The Application Of Improving Fuzzy Clustering Algorithm In Power Plant Operation Optimization

With the rapid development of the power industry for informatization and automation in our country, power plant has accumulated vast amounts of historical operating data, which will inevitably contain a great deal of valuable knowledge, these provide the foundation of data mining technology to operation optimization process. But we are also facing the "information rich, lack of knowledge" issue. The missing value in data make data mining in power plant much difficult, the mishandling of missing value may even lead to the failure of data mining. Therefore, how to impute the missing value efficiently and accurately become an urgent task.Fuzzy clustering algorithm reflects the thought of "Like attracts like" and "fuzzy-based". The running mechanism of power plant is quite complex, and the parameters coupled to each other. The operating data under different conditions differs a lot, and under similar conditions presented a certain similarity. These are the theoretical basis of fuzzy clustering applied to the plant data analysis process. In this paper, the improved fuzzy clustering algorithm for multi-attribute data to impute missing values. Fuzzy clustering algorithm is also used to the attributes discretization. Then through the analysis of association rule, we can guide the operation of power plant.The main content of this paper include:1. The study proposes a novel algorithm(SVR-OCSFCM) that applied to impute multi-attribute missing values. In order to improve the efficiency of imputing of missing values, the paper combines the linear algorithm and SVR-OCSFCM algorithm as the imputation strategy.2. The plant data is discretized by the improved fuzzy clustering algorithm. Discretization apply the consistency of the decision table as criteria, and adjust clustering parameters dynamically to optimize effects of discretization.3. The framework and interface of the data preprocessing is developed. And we use Microsoft Visual Studio development environment to achieve the various algorithms and functional modules.4. With the objective of minimizing the coal consumption rate of power supply, the historical data of power plant is imputed and discretized. The optimal oxygen content of various conditions is identified through association rules, which can provide guidance to the optimal operation of the power station.