The Research Of Fault Dignosis And Alarm Analysis For District Power Grid

In industrialized society, customers require power with better quality and higher reliability. However, it is inevitable that failure and faults occur in power system operation, especially in district power grid. A good fault diagnosis system for districts is required to eliminate the faults, ensure the safe and stable operation of power system, enhance the reliability and consistency, realize the quick positioning of fault and diagnosis of fault category for the benefit of the maintenance and the rapid restoration after the outage.This paper focuses on the fault diagnosis for the district power grid faults. It firstly introduces the definition and structure of ES (Expert Systems) which is one of applications of Artificial Intelligence (AI) techniques, and features of different compositions. The new expert system mentioned in this paper comprises of four parts, including inference machine, rules translating tool, knowledge base and API(Application Programming Interface). The knowledge base is a key step in the whole process. The reason is that a knowledge base with sound structure and complete content directly influences the inference speed and the validity of result in Expert Systems. In this paper, 2 chapters are devoted to illustrate the structure and classification of the knowledge base and the conveyance of knowledge. A new method is also proposed in the 2 chapters for establishing the knowledge base that is to divide the knowledge base into two parts, rule base and case base. In this way, the rule base is applied for the universal faults diagnosis in districts while the case base for special faults diagnosis in districts with specifically structured district power grid. Consequently, the expert system will be easy to be implemented and transplanted into different power grids.Secondly, according to the characters of the knowledge base, a new fault diagnosis is presented based on fuzzy set covering theory. It utilizes incomplete operational information of protective relays and tripping information of circuit breakers to deal with uncertainties involved in the fault diagnosis in power systems, i.e., the operating reliabilities of protective relays and circuit breakers, and the correctness of the received and non-received alarm signals. The test results of a sample power system shows that the fault diagnosis methods are correct and efficient, and are of promise for on-line application in actual districts.Then it illustrates the expert system design approach and the implementation details. In this system, the inference machine and knowledge base are independent, as a result neither of them will influence the other in the self-perfection process.This paper is finished on the bases of the expert system tests over ten thousand times. The test results are presented in the conclusion part.