Operational reliability enhancement in wave energy converters: Artificial intelligence-based rotational speed control for vibration mitigation in oscillating water columns
针对波浪能转换器结构稳定性问题，提出基于人工神经网络的转速控制方法，通过调整涡轮发电机角频率避开临界转速，减少轴承、共振和不平衡引起的振动，同时提升发电功率。
Wave Energy Converters (WECs) play a crucial role in the transition toward sustainable energy sources. However, their efficiency and reliability are often hindered by challenges in structural stability, which impacts energy extraction performance. While various control strategies have been proposed for WEC in general, existing methods often fail to adequately address real-time adaptability and robustness under fluctuating wave conditions. Moreover, there’s almost no control strategies using real measured vibration for data-driven rotational speed control or control strategies using PTO’s rotational speed as an active structural control specific in WECs in general or in OWCs. Optimization techniques and AI can be applied to develop more reliable and sustainable energy generation and storage systems, contributing to the goal of carbon neutrality. This study addresses this gap by developing an Artificial Neural Network-based rotational speed control approach tailored for WECs. The proposed methodology introduces the modeling of a Wells turbine-based OWC and integrates a novel rotational-speed-control using an Artificial Neural Network (ANN) for vibration reduction using premeasured data of bearing, unbalance and resonance-induced vibrations. The proposed ANN-based rotational-speed-control will adjust the turbo-generator angular frequency to evade the critical speed and increase the generated power. Through extensive simulations and validation, a comparative study has been carried out between the controlled and uncontrolled OWC. The obtained results demonstrate the effectiveness and the superior performance of the suggested ANN-based rotational speed control against the uncontrolled case by reducing structural vibrations by 77.53 %, 52.91% and 36.10% and maximizing power generation up to 65.75%, 61.55 % and 26.85% for the bearing, resonance and unbalance problems, respectively. These findings highlight the potential of machine learning techniques in advancing WEC technology, ultimately contributing to the reduction of the Levelized Cost of Energy (LCoE) in wave energy applications.