控制理论与应用2026，Vol.43Issue(8)：1639-1648,10.DOI:10.7641/CTA.2025.50016
多级高速压气机喘振失稳的确定学习建模与预警
Deterministic learning and surge warning for a multistage high-speed compressor
摘要
Abstract
Aerodynamic instability warning and safe operation monitoring of compressors are important and challeng-ing problems in high-performance aero-engine research.For issues such as weak characteristics,complex mechanisms,and the difficult-to-predict nature of multistage high-speed compressor surges,this paper presents a unified learning and warning method based on single-sensor data.Firstly,a sampled-data observer-based deterministic learning algorithm is used to identify the compressor dynamics,forming a pattern library that characterizes the dynamic information of surge evolution.Subsequently,this library is used to design dynamical estimators,which utilize real-time input from single-sensor data to achieve surge warning.Finally,offline and online warning experiments at different speeds are carried out on a five-stage high-speed compressor test rig.Preliminary experimental results show that based on the single-sensor data,the proposed method can achieve a technological breakthrough from millisecond-level detection after the surge occurs to second-level warning before the surge occurs.In summary,the proposed learning method can utilize compressor data of the multistage high-speed compressor to obtain the essential and comprehensive dynamic characteristics of the compressor surge,which has favorable interpretability.The warning method developed on this basis only uses single-sensor data to achieve online early detection of sudden surges.It is expected to provide a real-time monitoring method for the safe and stable operation of high-performance aero-engines,which has a certain application value.关键词
多级高速压气机/气动失稳/动态系统/确定学习/径向基函数网络Key words
multistage high-speed compressor/aerodynamic instability/dynamical system/deterministic learning/radial basis function network引用本文复制引用
胡竞涛,潘若痴,王聪,吴伟明,张志博,张付凯,朱泽键,韩帅,陈禹西,贾博博,刘世官..多级高速压气机喘振失稳的确定学习建模与预警[J].控制理论与应用,2026,43(8):1639-1648,10.基金项目
中国博士后科学基金资助项目(2024M761803),国家资助博士后研究计划项目(GZC20231451),山东省博士后创新项目(SDCX-ZG-202400315),国家自然科学基金项目(62350083,62203262,62203263)资助.Supported by the China Postdoctoral Science Foundation(2024M761803),the Postdoctoral Fellowship Program of CPSF(GZC20231451),the Shan-dong Postdoctoral Science Foundation(SDCX-ZG-202400315)and the National Natural Science Foundation of China(62350083,62203262,62203263). （2024M761803）
