Research On Closed-loop Control Of Diesel Engine Combustion Based On Model Prediction

Diesel engine will still be the main power plant of ships for a long time because of its advantages of high thermal efficiency and strong adaptability.The market has higher and higher fuel economy index of diesel engine,and the society has more and more strict regulations on emissions.The traditional open-loop combustion control technology cannot avoid the combustion imbalance in the process of diesel engine operation.The coupling phenomenon between multiple control rings of PID combustion closed-loop controller will lead to the degradation of diesel engine performance and emission deterioration,which cannot meet the requirements.Model predictive control(MPC)is one of the most effective and feasible technology paths to realize closed-loop control of diesel engine combustion and solve combustion non-uniformity problem by considering the interaction between system variables.Based on the above background,this paper will explore the application of MPC algorithm in Marine diesel engine combustion closed-loop control to reduce combustion unbalance and improve engine dynamic performance.In view of the characteristics of the diesel engine,such as highly nonlinear and strong coupling,this paper selects the neural network algorithm as the identification method to collect the data of the design working conditions of a certain six-cylinder diesel engine.Through the data set training,the diesel engine performance prediction model meeting the requirements is obtained.The nonlinear prediction model of diesel engine performance is linearized by using Taylor expansion to solve the problem of nonlinear model solving.Further combined with the actual situation of the diesel engine,the cost function is designed,the basic parameters of the controller are set up,the control strategy is formulated,and the MPC controller is designed.The solution of the controller is transformed into a standard quadratic programming(QP)problem which is easy to be solved by computer,and a diesel engine mechanism model at ignition scale is established to verify the designed controller off-line.The tracking speed,tracking CA50 and anti-jamming ability of the controller are verified offline by simulation.The results show that the controller has better control performance.A semi-physical simulation platform was built based on c RIO and PXI,and six experimental conditions were designed to verify the controller online.The experimental results show that the diesel engine under MPC control has a smooth transition from high-speed start-up to idle running,no overshoot,no static error in steady state,and the fluctuation of steady speed is less than 1% of the target value.The MPC controller has better tracking ability for speed and CA50 than PID controller,and the MPC has stronger robustness under sudden load increase and sudden load reduction.In terms of reducing combustion unevenness,the variation coefficient and unevenness of CA50 under MPC control are significantly reduced compared with PID control and open-loop control.By comparing with PID,MPC algorithm is proved to have certain ability of automatic decoupling.The above results show that the MPC controller has the advantages of good tracking performance,strong robustness,automatic decoupling ability,and significantly reducing the non-uniformity and cyclic variation between cylinders in the combustion process in the double closed-loop control of CA50 and rotational speed.