Fault Diagnosis Methods Based On Data Mining

With the rapid development of modern technology, production process is increasingly complexed. Equipment system running continuously at the conditions of high load and high-power, so some failures happens inevitably, which maybe cause serious economic losses and casualties. Fault diagnosis system gives very important means of esuring the equipment running at the condition of efficient, safe and reliable, which is of great significance to reduce the accident rate, extend the service life, and increase the economic efficiency of enterprises. Traditional fault diagnosis methods have played an important role in early fault diagnosis. However, there were some shortcomings more or less due to technical limitations. Applying the data mining techniques into the fault diagnosis system could overcome these shortcomings very well.In this paper, the research current situation both in domestic and foreign of fault diagnosis methods is discussed firstly, and then analyses the limitations and inadequate of traditional methods, made a detailed introduction of DM (Data Mining).A detailed research in the association rules algorithm and decision tree was taken, which are common used in DM. Trying to present improved algorithms of this two algorithms by analyzing the two algorithm's advantages and disadvantages, which were applied in the fault diagnosis system.This paper tries to apply the CA (competition condensed clustering algorithm) algorithm into the feature classification of fault diagnosis. The specific process of fault diagnosis based on fuzzy association rules algorithm was detail descried. According to the characteristics of decision tree algorithm, a decision tree based on ant colony algorithm was proposed. This algorithm ascension the accuracy of classification effectively, and the result shows the improved algorithm is feasible and effective.In the end of the paper, both the improved fuzzy association rules and improved decision tree algorithm applied in a fault diagnosis system, applying SQLserver2000 to create Database, storage process historical data, and take a preliminary treatment to the data. MATLAB GUI is used as a develop tool, comprehensive application Database toolbox, to establish a simplified fault diagnosis system, and verify the correctness and feasibility of the fault diagnosis method by a real example.