LAUSR

Abstract The operation state of coal mill is related to the security and stability operation of coal-fired power plant. In this paper, a fault diagnosis method of coal mill system based on the simulated typical fault samples is proposed. By analyzing the fault mechanism, fault features are simulated based on the model of coal mill, and massive fault samples are obtained. Then, according to the time series difference of different types of fault, the stacked automatic encoder is established to reduce the dimension of fault data. Furthermore, by learning and classifying the features of fault data after dimensionality reduction, the fault identification of coal mill system can be realized. Simulation results show that the model-based fault simulation can effectively solve the problem of fault samples acquisition, the proposed fault diagnosis method has highly accuracy and can effectively monitor the running status of the coal mill system.