涡轮增压汽油机气路预测模型的建立与预测控制
Airpath prediction model and predictive control of turbocharged gasoline engine

针对涡轮增压汽油机气路系统中节气门与废气旁通阀动力学耦合、机理建模复杂的问题, 本文提出基于神经网 络模型的气路系统预测控制方法，实现了节气门与废气旁通阀的协调控制. 首先, 针对涡轮增压汽油机气路系统map与 机理混合描述的特性, 利用系统的输入输出数据，采用反向传播神经网络(back propagation neural network, BPNN)训 练得到一个非线性气路模型; 其次, 基于泰勒展开式对预测模型进行线性化, 并对模型的精度进行了验证，进而利用该 模型预测系统的未来动态; 然后, 在考虑系统存在输入约束的条件下, 设计了一个线性模型预测控制器对节气门与废气 旁通阀进行协调控制, 实现了进气歧管压力和升压的跟踪控制进而满足发动机的扭矩需求; 最后, 通过离线仿真和基 于dSPACE的快速原型实验(rapid control prototyping, RCP)验证了控制系统的有效性和实时性.
In this paper, a neural network based model predictive controller is developed for the coordinated control of the throttle and wastegate in a turbocharged gasoline engine airpath system. Firstly, considering the mixed description of map and physical for engine airpath system, a data-driven nonlinear airpath model is trained using back propagation neural network (BPNN) to predict the future dynamics of the turbocharged engine. Secondly, the prediction model is linearized based on Taylor expansion and the feasibility of this simplification is assessed. Thirdly, in order to satisfy the engine torque demand, a linear model predictive controller is designed to manage the throttle and wastegate so that the engine tracks the setpoints of the intake manifold pressure and boost pressure considering the system constraints. Furthermore, simulation results are presented to verify the effectiveness of the controller. Finally, a rapid control prototyping (RCP) experiment based on dSPACE is further implemented to test the real-time performance.