of the main protection and backup protection as well as remote backup to the two ends of the lines and transformers are established firstly, and then the bottom subnets are integrated to establish the comprehensive diagnostic subnet; the bottom subnet models of each lead direction of the bus are established firstly, and then the bottom subnets are integrated to establish the comprehensive diagnostic subnet. Multiple examples show that the scale of the model effectively and the matrix dimension with strong adaptability are reduced by the method.