Analysis And Implementation Of Fan Equipment Running Status Prediction System Based On Similarity-Regression

Fan as an important auxiliary engine of thermal power plants, whether it is able to operate in a secure and economical way, has a direct impact on the level of the cost of thermal power plants. Running trend of fan equipment parameters to predict is one of the key technology to solve device fault, abnormal halt, and can provide powerful data analysis and decision-making basis for decision makers. At present, the domestic is not deep enough in the research of prediction approaches of fan equipment parameters in thermal power plant and the mature prediction system relatively is less. Therefore, fan equipment running status prediction approaches have greatly theoretical and practical research needs.The paper is to do further researches for fan equipment running status prediction approaches, in order to improve the accuracy of prediction results and provide guidance for the development of prediction systems. From the perspective of software engineering, this paper detailed describes the development and implementation of fan equipment running status prediction system. The main work includes the following three aspects:Firstly, Birch clustering algorithm only needs to scan the database once and it does not require to input parameters and easily calculate the center, radius, diameter, in-class distance and between-class distance.In this paper, we use Birch algorithm to preprocess massive historical data, which is to lay the foundation for the study of similarity-regression prediction algorithm.Secondly, we propose prediction algorithm based on similarity-regression. Its key idea is that we introduce variable coefficient weighting and regard it as weight value of similarity calculation in Birch clustering algorithm, which can avoid having the bad effect on clustering results because of big deviation in certain components. After calculating the similarity, we select the nearest cluster and build its predictive model. This not only reduces the time complexity of computing, but also improves the prediction accuracy of the system.Thirdly, we develop fan equipment running status prediction system based on similarity-regression, which is implemented by the Java language and SQL Server database. According to the design principles of software engineering, we complete the system steps of requirement analysis, general design, encoding implementation, and test the whole system and each function at the same time, which verifies the validity of prediction algorithm based on similarity-regression and ensures the stability of the system.