Prediction of Boiler Control Parameters Based on LSTM Neural Network

A prediction method of main control parameters of boiler based on long short-term memory neural network (LSTM) model was proposed in this paper. The maximum information coefficient (MIC) is used to analyze the nonlinear correlation variables in the boiler, and the optimal input feature set is determined with the prediction accuracy as the goal. Combined with engineering practice, multiple LSTM step-by-step prediction models are established between input and output variables. Based on the maximum information coefficient, the time delay estimates between output parameters of each algorithm are obtained. The method is verified by the data of the coal water slurry boiler in a thermal power company. The results show that the method has high prediction accuracy and can meet the needs of boiler parameter prediction.