Artificial Intelligence-based Modeling and Control for Scramjet Engines

Hypersonic weapon systems are a strategic need for the USAF and DoD. One particular challenge is the development of efficient and reliable air breathing engines such as scramjets for hypersonic systems. To address this need, we propose to develop an artificial intelligence-based control system for scramjets. The controller will build upon SE’s existing control strategies of regulating the fuel-air equivalence ratio delivered through the injectors to adjust the leading-edge position of the shock train in the isolator of scramjet engines. The controller will focus on making engine operation robust against nonlinearity inherent in system models varying with input signal in addition to disturbances. Since existing control schemes that do not properly adapt to such uncertainty and disturbances easily fail to stabilize the control system, an advanced control method using AI/machine learning will be developed to compensate for the nonlinearity of the scramjet. Ultimately, this will mitigate inlet unstart, which is a phenomenon where perturbations inside a hypersonic inlet cause the shock structure to move upstream of the throat and out of the inlet. Mitigation of inlet unstart will make the scramjet robust over a large range of flight regimes and enable high-maneuverability while still providing excellent efficiency and durability.