Turbulent combustion is one of the key processes in many energy conversion systems in modern life. In order to improve combustion efficiency and suppress emission of pollutants, many efforts have been made by scholars to investigate turbulent flames. In the present study, Artificial neural network (ANN) was first employed for the storage and interpolation of the flamelet library in flamelet generated manifolds (FGM) model, in which Eulerian stochastic field (ESF) model was used to directly consider the probability density function of the control variables. This new model had been implemented in OpenFOAM and was validated by simulation of the Sandia Flame D under consideration of the detailed chemical reaction mechanism. By comparing the results of numerical simulations and experimental measurements of the temperature and the mass fraction of main components, the accuracy of the proposed ANN-ESFFGM model was verified. Through the use of ANNs to characterize the chemical reactions, the flame simulation accuracy of the new model is higher than that of the original ESFFGM model, especially in the prediction of the ignition position. With the increase in the number of stochastic fields, the simulation accuracy of the new turbulent combustion model is continuously improved until a certain value of stochastic fields was reached. Moreover, excessively high FGM table resolution has limited improvement in numerical simulation accuracy.

Large eddy simulation is used to study the gas-particle two-phase transverse jet in supersonic flow. In this paper, the Eulerian method is used to simulate the fluid, and the Lagrangian method is used to simulate the particles. The numerical study is carried out for different particle mass fractions and Stokes numbers with the jet-crossflow momentum flux ratio and the Mach number of crossflow being kept constant. It is found that the larger particles (St = 5) are mainly located above the counter-rotating vortex pair, and the smaller particles (St = 0.9) are mainly located in the counter-rotating vortex pair. Due to the Kelvin–Helmholtz instability, several characteristic phenomena have been observed, including particle trailing and the wave structure constructed of the particle cloud, and the particles also make the Mach disk lower.

This paper presents numerical and experimental investigations of the two-phase underwater ramjet. Unlike giving a fixed initial bubble radius in previous studies, the more advanced numerical model, which can calculate the initial bubble radius on the basis of the flowfield parameters, was developed by considering the process of bubble formation. A scaled-down experimental ramjet engine was tested in a direct-connect ground testing system to characterize the factors influencing the engine performance. The results show that, with the gas/water ratio rising up, the dimensionless thrust was almost linearly increased, and a bubble injector with a smaller orifice diameter was only able to promote the engine performance slightly. The predictions with the present model compare very favorably with experimental results, and the net thrust of the two-phase underwater ramjet can be predicted with an accuracy of less than $\pm 20\%$.

As a promising research direction in recent decades, active learning allows an oracle to assign labels to typical examples for performance improvement in learning systems. Existing works mainly focus on designing criteria for screening examples of high value to be labeled in a handcrafted manner. Instead of manually developing strategies of querying the user to access labels for the desired examples, we utilized the reinforcement learning algorithm parameterized with the neural network to automatically explore query strategies in active learning when addressing stream-based one-shot classification problems. With the involvement of cross-entropy in the loss function of Q-learning, an efficient policy to decide when and where to predict or query an instance is learned through the developed framework. Compared with a former influential work, the advantages of our method are demonstrated experimentally with two image classification tasks, and it exhibited better performance, quick convergence, relatively good stability and fewer requests for labels.