Nonlinear Model Predictive Control for Premixed Turbocharged Natural Gas Engine
Qichangyi Gong, Jinbang Xu, Jie Ye, Feng Han, Anwen Shen
Abstract
In this article, a nonlinear model predictive control (N-MPC) strategy for premixed turbocharged natural gas engine has been proposed. It is applied to genset and the control objective is to track the target speed and air-fuel ratio (AFR), while resisting load disturbance. The main control difficulties are the engine’s strong couplings and variable large AFR delay, and conventional methods (PI) cannot solve these problems well. A premixed fuel model, which can specifically reveal the couplings, has been proposed in this article. N-MPC controller uses this model combined with AFR prediction, coordinating the actuators according to a specified cost function, showing advantages in dealing with this problem over conventional methods. The nonlinear optimization control problem is solved by classic Wilson–Han–Powell sequential quadratic programming method combined with a modified warm start strategy, which uses input compensation to reduce the number of calculation iterations. The simulation results show that the N-MPC has faster response than linear parameter varying model predictive control (LPVMPC), and the warm-start strategy significantly improves computing efficiency. The N-MPC is also verified and compared with conventional methods (PI) and LPVMPC in experiments, which proves that N-MPC is superior to LPVMPC in speed transient performance, and the two MPC are much better than PI in terms of AFR control.