Analysis of boiler operation data in a thermal power plant based on the K-means algorithm

Abstract The performance parameters of power plant operation reflect the real-time situation of unit operation. In order to ensure the accuracy of performance parameter calculation and the authenticity and reliability of calculation model establishment and system identification, it is necessary to distinguish the steady-state data and unsteady-state data collected by the power plant. Steady-state detection is performed on massive operating data, and steady-state data is extracted as the object of subsequent analysis and research. In this paper, the multi-parameter intelligent K-Means clustering algorithm is used to cluster the peak-shaving conditions of a power plant during the heating period. The results show that the number of samples is reduced by about 8 % after cluster analysis. Clustering analysis can significantly reduce the number of sample parameters and remove the large-scale fluctuation of data caused by sensor errors in the process of data acquisition, which provides a research basis for subsequent data mining.