Research And Implementation Of Electric Power Fault Analysis And Early Warning System Based On Data Mining

Electric energy is the most important and convenient energy in modern society, which has been very widely used in various industries. Electric Energy Collection and Billing System is the subsystem of Electricity Market Operation Support System. Its main function is the remote automatic collection, storage, transmission, statistics and billing of electric grid electric energy data, and it is the guarantee of economical operation, scientific scheduling and assessment settlement of electric grid. If some fault of Electric Energy Collection and Billing System occurs, there will be seriously bad consequences in economic benefits and scheduling management.The actual methods of Electric Energy Collection and Billing System fault monitoring, early warning and diagnostic have some defects such as slow response rate, lag, lack of fault analysis. So it's needed to research and develop a kind of new Fault Analysis and Early Warning System, which can monitor real-time data automatedly, find out abnormal data, intelligently analyse abnormal data to find out the most possible type of fault to provide to those who need it as reference.This paper brings forward the data mining technology route for Electric Energy Collection and Billing System fault analysis and early warning. This paper studies the application of outlier analysis and data classification and prediction in this issue: firstly, this paper improves distance-based outlier analysis algorithm, which achieves satisfactory experiment results; secondly, this paper makes the experiment of decision tree algorithm on real data, which achieves satisfactory results. Meanwhile, this paper discusses the applicability of the strategy pattern, one of the object-oriented software design patterns, on data mining software development. On the basis of the needs analysis, this paper designs and implements Electric Energy Collection and Billing System Fault Analysis and Early Warning System based on Data Ming and strategy pattern, and introduces its architecture, modeling system module, real-time system module, knowledge database and model database.