ANN/GA-ANN modeling study on operating parameter prediction for waste-to-energy plant

Automatic combustion control (ACC) system is the main control system of the current municipal solid waste (MSW) incineration power plant. In this work, the artificial neural network (ANN) model was first used for prediction and guiding of operation of the incineration system. Results showed that BP-ANN model was an effective tool for operation parameter prediction, especially the crucial parameters of steam mass flow and furnace temperature. The Log-Sigmoid function was more suitable to be used as the transfer function in hidden layer. Comparison of the input parameter showed that furnace temperature, grate speed, primary air, and secondary air are the most important and relevant parameters for prediction of steam mass flow in MSW incineration system. The genetic algorithm (GA) could be used to optimize the initial weight and threshold values of the BP-ANN network, which largely improved the prediction accuracy. Impact weight analysis results show that furnace temperature and secondary have the largest influence on the steam flow. With regard to furnace temperature, secondary air, primary air, and oxygen concentration in exhaust gas play the leading roles. The largest error for steam mass flow prediction of the real-time operation was lower than 8%.