In order to improve the ability to predict the pulverizer fault in advance

combined with the advantages of deep learning method

a pulverizer fault early warning method based on convolutional neural network-long and short-term memory(CNN-LSTM) neural network is proposed on the basis of conventional LSTM network. The time series of coal blockage is selected as the input variables of multivariate prediction model. The deviation function between the predicted value of the model output and the operation data of the pulverizer under normal working state is obtained

and the early warning threshold is determined by using nuclear density estimation method to realize the early warning of coal blockage fault of the pulverizer. Taking the medium speed coal mill of a 660MW thermal power unit as the research object

the CNN-LSTM network model is established and the fault early warning experiment is carried out. The experimental results show that

this model can accurately predict the change trend of parameters at multiple measuring points of the coal mill

and has higher accuracy than the LSTM model. This method can give an effective early warning of coal blockage fault of pulverizer in advance.