The Research And Application Of Data Mining In Power Plant Operation Optimization

Data mining is one of important parts of intelligence system. For its mighty function in finding valuable and potential knowledge among large amount of data by means of many advanced technologies such as artificial intelligence, pattern recognition and mathematical statistics, it has been widely developed in many fields. Data mining improves the application of data from primary searches to advanced decision making.Power plant operation optimization involves many fields and it is an important method to keep the unit operating safely and economically. But up to now, many noduses have been left unsolved such as decision of optimal target value and optimization operation guide.With the development of power plant information technology, more and more field data were saved to MIS or SIS. Power enterprise gathered enough data and abundant knowledge was contained in these data. It will be valuable and significative if we can find knowledge from operation data to improve the operation level of the system. While power plant operation data is complex, dynamic and interactive, it is difficult to analyze data artificially. So this paper proposed a new method based on data mining to optimizing operation. The new method combined pivotal data mining theory and power plant operation optimization and induced data mining technology to operation optimization process. Aiming at the application of data mining in power plant operation optimization, this paper mainly studied the following three problems: the decision of optimal target value based on data mining, operation guide based on variance analysis, the construction of power plant operation optimization system.1. Reviewed and analyzed power plant operation optimization deeply, especially the decision of optimal target value. This paper discussed current methods to decide optimal value and analyzed their disadvantage. Then proposed a new method based on fuzzy association rule to decide optimal target value. This method took full considerations of virtual condition, so it could be used to instruct operation. And because the value came from real operation data, it was realizable in operation.2. Analyzed the problems of operation optimization based on data mining in detail. Indicated the importance of data selecting and in order to avoid blindness in...