According to the multivariable coupling、 large time delay, non-linearity and time-varying and other difficulties of circulating fluidized bed boiler combustion system, a kind of control technology based on neural network to circulating fluidized bed boiler combustion system was presented. Actual parameter data of a paper mill in Kunming and neural network control principle were used in the establishment of a circulating fluidized bed boiler combustion system mathematical model and modified BP neural network algorithm training. Results of MATLAB simulation show that boiler combustion system control precision was effectively improved and good effects in production and application were got.
For study on SO2 and NO emission characteristics of circulating fluidized bed boiler experiments were carried out in a 300MW circulating fluidized bed boiler. In these experiments, the variables were the amount of limestone and bed temperature. While conducting the test, the reasons for the changes in emission were analyzed, and some advices about combustion adjustment were proposed.