Research on nonlinear model predictive control of regulated two-stage turbocharging system of diesel engine at high altitudes

Increasingly complex air path concepts are investigated to achieve improving the power while reducing fuel consumption of diesel engines at high altitudes. One promising technology is the Twin-VGT (variable geometry turbocharger) for diesel engines. A control concept has to be developed to exploit boost potential by coordinated management of the two turbocharger stages at different altitudes. In this paper, a nonlinear model prediction control (NMPC) algorithm based on back propagation neural network (BPNN) was proposed to purpose multi-parameter control of turbocharging system at high altitudes. Optimal control sequences of NMPC were solved by improved particle swarm optimization (PSO), and boost pressure and intake flow achieved good dynamic tracking performance by collaborative control high-pressure VGT (HVGT) and low-pressure VGT (LVGT) under whole operative conditions at high altitudes. NMPC achieved better step response performance compared with PID controller at different altitudes. NMPC control error of intake flow and boost pressure are within 0.26% under steady and transient conditions, exhibiting higher control accuracy and responsiveness even under transient operating conditions at high altitudes.