Research On SCR System Control Strategy Based On DSPACE

With the development of ocean transportation,the problem of exhaust emissions from marine diesel engines has become increasingly serious.International Maritime Organization（IMO）regulations for limiting ship’s nitrogen oxides（NOx）are becoming stricter,and the selective catalytic reduction technology（SCR）can effectively reduce NOx emissions from ships.SCR system control technology has become an important research topic in the field of marine diesel engine NOx emissions.In this paper,the SCR electronic control system are designed based on Siemens PLC S7-1200 and setting up the SCR system experimental bench.Aiming at the SCR catalytic reaction model and the urea injection control technology,perform an open-loop parameter identification experiment and a rapid control prototype experiment with this bench.The main contents and the main results are as follows:Analyzing each chemical reaction process in the SCR catalytic reaction process under the premise of simplifying certain assumptions,SCR catalytic reaction model was established.Through the open-loop parameter identification experimental data,the nonlinear least squares method and particle swarm optimization algorithm were used to solve the unknown parameters identification problem in SCR catalytic reaction model,and the identification results were verified using experimental data under different operating conditions.It was proved that particle swarm optimization algorithm can accurately identify the unknown parameters of SCR catalytic reaction model and improve the accuracy of SCR catalytic reaction model,the SCR catalytic reaction model can accurately predict the NOx concentration and NH₃ concentration of SCR reactor outlet.Aiming at the difficulty in finding the optimal PID parameters for PID control,the BP neural network PID control was proposed;and in response to the problems of high urea injection volume and high ammonia escape in the early stage of closed-loop control,feedforward control and BP neural network PID combined control Methods were proposed.Simulation results showed that the BP neural network PID control had better control effect than PID control,and the comprehensive control effect of the feedforward control and BP neural network PID control effectively solved the problem of cross-sensitivity of NOx sensor caused by early overspray and had better overall control.Based on the development process of the V mode,a SCR semi-physical simulation platform system was built based on d SPACE.The rapid control prototype（RCP）was used to test the designed control algorithm in real machine,which verified the effectiveness of the control strategy.