The Research Of Expert System Based On Fuzzy Theory In Power System Fault Diagnosis

Power system plays an important role in the national economy and our daily life. Asthe development of economy and technology, power system merged intelligent technologybecomes an important trend. Many intelligent technologies are put in use of the powersystem, such as sensing technology, control method, decision and support technology,communication technology, energy and electricity. The technologies from varies areas havegreat progress and are used in traditional power system. The modern smart grid usually hasa huge scale and its structure is very complicated. As a result of so many precisioninstruments involved in the power system, higher requirement of technologies is needed.The modern smart grid has become a huge and complex system which owns many technicaldisciplines.As the key in power system, transformer is very important both in the transmissionand use courses. It increases the voltage to reduce loss during transmission and in the clientpart, transformer reduce the voltage to satisfy the need of users. The voltage of industrialelectricity and living power is different, which is complied by transformer. Because of thecomplex structure of itself, also the transformer is easy to have problems, which willusually cause an serious accident in the whole power system.To improve the efficiency of transformer and ensure it running in a safe and stablestate for a longer time, we can do some work in the following parts. First of all, using thetransformer produced strictly and in the manufacturing sector with higher level oftechnology. So we can make sure that the transformer is of higher quality congenital andwhen some accidents come about, it can be deal with proper. Next when the quality isinsured, the using life and failure frequency are mainly decided by the maintenance. Manyresearches is done for this. How to make sure the transformer run safely and stable, andforecast the probable fault in early time, is becoming an important. At the same time, wealso need to avoid the possible fault and deal with it in the initial stage. Many scholars aredoing research on this problem, and kinds of smart technologies are to be used in this area. Still there are many shortcomings. In this paper, we propose an expert system based onfuzzy theory. The traditional machine reasoning is mechanic while human language isambiguous, which is a big contradiction existing for a long time. The fuzzy theory cansolve this problem effectively. We consider both the influence of many factors and a certainfactor may have decisive influence to the final output in the algorithm proposed in thispaper. The experimental results show this algorithm is effective and accurate. Thisalgorithm is applied in the project of Research and Application of Intelligent Method onFault Diagnosis in Power System, which is sponsored by Science and TechnologyDepartment of Jilin Province. We adjust this algorithm in the actual test and its feasibility isincreased greatly. At the same time, the speed of diagnosis and accuracy are also improved.In the algorithm proposed in this paper, we choose the following factors for thecomprehensive evaluation based on the common transformer faults, gas test, electrical test,fault history, and the state of itself. Gas test and electrical test are decided by other4and5sub-factors dividedly. In the progress of data preprocessing, we standardized the raw datathrough the fuzzy theory. The results are numbers between0and100with no unit. Throughthis algorithm we obtain a fuzzy comprehensive evaluation matrix for the problem oftransformer fault diagnosis. The matrix is calculated by weight matrix and relationshipmatrix. By such kind of calculation, two sides of the issue are taken into account. On onehand, what kind of influence the every single factor to the whole question is considered; onthe other hand, the relationship between certain a factor and certain kind of the results isconsidered. So the algorithm overcome the malpractice of common algorithm that when anextreme case come out, the system may be helpless. Experiment results show that thisalgorithm performs better compared with the common algorithm of fuzzy expert system. Atthe same time, the time of diagnosis and accuracy are also improved.