Asynchronous model predictive control in aircraft engines

Future aircraft engines feature numerous safety constraints and a large number of control variables, making traditional PID with gain scheduling costly to design and verify while also challenging to ensure optimal performance. Model Predictive Control (MPC) can explicitly handle multivariable system constraints; however, its high computational burden poses challenges for ensuring real-time feasibility in fast dynamic systems. To address this issue, this paper proposes an Asynchronous MPC, which increases the prediction sampling time while reduces the prediction horizon, thereby reducing the number of prediction steps and subsequently lowering the optimization problem's dimensionality. This approach simultaneously enhances real-time computation and ensures fast system response. However, the relationship between sampling times, controller performance, and computational efficiency is complex. This paper presents the architecture of an Asynchronous MPC, analyzes its computational load within a single sampling period, and utilizes multi-objective optimization to determine the relationship between computational load and control performance. Results demonstrate that compared to traditional predictive controllers, the proposed method can reduce the prediction horizon by up to four times while maintaining equivalent performance, significantly reducing the computational load of real-time optimization.