Rule-based And Case-based Reasoning For Distribution System Fault Diagnosis

As the key part of power system, the stable of electric distribution network is an important guarantee for sustainable development of social production and people's lives. Because of the complicated structure and wide coverage area, the operation of distribution network is easily influenced by natural disasters and its impact factors, which can cause faults of power system. The development of fault diagnosis technology for distribution network provides useful method for this problem.Based on the analysis of mechanism and difficulty for fault diagnosis technology, this paper introduces expert system diagnosis method that is studied and applied earlier and more mature. By synthesizing the advantages of case and rule reasoning, a mixed fault diagnosis strategy is introduced to raise the efficiency of fault diagnosis and remedy the shortage of single method.First, after discussing the trait of rule and case reasoning, a new combined method is proposed. Second, the decision table in rough set theory is utilized to describe the diagnosis knowledge, and an attribution reduction algorithm based on discernibility matrix guides the formation of the rule base, which is important for preparation in rule and case reasoning. Finally, in accordance with the reasonable representation and organization of rule and case base, case indexing and matching are finished by grey relevancy degree. The same or similar solution with goal case can be searched in rule and case base, which provides important reference to solve distribution network fault.Analysis of distribution network failure diagnosis system, hardware platform is established by C/S construction, which is the foundation for database storing rules and case. Based on rule and case reasoning strategy, the fault diagnosis software is developed and various of distribution network fault are simulated and experimented. Actual operation shows that, the strategy maintains the advantages of rules easily accessible and inherites the high accuracy and rich resource of case-based reasoning. This method provides an available approach to solve distribution network fault diagnosis problems.