Applied Research On Expert System Of Intelligent Power Grid Diagnosis Based On Data Mining Technology

Recently, with the innovation of information technology, the application of database and Internet is widely used in all sorts of business. More and more information come into our work and life, and its amount is growing exponentially. At the same time, the data processing amount in industry and commerce is also increasing fast. Now we are in a society of information explosion. So how to deal with the massive and growing information efficiently and find out valuable information from it in this era becomes a focus that all the people are paying attention to. The gaining and extraction of information plays a more and more important role in the field of power, business, and automation control etc. The data mining technology even enables people to fetch the valuable knowledge and rules rapidly in numerous and disorderly information.In this paper, we apply the decision tree algorithm of data mining technology to intelligent power grid fault diagnosis system, and build an intelligent trouble diagnosis expert system based on uncertain factors. Our model combines knowledge gaining and expression together by the decision tree, so it overcomes its shortage in traditional method, making the knowledge expression synchronize with knowledge gaining. In this model, we use the way of "if-then" to express knowledge with advantages as follows:reasoning clearly, easy to understand, explaining conveniently, at the same time, knowledge gaining is accomplished by improved decision tree algorithm, so the two courses can combine together perfectly. Also in this model, we improve decision tree classification algorithm in data mining technology. At first we need to select proper data set of training sample by the optimal screening sample technology. Then improve the selection criteria of test property. Finally we also improve the discretization method of binary search to continuous attributes. Our improved decision tree algorithm is able to produce efficient, clear, and brief "if-then" rules quickly. The rule tested by sample set is of high accuracy and its classification is satisfying. By applying the improved algorithm to intelligent power grid fault diagnosis system, the efficiency and stability of our system is improved.The experiment system raised in this paper is composed of four parts, the inference engine, the interpreter, the decision tree algorithm, and the HMI-human machine interface. The inference engine part takes charge of reasoning the rules decision tree algorithm produces. The interpreter part mainly translates rules the inference engine returns or knowledge engineers write in. The decision tree algorithm will build a fine decision tree and produce rules which can classify exactly so that the inference engine can use it. The HMI part realizes the function of man-machine communication. The experiment system is based on ".NET" platform, so its safety performance is excellent and it is also easy to extend, at the same time, if we use proper data sample set or other data mining algorithm, it can also deal with problems in a wider range of field. From the results we can see this model not only realizes knowledge expression and knowledge gaining automatically but also owns many advantages, such as low computational complexity, simpler course, and good applicability. This model can create the decision tree quickly and the knowledge it gains in the next step, namely the rules, can reach higher reasoning efficiency. So it can solve many problems in intelligent power grid decision. In this way it can solve many problems effectively in power grid decision process. It is of high value in use and full of practical significance.