隔离段作为连接双模态超燃冲压发动机进气道和燃烧室的重要部件，用于稳定预燃激波串，避免进气道不起动。隔离段激波串的非定常特性使得发动机承受着较大的载荷波动，导致发动机稳定性难以满足飞行要求，因此研究隔离段激波串非定常特性的影响是十分必要的。脉动反压下的受迫振荡和稳态反压下的自激振荡是导致隔离段激波串非定常现象的两个重要原因。本项目将利用数值仿真和实验研究方法，分别对受迫振荡和自激振荡开展研究工作，搭建基于人工神经网络的非定常特性预测模型，并提出人工神经网络模型与全局灵敏度分析相结合的方法。探索反压脉动参数与激波前流动参数对激波串受迫振荡特性的影响规律，分析激波串结构中的脉动压力来源，揭示受迫振荡的形成机理；探索激波前流动参数和激波后反压对激波串自激振荡特性的影响规律，分析对自激振荡特性贡献最大的压力点及多压力点耦合作用，揭示自激振荡的形成机理。

As the key part for dual-mode scramjet, the isolator is used to connect supersonic inlet and combustion chamber. The isolator is mainly used to stabilize the shock train upstream of the combustion chamber, and it can avoid the unstart phenomenon. The unsteady characteristics of shock train in isolator lead to major load fluctuation of the engine. As a result, the stability of the engine is difficult to meet the flight requirement. So it is necessary to study the influence of the unsteady characteristics of shock train in isolator. The forced oscillation under fluctuating back pressure and the self-excited oscillation under fixed back pressure are the important reasons of the unsteady phenomenon in isolator. The project is focusing on unsteady characteristics of forced oscillation and self-excited oscillation by numerical simulation and experimental research. The prediction models of unsteady characteristics have been made based on artificial neural network. A method combining artificial neural network with global sensitivity analysis is proposed. The influences of fluctuating back pressure and upstream flow parameters on unsteady characteristics of forced oscillation are investigated. The fluctuation sources and the forming mechanism of forced oscillation are revealed. The influences of upstream flow parameters and back pressure on unsteady characteristics of self-excited oscillation are investigated. The coupling effects of the wall pressure points on the unsteady characteristics of self-excited oscillation are studied. The fluctuation sources and the forming mechanism of self-excited oscillation are found.