一种双回路VCE加速控制计划优化方法研究  

Research on Acceleration Control Schedule Optimization Based on Dual-loop VCE

作　　者：赵兴宇 李秋红[1] 庞淑伟 顾子渝 刘鑫洋 ZHAO Xingyu;LI Qiuhong;PANG Shuwei;GU Ziyu;LIU Xinyang(College of Energy and Power,Nanjing University of Aeronautics and Astronautics,Nanjing 210016,China)

机构地区：[1]南京航空航天大学能源与动力学院、江苏省航空动力系统重点实验室,南京210016

出　　处：《计算机测量与控制》2025年第3期113-123,共11页Computer Measurement &Control

基　　金：中国航发产学研合作项目(HFZL2021CXY007);航空发动机及燃气轮机基础科学中心项目(P2022-B-V-002-001)。

摘　　要：为提高变循环发动机加速性能,提升推力的响应速度,提出了内外回路耦合的加速控制计划优化方法;内回路采用模型预测控制,以增加乘法层的神经网络结构在线构建线性化预测模型,采用交替方向乘子法对约束条件下的闭环控制变量进行优化;在外回路,采用樽海鞘链群智能算法,以推力响应最快为目标,优化由贝塞尔曲线构造的开环控制规律;内回路的优化基于外回路开环几何机构的调节规律展开,外回路的优化以内回路的推力响应为评价,形成内、外回路耦合优化结构;以最优个体对应的发动机输入输出构造基于换算油气比的加速控制计划以及开环几何机构控制计划,开展加速过程的仿真验证;结果表明,在线学习预测模型5步预测最大动态误差低于0.8%;相比原有控制计划,优化后的控制计划充分利用了开、闭环控制变量的有利耦合,加速过程中,转速的响应时间缩短14.6%以上,推力响应时间缩短15%以上,验证了所提出的控制计划优化方法的有效性。To improve the acceleration performance of variable cycle engines(VCEs),and enhance the response speed of thrust,an acceleration control schedule optimization method based on coupled internal and external loops is proposed.In the inner loop,the neural network state space model with an additional multiplication layer was used to construct the prediction model online,and the alternating direction multiplier method was used to optimize the closed-loop control variables under constraint conditions.In the external loop,by using the salp swarm algorithm,the open-loop control law constructed by the Bessel curve was optimized with the goal of thrust response speed.The inner loop was optimized by the regulation law of the open-loop geometric mechanism and the external loop was optimized by the thrust response of the inner loop,which formed a coupled optimization structure.The control schedules for the corrected fuel-air ratio acceleration and open-loop geometric mechanism are constructed through the input and output of the engine corresponding to the optimal individual,achieving the simulation of acceleration process experiments.The results show that the maximum 5-step output prediction error of the online learning prediction model is below 2%.Compared with the original control schedule,the optimized acceleration control schedule makes full use of the favorable coupling of the open and closed loop control variables,and in acceleration process,the response time of rotational speed is shortened by over 14.6,and the response time of thrust by over 15%,which verifies the effectiveness of the proposed control schedule optimization method.

关 键 词：变循环发动机 加速控制计划 神经网络状态空间模型 贝塞尔曲线 模型预测控制 

分 类 号：V233.7[航空宇航科学与技术—航空宇航推进理论与工程] TP273[自动化与计算机技术—检测技术与自动化装置]