Bed temperature modeling and forecasting of Circulating Fluidized Bed boiler based on SMO algorithm

Considering the properties of Circulating Fluidized Bed boilers (CFB) such as distribution parameters, time variation, nonlinearity, multivariable coupling, etc, A new bed temperature modeling and forecasting method of CFB based on SMO algorithm is proposed. Moreover, as to the real bed temperature data of a 300 MW coal gangue power plant, the proposed method was used to get the dynamic model. Simulation results show that this algorithm has higher prediction accuracy and faster computational speed than the traditional BP neural network algorithm; it also has a significant practical and application value to the control of bed temperature of CFB.