A novel method is proposed for accurately determining the local wall friction through the near-wall measurement of time-average velocity profile in a Type-A turbulent boundary layer (TBL). The method is based on the newly established analytical wall-law in Type-A TBL. The direct numerical simulations (DNS) data of turbulence on a zero-pressure-gradient flat-plate (ZPGFP) is used to demonstrate the accuracy and the robustness of the approach. To verify the reliability and applicability of the method, a two-dimensional particle image velocimetry (PIV) measurement was performed in a ZPGFP TBL with a low-to-moderate Reynolds number (Re). Via utilizing the algorithm of single-pixel ensemble correlation (SPEC), the velocity profiles in the ZPGFP TBL were resolved at a significantly improved spatial resolution, which greatly enhanced the measurement accuracy and permitted us to accurately capture the near-wall velocity information. The accuracy of the approach is then quantitatively validated for the high Reynolds number turbulence using the ZPGFP TBL data. The research demonstrates that the current method can provide the precise estimation of wall friction with a mean error of less than 2%, which not only possesses the advantage of its insensitivity to the absolute wall-normal distance of the measuring point, but also its capability of providing an accurate prediction of wall shear stress based on fairly sparse experimental data on the velocity profile. The current study demonstrates that the wall shear stress can be accurately estimated by a velocity even at a single-point either measured or calculated in the near-wall region.

In this paper, we construct a pretraining framework for fluid environment perception, which includes an information compression model and the corresponding pretraining method. We test this framework in a two-cylinder problem through numerical simulation. The results show that after unsupervised pretraining with this framework, the intelligent agent can acquire key features of surrounding fluid environment, thereby adapting more quickly and effectively to subsequent multi-scenario tasks. In our research, these tasks include perceiving the position of the upstream obstacle and actively avoiding shedding vortices in the flow field to achieve drag reduction. Better performance of the pretrained agent is discussed in the sensitivity analysis.

The aspect ratio dependence of the electroconvection phenomenon in a suspended nematic liquid crystal film with a rectangular boundary is investigated. Two-dimensional global stability analysis is carried out on the coupled electrohydrodynamic system to calculate the instability boundary of the phenomenon for different aspect ratios. The calculated critical $R$ number (Rayleigh-like number) shows a rapidly decreasing trend in the low-aspect-ratio region (roughly $\unicode[STIX]{x1D6FE}<1.5$ , where $\unicode[STIX]{x1D6FE}$ is defined as the aspect ratio of the film), and then the variation becomes slow until $\unicode[STIX]{x1D6FE}\approx 2.5$ , where the critical $R$ number starts to increase slightly. Convective patterns of liquid films with different aspect ratios are also obtained from stability analysis and validated by particle image velocimetry measurement.

Achieving robust and energy-efficient navigation in unknown fluid environments remains a key challenge for bioinspired underwater robots. In this study, we develop a reinforcement learning-based control framework that enables a fish-like swimmer to autonomously acquire effective navigation strategies within a high-fidelity computational fluid dynamics environment. By shaping the reward function to favor energy efficiency, the agent spontaneously discovers different locomotion patterns, ranging from continuous bursting to burst-and-coast gaits, all without prior knowledge of fluid mechanics. Although the agent is trained in a quiescent fluid environment, the learned swimming policies are generalized well in various navigation tasks and remain robust under complex flow perturbations, including uniform currents and unsteady vortex wakes. In all test scenarios, the agent achieves a 100 % navigation success rate. These findings highlight the potential of integrating physics-based simulation with learning-based control strategy to advance the design of adaptive, efficient, and resilient aquatic robots inspired by biological swimmers.

Surface pressure measurements contain information about wake-induced flow disturbances and the motion of disturbance sources, yet this information is embedded in complex temporal pressure signals that are difficult to interpret. Extracting informative flow features from such measurements therefore remains challenging. Many existing approaches rely on task-specific mappings between pressure signals and target states, which limits their ability to learn transferable representations of hydrodynamic signals. In this study, we propose a self-supervised learning framework to learn compact temporal representations from sparse surface pressure measurements. Water-tunnel experiments are conducted using a National Advisory Committee for Aeronautics 0015 sensing model equipped with fourteen pressure sensors that record wake-induced pressure fluctuations generated by upstream bodies. A gated recurrent unit encoder is trained through a self-supervised objective to extract temporal features from the pressure sequences, and the learned representations are evaluated through downstream tasks that infer the motion parameters of disturbance sources. The model is first trained using pressure responses induced by the wake of a pitching airfoil and subsequently evaluated under a second tested disturbance scenario involving a translating D-shaped cylinder. Experimental results show that the learned representation improves reconstruction accuracy and accelerates training convergence compared with randomly initialized models, while retaining useful predictive performance in the tested transfer case. Furthermore, spatial channel ablation provides pressure-based evidence for the relative importance of different sensor regions. The ablation results indicate that the most informative sensor region changes between the tested wake-generation mechanisms. These results suggest that self-supervised representation learning can capture temporal pressure-response features that are transferable within the tested configurations, providing a practical framework for pressure-based inference of upstream hydrodynamic states from limited sensor measurements.