The present article proposes an active observation speed prediction control algorithm architecture for embedded applications, with the aim of addressing the problems of complex operating conditions, strong perturbations, and high control real-time requirements of high-pressure direct injection (HPDI) dual-fuel engines. A nonlinear speed prediction model with diesel and natural gas injection mass as inputs has been established, and the nonlinear model predictive control (NMPC) method is used to realize the optimized control of engine speed. In order to enhance the operational efficiency of the algorithm on the embedded platform, a system has been developed that includes an event triggering mechanism and a warm-start strategy. These mechanisms work in tandem to dynamically adjust the computation cycle. Additionally, a torque reduced-order expansion state observer (RESO) has been integrated to improve the accuracy of perturbation estimation and computational efficiency. The model-level experiments and hardware verification were carried out under the sudden load change operating condition and World Harmonized Transient Cycle (WHTC) test, respectively. The simulation results demonstrate that the proposed optimization strategy can effectively reduce the peak-to-peak value of speed control error to 118.73 rpm and shorten the stabilization time to 3.48 s. Furthermore, the tracking accuracies of the controller on the speed and torque targets in the hardware test reach 0.994 and 0.997, respectively, thereby substantiating the high accuracy and robust performance of the proposed algorithm.
Yang et al. (2026) studied this question.