Application Of Fuzzy Clustering Algorithm In Fault Diagnosis Of Gas-steam Combined Cycle Unit

Waste heat boiler is one of the main equipment of gas-steam combined cycle power generation,which is used to recover the waste heat and exhaust heat of gas turbine and improve the energy conversion efficiency of the whole unit.It is of great economic and social benefit to find and eliminate the failure of waste heat boiler soda water system in time and develop the fault technology of waste heat boiler soda water system.With the development of science and technology and the rapid progress of computer technology,many experts and scholars at home and abroad have done a lot of work in the leakage fault diagnosis of waste heat boiler.Common for waste heat boiler pressure soda system economizer leakage,leakage evaporator and superheater leakage fault,this paper proposes a hierarchical progressive fault diagnosis method based on fuzzy clustering,fuzzy clustering algorithm to get the first layer normal working condition of standard samples,the selection based on pattern recognition principle of nearly whether soda system leakage failure;In the second layer,the fuzzy clustering algorithm is combined with the nearest neighbor algorithm to diagnose the specific location of the fault.In the third layer,based on the first and second layers,the secondary feature screening is carried out first,and the standard vector center is obtained through fuzzy clustering.The distance threshold of the fault degree is calculated by using multiple groups of historical fault data to identify the severity of the leakage fault.In this process,the traditional fuzzy C-means algorithm(FCM)is improved in this paper.Aiming at its sensitive initial parameters and easy local convergence,a hierarchical genetic optimization of FCM initial clustering center is proposed,and the Gaussian diameter kernel function is used to replace the Euclidean distance formula,which effectively improves the clustering performance.In this paper,with the help of a simulator,simulation experiments were carried out on an 960 MW waste heat boiler with gas-steam combined cycle.A large number of simulation experiments were conducted on various leakage failures of waste heat boiler soda water system,and the variation rules of its parameters were studied and a large number of stock index samples were extracted.The hierarchical progressive fault diagnosis system was established by combining fuzzy clustering with pattern recognition based near selection principle and nearest neighbor algorithm.The diagnosis program was written in Matlab environment,and the real-time diagnosis was completed by communicating with the simulator.The effectiveness of the method was verified.