SCR Flue Gas Denitrification Control Based On Fuzzy-PID And RBF Neural Network Controller

Selective catalytic reduction(SCR)flue gas denitrification control,the most widely used denitrification method in China today,can effectively reduce nitrogen oxides(NOx)emissions from coal-fired power plants.Its reaction principle lies in that ammonia is sprayed into the flue gas and reacts with NOx to produce nitrogen and water,thus making it of great importance for spray ammonia quantity control.Currently,SCR flue gas denitrification systems mainly use a PID controller to control spray ammonia quantity,but they will exhibit the characteristics of strong nonlinearity and hysteresis under variable operating conditions,resulting in difficulties in the accurate control of spray ammonia quantity.Given the increasingly stringent environmental regulations,how to implement a more effective project to control NOx emissions has become vital.The paper delivered an introduction to the current three spray ammonia flow control strategies in China,and provided a summary of some domestic and foreign research on spray ammonia flow control projects,with a simple analysis on their advantages and disadvantages.The mathematical model of the spray ammonia flow control system was established through a mathematical derivation of the ammonia flow regulating valve and SCR denitrification reactor in the SCR flue gas denitrification system,on which a simulink model of a typical SCR flue gas denitrification cascade control system was built.Furthermore,an analysis was conducted on the model through MATLAB to determine its dynamic performance and robustness.In the paper,the PID parameters were corrected in real time in accordance with the principles of parameter self-tuning fuzzy PID to make it adaptable to changes in the parameters and meet the dynamic performance indicators required by the system.The simulation results showed that the parameter self-tuning Fuzzy-PID not only has good robustness and stability,but also has the advantages of a faster response speed,a stronger anti-interference ability and fewer oscillation times than the traditional one.Additionally,the idea of RBF neural network control was employed and a neural network controller was design based on the resource allocation network(RAN network)to replace the main PID controller in the cascade system.For better simulation,the original transfer function was fitted and discretized without changing the original transfer function.The simulation results showed that in addition to good robustness and stability,the neural network control performed better than traditional cascade PID control.