The accurate parameter prediction of a flow field is of practical significance to promote the development of hypersonic flight. Velocity field prediction using deep learning is a promising method to provide an accurate velocity field in a scramjet isolator. A new approach for the velocity field prediction in a scramjet isolator is developed in this study. A data-driven model is proposed for the prediction of the velocity field in a scramjet isolator by convolutional neural networks (CNNs) using measurements of the pressure on the isolator. Numerical simulations of flow in a three-dimensional scramjet isolator at various Mach numbers and backpressures are carried out to establish the dataset capturing the flow mechanism over various operating conditions. A CNN architecture composed of multiple reconstruction modules and feature extraction modules is designed. The CNN is trained using the computational fluid dynamics dataset to establish the mapping relationship between the wall pressure on the isolator and the velocity field in the isolator. The trained model is then tested over various Mach numbers and backpressures. The data-driven model successfully learns the relationship between the velocity field and pressure experienced on the wall of the isolator, i.e., the trained CNN model successfully reconstructed the velocity field based on the wall pressure on the isolator with high accuracy.

The oscillation characteristics of the shock train in an isolator have been investigated in a direct-connect wind tunnel at Mach 2.7. High-speed schlieren imaging and high-resonance frequency pressure measurements were used to capture the flow features during the shock train movement. The oscillation features without the effects of incident shocks were analyzed first. As the shock train moved upstream, the low-frequency part of the oscillation was found to develop. The analysis was then extended to complex situations with incident shocks. It was revealed that the shock wave-boundary layer interactions considerably influence the shock train behavior. The interactions were classified into three patterns: (I) single interaction, (II) multi-interactions on the same side, and (III) multi-interactions on different sides. Experimental results indicated that the oscillation could be affected in temporal scale by pattern II and enhanced in spatial scale by pattern III. The data also showed that the pressure rise induced upstream propagates to the exit, causing phase offsets in the wall pressure histories and making the pressure distributions diverge from their stable state. This phenomenon suggested a possible physical mechanism for the oscillation during shock train movement, which was verified by additional tests with large backpressure rising rate. It was found that there exists a critical frequency which is related to the pressure ramping rate during the oscillation. If the dominant frequency of the backpressure varies beyond this critical frequency, the pressure distribution could be forced into a steady state before the oscillation was induced. Otherwise the oscillation could not be suppressed.

Oblique shock waves are unavoidable in a rectangular hypersonic inlet, leading to a non-uniform flow field. While a significant body of the literature exists regarding the shock train modeling in a uniform incoming flow condition, few efforts have focused on the shock train behavior considering the influence of the shock wave boundary layer interactions. A low-order dynamic model of the shock train has been constructed with the help of the free interaction theory and a 1-D analysis approach. Experimental and numerical investigations have been carried out to evaluate the low-order model. The results show that the model has the capability of qualitatively analyzing the shock train behavior. In the cases with incident shocks, the rapid forward movement of the shock train has been observed by experiment. Besides this phenomenon was also modeled using the low-order model. Schlieren images show that when the shock train approaches the interaction zone, its behavior is characterized by oscillation and then follows a rapid forward movement with a linear increasing backpressure at 2.7 Ma. This phenomenon is analyzed theoretically based on the free interaction theory. Meanwhile with the help of the direct numerical simulation results from some existing studies, the flow structures in the interaction region and the following boundary layer also provide the evidence.

Low-frequency unsteadiness is investigated through wind tunnel experiments and numerical simulations of the internal flow in a supersonic isolator with background waves generated by a 14° wedge in a freestream with a Mach number of 2.94. The power spectra, coherence, and phase analyses of high-frequency pressure signals and schlieren images provide a local and global description of the unsteadiness. The upstream mechanism exhibits a significant influence on the unthrottled flow field. In the weak interactions of small separation flow, the pressure fluctuation between two adjacent incident points has a strong correlation in a large frequency range, while only large-amplitude shock oscillations are exhibited in the pressure fluctuations at the boundary layer. The downstream mechanism dominates the asymmetric shock motion in the throttled flow field. The profiles of the power spectrum and standard deviation both exhibit two peaks at the upstream and downstream peripheries of the wall separation patterns. Two types of oscillations can be identified through the pressure data, and type III is established from the analysis of schlieren images. The oscillation behavior of the three types is obtained through the power spectral analysis of a series of schlieren snapshots. The frequency of the occurrence and the one-cycle amplitude of different oscillation types are significantly different. By combining the coherence and phase analyses with the corresponding schlieren images and pressure data, the feedback mechanism of the three oscillation types is determined. This study combines the low-frequency unsteadiness in supersonic internal flows with the multiple separation regions caused by complex background waves.

A general super-resolution (SR) reconstruction strategy is proposed to address the super-resolution reconstruction of temperature fields from low-resolution coarse temperature field data using convolutional neural networks. Two deep learning (DL) models were applied to augment the spatial resolution of temperature fields. One is the classical super-resolution convolutional neural network, and the other is the novel multiple path super-resolution convolutional neural network (MPSRC). Three paths with and without a pooling layer are designed in the MPSRC to fully capture spatial distribution features of temperature. Numerical simulations of combustion in a strut scramjet combustor at various Mach numbers are carried out to establish a dataset for network training and testing. The corresponding high-resolution temperature fields were successfully reconstructed with remarkable accuracy. The reconstruction performances of those models were comprehensively investigated and compared with the bicubic interpolation method. The results demonstrated that both DL methods can greatly improve the super-resolution reconstruction accuracy and the MPSRC can provide a better reconstruction result with a lower mean square error and a higher peak signal-to-noise ratio.

The pursuit of high-resolution flow fields is meaningful for the development of hypersonic technology. Flow field super-resolution (SR) based on deep learning is a novel and effective method to provide HR flow fields in a scramjet isolator. Single-path and multiple-path network models based on convolutional neural networks (CNNs) have been developed to augment the spatial resolution of the experimental supersonic flow field. The single-path model uses a simple convolutional layer and fully connected layer serial architecture, and the multiple-path model increases the branch path by adding pooling layers to achieve a fusion structure architecture. Ground experiments of flow in a supersonic isolator at various working conditions are conducted to establish an experimental dataset. The trained single-path and multiple-path CNNs are compared with the traditional interpolation method on the flow field SR reconstruction accuracy. The results demonstrated that single-path CNNs have certain learning ability, but the SR accuracy is not satisfactory; multiple-path CNNs significantly improve the accuracy of flow field SR, and the multiple-path CNN with one branch path achieves the best SR performance.