Deep neural network based the oxygen content of boiler flue gas

The oxygen content of boiler flue gas is a highly important indicator of boiler efficiency. Current measurements have high cost but low accuracy, in this paper a deep neural network (DNN) algorithm is proposed to solve these problems. Firstly, the variables from distribution control system highly related with oxygen content are selected as inputs of the prediction model by a lasso algorithm. Then, the prediction models based on deep neural network are established with the inputs selected before. The experimental data are obtained from the historical data of coal-fired power plants. Results compared with 3 data-driven algorithms, and DNN algorithm gets better prediction accuracy with mean absolute percentage error of 1.2536.