A predictive control method for selective catalytic reduction of(SCR) flue gas denitrification is proposed

which uses the radial basis(RBF) neural network of particle swarm optimization to predict the mass concentration of NOx at entrance of the SCR denitrification system. The overall optimization space is discretized. The optimization model is reconstructed by combining node optimization with the fastest gradient method. Meanwhile

the prediction results of the RBF neural network are introduced into the denitrification control strategy

and applied to ammonia injection control of thermal power units. The practical application of this new developed control strategy in a 350 MW unit show that

this method can not only realize the automatic control of ammonia injection valves

but also improve the control precision of the ammonia injection system.