Chen Kong

Rapid flowfield prediction is essential for scramjet-related hydrogen injection and mixing under varying Mach numbers. This study develops a physics-informed neural network for a Deutsches Zentrum für Luft- und Raumfahrt-type supersonic strut-injector configuration under non-reacting hydrogen–air mixing conditions. Validated two-dimensional Reynolds-averaged Navier–Stokes simulations are used to generate reference data. Taking spatial coordinates and incoming Mach number as inputs, the network predicts pressure, temperature, velocity components, and species mass fractions. Continuity and momentum residuals are combined with the supervised loss in the composite training objective to improve physical consistency. A Schlieren-informed stratified weighted sampling strategy further emphasizes shock-dominated and high-gradient regions. Results show that partial differential equation constraints enhance extrapolative robustness, suppress nonphysical artifacts, and better preserve flowfield morphology under unseen Mach numbers. The proposed method provides an effective physics-constrained surrogate for complex supersonic mixing flows.

Accurate reconstruction of transient velocity fields in supersonic isolators under variable operating conditions is critical for understanding shock train dynamics and flow unstart mechanisms. Wall pressure and Schlieren measurements encode complementary physical information about shock–boundary layer interactions and flow separation: pressure signals capture the temporal evolution of streamwise pressure gradients driving shock migration, while Schlieren images reflect spatial density gradients associated with shock structure morphology. Fixed-weight multi-source fusion models fail to adapt to the dynamic evolution of transient supersonic flows, suffering marked accuracy degradation under variable operating conditions. To address this limitation, a condition-guided dynamic weighting mechanism is proposed, which adaptively allocates contribution weights between Schlieren images and wall pressure signals via a gating module. Leveraging dynamic features from Mach number (Ma)/back pressure (BP) variation rates and amplitudes, weights adjust dynamically with flow evolution and directional specificity: pressure feature weight consistently exceeds Schlieren's, decreasing during Ma rising phases and increasing during falling phases, and is higher for y-direction than x-direction velocity reconstruction. Quantitative validation shows the proposed model reduces x/y-velocity root mean square error (RMSE) by 8.2%–11.5% relative to fixed-weight fusion models and outperforms competing approaches (convolutional neural network-long short-term memory, gated recurrent unit, traditional fusion baselines) with 5.3%–31.4% lower RMSE in velocity field reconstruction. This work demonstrates that physics-guided adaptive multi-modal fusion enables physically interpretable reconstruction of complex shock-dominated flows, supporting real-time flow state estimation and unstart avoidance in supersonic propulsion systems.

This study investigates the flow field behavior during the mode transition of a turbine-based combined cycle (TBCC) inlet, focusing on the coupling dynamics between turbine and ramjet channels under backpressure conditions. It reveals the dual-channel interaction mechanism and identifies three distinct coupling phenomena: when both channels are started, they operate independently; when the ramjet channel is unstarted, it has no effect on the upstream flow or turbine channel; when the turbine channel is unstarted, it alters the flow structure of the ramjet channel. This study also examines flow evolution and performance parameter changes during both forward and reverse transitions. Under throttled ramjet conditions, two distinct flow behaviors emerge depending on the initial state. In forward transitions, when the inlet is starting, hysteresis occurs in the ramjet's shock train movement, and the performance trends of the turbine and ramjet channels exhibit strong symmetry. However, when the ramjet channel fails to start, the inert flow field cannot return to its original state, and this symmetry is lost.