Abstract
In order to elaborate and predict the NO_x emission characteristics, N⇾NO_x of six representative single-component waste variation with bed temperature and excess air was studied in the ^(∳)150mm fluidized bed. The results show paper and wood have the maximum, but rubber and plastics have the minimum N⇾NO_x. Fluidized bed temperature has more influence on the coal and rubber than other waste because its fuel-nitrogen chemical construction is stable, N⇾NO_x of every waste increases with the excess air coefficient increasing. We construct a 12×7×1 back-propagation neural network according to a lot of experiment analysis. Stop criteria MSEREG=49 could reach after 186 iterations. SCG algorithm was designed because of its good performance. The simulation results indicate approximation error between actual values and simulation value of outputs is relatively low for most patterns. This indicates BP neural network has been trained well and it has good generality capability and memory capability while in usual working parameters.
