基于樽海鞘群极限学习机的进/发一体化性能寻优控制模型研究  

An integrated inlet/engine performance seeking control modelbased on salp swarm algorithm extreme learning machine

作　　者：于子洋 王晨 杜宪 聂聆聪[2] 孙希明 YU Ziyang;WANG Chen;DU Xian;NIE Lingcong;SUN Ximing(School of Control Science and Engineering,Dalian University of Technology,Dalian 116024,China;Beijing Power Machinery Institute,Beijing 100074,China)

机构地区：[1]大连理工大学控制科学与工程学院,辽宁大连116024 [2]北京动力机械研究所,北京100074

出　　处：《推进技术》2024年第5期236-249,共14页Journal of Propulsion Technology

基　　金：国家自然科学基金(61890921,61890924);国家科技重大专项(J2019-I-0019-0018);中央高校基本科研业务费(DUT22QN204)。

摘　　要：为充分发挥航空推进系统的性能,提高性能寻优控制的实时性,将樽海鞘群算法(SSA)与极限学习机(ELM)相结合,基于进/发一体化部件级模型建立数据集,提出一种基于SSA-ELM的数据驱动模型。将该建模方法与广义回归神经网络(GRNN)、BP神经网络(BPNN)和极限学习机(ELM)比较,结果表明,相比于BPNN,ELM,GRNN,SSA-ELM用于预测可以使安装推力的均方根误差(RMSE)分别降低7.41%,17.01%,72.57%,安装油耗的RMSE分别降低4.32%,19.41%,66.77%,具有更高的预测精度。将基于SSA-ELM的数据驱动模型作为机载模型应用到性能寻优控制,结果表明,该机载模型能够维持理想的寻优效果。针对最大安装推力模式开展实时性分析,该机载模型相比于进/发一体化部件级模型,平均计算时间由184.05 ms缩短至1.357 ms,实时性得到显著改善,大大提高了寻优效率。To fully exploit the performance of the aero-propulsion system and improve the real-time perfor⁃mance of performance seeking control,a data-driven model based on SSA-ELM is proposed by combining Salp Swarm Algorithm(SSA)and Extreme Learning Machine(ELM)and establishing data sets based on integrated in⁃let/engine component-level model.The modeling method was compared with General Regression Neural Network(GRNN),BP Neural Network(BPNN)and Extreme Learning Machine(ELM).The results show that compared with BPNN,ELM and GRNN,SSA-ELM for prediction can reduce the Root Mean Square Error(RMSE)of in⁃stalled thrust by 7.41%,17.01%,and 72.57%,respectively,and the RMSE of installed fuel consumption by 4.32%,19.41%,and 66.77%,respectively,which has higher prediction accuracy.The data-driven model based on SSA-ELM was applied to the performance seeking control as on-board model.The results show that the on-board model can maintain the ideal seeking effect.In the real-time performance analysis of the maximum in⁃stalled thrust mode,the average computation time of the on-board model is reduced from 184.05 ms to 1.357 ms compared to the integrated inlet/engine component-level model,which significantly improves the real-time per⁃formance and greatly enhances the seeking efficiency.

关 键 词：航空发动机 进/发一体化 樽海鞘群优化算法 极限学习机 数据驱动模型 性能寻优控制 

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]