Reinforcement learning vibration control of a multi-flexible beam coupling system

/pdf/a-review-of-machine-learning-methods-applied-to-structural-1cu7fjj8.pdf

A Review of Machine Learning Methods Applied to Structural Dynamics and Vibroacoustic

Trajectory planning and vibration control of translation flexible hinged plate based on optimization and reinforcement learning algorithm

High-dimensional high-frequency continuous-vibration control problems often have very complex dynamic behaviors. It is difficult for the conventional control methods to obtain appropriate control laws from such complex systems to suppress the vibration. This paper proposes a new vibration controller by using reinforcement learning (RL) and a finite-impulse-response (FIR) filter. First, a simulator with enough physical fidelity was built for the vibration system. Then, the deep deterministic policy gradient (DDPG) algorithm interacted with the simulator to find a near-optimal control policy to meet the specified goals. Finally, the control policy, represented as a neural network, was run directly on a controller in real-world experiments with high-dimensional and high-frequency dynamics. The simulation results show that the maximum peak values of the power-spectrum-density (PSD) curves at specific frequencies can be reduced by over 63%. The experimental results show that the peak values of the PSD curves at specific frequencies were reduced by more than 47% (maximum over 52%). The numerical and experimental results indicate that the proposed controller can significantly attenuate various vibrations within the range from 50 Hz to 60 Hz.

/pdf/a-new-vibration-controller-design-method-using-reinforcement-3ob4uwst.pdf

A New Vibration Controller Design Method Using Reinforcement Learning and FIR Filters: A Numerical and Experimental Study

A hybrid method incorporating the simulations of noise sources with delayed detached eddy simulation (DDES) and calculations of far-field noise with the Ffowcs Williams–Hawkings (FW-H) equation is used to study the suppression technique for the aerodynamic noise of a Faiveley CX-PG pantograph. Considering that China’s Fuxing bullet trains operate at 350 km/h, the inflow velocity of 350 km/h is applied in this paper. The noise radiated from the panhead area, middle area, and bottom area at an inflow velocity of 350 km/h is distinguished. The noise intensities at the standard observer show that the noise radiated from the panhead area is the strongest, and the sound pressure level spectrum value is larger than the other two in the range above 500 Hz. The influence of applying the wavy rods and modifying the contact strip shape on the aerodynamic noise is discussed in detail. By comparing the acoustic source distribution and the far-field noise intensity, it is found that applying the wavy rods can effectively reduce the panhead noise, especially around the peak frequency. Modifying the shape of the contact strip to a hexagon can suppress the vortex shedding, leading to a lower surface pressure level. Combining the strip modification and wavy rods, the total noise intensity can be diminished by about 3.0 dB.

/pdf/numerical-study-on-aerodynamic-noise-reduction-of-pantograph-34ggr20f.pdf

Numerical Study on Aerodynamic Noise Reduction of Pantograph

Generation of noise caused by the flow around a cylinder and its control are important in various engineering applications. Based on computational fluid dynamics with acoustic analogy and the vortex dynamics theory analysis, this study aims at investigating the ability of the wavy cylinder in improving aerodynamic performance and reducing aerodynamic noise. Noise control mechanisms with different Reynolds numbers are analyzed. The results show that the wavy cylinder is helpful for the reduction of the average drag coefficient and is efficient in suppressing fluctuation of the lift coefficient; consequently, the overall noise of the wavy cylinder is reduced. Specifically, the tonal noise is significantly suppressed or even eliminated under proper configurations. To explore the underlying noise suppression mechanisms, the process of vorticity generation around the wavy cylinder surface is examined in detail. The vorticity distribution on the surface of the wavy cylinder is profoundly improved, and the distribution of the boundary vorticity flux and boundary enstrophy flux is also remarkably weakened. As a result, the generation of vorticity near the wavy cylinder wall is diminished. These directly lead to a significant contraction of the vortex structure distribution in the wavy cylinder wake, especially for some large-scale vortex structures. Moreover, periodic vortex shedding is significantly suppressed in the case with high Reynolds numbers, which might be the main reasons for noise reduction. The interaction area of the positive and negative Lamb vector divergence, which is closely related to the noise generation, is decreased. This contributes to drag reduction and noise attenuation. This indicates that drag reduction and noise suppression are closely bounded in the wavy cylinder.

Guo-hao Chen

Papers

Numerical study on the flow and noise control mechanism of wavy cylinder

Reinforcement learning vibration control for a flexible hinged plate

Effects of slat track on the flow and acoustic field of high-lift devices