Performance Prediction And Optimization Of Utility Boiler Based On Big Data Analysis

With the gradual improvement of automation technology and information processing speed of coal-fired power plants in China,many power plants collect a lot of boiler operation data,and these data contain very critical information.Using big data method to analyze these data can obtain the professional knowledge about boiler operation improvement,which is helpful to guide the operation optimization scheme of power plant boiler in the future.In this paper,660MW supercritical unit is taken as an example.Firstly,the data of relevant measurement points of the boiler of the corresponding unit is extracted,and the abnormal value of the operation data is processed.Because the input parameters of the power station are many and complex,the average influence value method based on support vector regression machine is used to screen the input variables.Then,according to the situation that the boiler has multi working conditions,the optimization should be based on the actual situation of the boiler In this paper,the K-means clustering algorithm is used to classify the data,and the elbow chart method is used to select the K value,and the working conditions are divided into three categories.In addition,two kinds of commonly used support vector regression algorithm and BP neural network algorithm are selected to establish the boiler efficiency and unit load forecasting models corresponding to the corresponding classification conditions,and the parameters in support vector machine and BP neural network are optimized.The results show that the error of each model is far less than 2%,and the two algorithms are compared.Finally,based on the actual operation conditions,the differential evolution algorithm is proposed to improve the efficiency by 1.0%.In this paper,a modeling and optimization method based on big data analysis is proposed for 660MW supercritical unit of a power plant in Hebei Province,which provides a certain reference for the establishment of data model and operation optimization of power plant boiler in the future.