Research Of Data-Driving In Power Plant Operation Optimization

Operation optimization on power plant is on the basis of analyzing equipments and systems off line, measuring parameters on line, evaluating and making decisions with the aim of security and economy. Huge amounts of data stored in real-time historical database saves a lot of information and can be utilized for find the operation law and the way to improve the system. In the paper, based on historical data, useful information is extracted from the database and optimized value of target parameters is searched..Rough sets theory can find connotative knowledge by information of data and produce decision rules. Pawlak pointed out that, rough sets can compute the dependence of different attributes and provide the most important ones. With this theory we can reduct the parameters affecting the efficiency and optimize the reduction results.However, this theory can only deal with discrete data and most attributes monitored in power plant are continuous. In the paper we discretize data by fuzzy clustering method. Number of classification and centers are initialized. The results show the final centers and membership matrix and classes data belong to.It exists complicated correlation among operating data. The degree of correlation can be identified to reflect the linear relationship between different attributes and provide the running regulation.Computation model and affecting factors of economical targets are introduced and a 600MW coal-fired power plant is given as an example. Decision rules are evaluated by confidence and support.