Auto-optimization of Energy Generation for Wave Energy Converters with Active Learning
Siyang Tang, Wen-Hua Chen, Cunjia Liu · 2025
Read Auto-optimization of Energy Generation for Wave Energy Converters with Active Learning by Siyang Tang, Wen-Hua Chen, Cunjia Liu (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2507-01727.txt.
Introduction
Siyang Tang, Wen-Hua Chen, and Cunjia Liu present a 2025 control system for wave energy converters. A wave converter is a device that turns ocean wave motion into power. Its output swings as sea states change, and the best settings shift with the waves. The authors aim to squeeze the most energy from seas that are not known in advance.
The system works on two levels. A slow top level plans the force profile the device should follow over several waves. A fast lower level makes the hardware track that plan. The top level learns the unknown wave pattern by probing the sea while it works, and it counts its own doubt in each choice. In computer tests with plain and rough waves, the system beat three rival control methods.
Catalog record
| CREATOR | Siyang Tang, Wen-Hua Chen, Cunjia Liu |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 48 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2507-01727
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Auto-optimization of Energy Generation for Wave Energy Converters with Active Learning Siyang Tang, Wen-Hua Chen, Cunjia Liu Abstract This paper presents an auto-optimization control framework for wave energy converters (WECs) to maximize energy generation under unknown and changing ocean conditions. The proposed control framework consists of two levels. The high-level controller operating at a longer time scale aims to maximize the average energy generation over several wave periods. The generated Power Take-Off (PTO) profile as the reference for the low-level physical system to follow. The new auto-optimization process leverages the parameterization of the non-stationary operation condition in WECs, establishing the relationship between the average energy generation and the key design parameters of the PTO force subject to the unknown wave parameters. The high-level controller is designed based on the concept of Dual Control for Exploration and Exploitation (DCEE) to quickly learn the unknown wave parameters by actively probing the ocean condition, while generating the optimal PTO profile. During this process, the uncertainty of the estimated wave condition is quantified and embedded in the optimization cost function to enable active learning. Simulation results under unknown regular and irregular waves demonstrate the effectiveness and robustness of this novel auto-optimization WEC systems with active learning, outperforming model predictive control, extremum seeking and classic Bang-Bang control approaches. 1 Introduction Renewable alternatives to fossil fuels are gaining momentum in response to the growing urgency of climate change. With approximately 71% of the Earth’s surface covered by oceans, these bodies of water offer an immense and largely untapped source of clean energy. Among various marine energy sources, wave energy stands out due to its superior power density compared to solar and wind energy [ 1 ] . Wave Energy Converters (WECs), which transform ocean wave energy into electricity, play a central role in harnessing this resource. Automatic control is pivotal for optimizing energy extraction in WECs by manipulating the Power Take-Off (PTO) mechanism. Numerous control strategies have been developed to maximize energy capture [ 2 , 3 , 4 , 5 , 6 ] . Among them, complex conjugate control determines the optimal velocity trajectory based on the frequency characteristics of the hydrodynamic excitation force. Specific implementations include derivative latching [ 2 ] and de-latching control [ 3 ] . Reactive control further extends passive strategies by adjusting both PTO resistance and reactance [ 5 ] . More recently, data-driven approaches such as reinforcement learning [ 4 ] and neural networks [ 7 ] have been explored to adapt to unknown and time-varying sea conditions. Model Predictive Control (MPC) and its variants [ 8 , 9 ] are recognized for their ability to handle system constraints while achieving promising control performance. Substantial efforts have been made to tailor MPC methods for WECs, as surveyed in [ 10 ] . With the development of wave forecasting technologies [ 11 ] , predictive control methods can incorporate future wave information to enhance energy harvesting [ 12 ] . However, the high computational burden of MPC often limits its applicability in real-time scenarios [ 10 ] . Additionally, the control performance index for WECs is frequently non-convex, posing further challenges to optimization [ 13 ] . Wave energy generation is fundamentally different from solar and wind power systems. Whereas the latter typically operate around a static maximum power point [ 14 ] , WECs must synchronize their motion with the oscillatory nature of ocean waves. As such, their optimal operating condition is inherently non-stationary and periodic, and cannot be described by a fixed point in the control space [ 15 ] . This unique challenge motivates the need for autonomous control systems capable of learning and adapting to changing ocean conditions. In parallel to these physically motivated strategies, learning-based approaches have gained increasing attention. A notable example is the recent work on deep reinforcement learning (DRL)-based non-causal control for wave energy conversion [ 16 ] , which proposes a data-driven scheme that combines wave prediction with adaptive control gain tuning. Such approaches showcase the potential of learning-based intelligence in WEC control. This paper investigates the problem of maximizing wave energy generation from the perspective of autonomous learning and adaptation. We consider the WEC as an intelligent agent operating in a partially unknown and dynamically changing environment, with the goal of maximizing its energy output by actively learning the characteristics of the wave field. Ideally, such a control system should not only adapt to the environment but also explore it efficiently to accelerate the learning process. This requires the WEC to interact with ocean waves in a way that both improves energy capture and yields informative data about the wave conditions. To achieve this, we adopt the Dual Control for Exploitation and Exploration (DCEE) framework [ 17 , 18 ] , a recently developed approach in autonomous control. Originally introduced for source-seeking problems in atmospheric dispersion [ 19 ] , DCEE enables an agent to simultaneously exploit current knowledge to maximize performance while exploring the environment to reduce uncertainty through active learning. The framework has demonstrated promising results in other energy optimization tasks under environmental uncertainty, such as maximizing photovoltaic (PV) power generation [ 20 ] , and has been shown to outperform conventional reinforcement learning in some scenarios [ 21 ] . Inspired by these successes, this paper applies DCEE to the WEC system to tackle the challenge of maximizing energy generation under uncertain and time-varying sea conditions. However, a key distinction lies in the non-stationary nature of the optimal operating condition in WECs—something not encountered in prior DCEE applications. To address this challenge, we propose to parameterize the periodic behavior of the optimal PTO profile using trigonometric functions, consistent with the wave-induced motion of the buoy. While previous works have used truncated Fourier series or orthogonal polynomial projections to approximate system states and controls [ 22 , 23 ] , we adopt a more concise parametrization based on key trigonometric parameters. Through analytical derivation, we establish a functional relationship between the average generated energy and these key parameters, thereby transforming the problem of control in an unknown environment into one of learning the wave characteristics from observed energy output. Unlike prior approaches that estimate wave parameters such as frequency using spectral or short-horizon estimation methods [ 24 , 25 ] , our framework employs Bayesian inference to estimate these parameters while simultaneously quantifying the associated uncertainty in real time. This uncertainty estimation is crucial to the DCEE strategy, where control actions are optimized not only to track the estimated optimal behavior (exploitation), but also to probe the wave conditions more effectively (exploration). By leveraging the DCEE framework and the proposed parametrization, we develop a novel auto-optimization control system capable of synchronizing WEC dynamics with unknown and varying wave fields to maximize energy capture. The contributions of this paper are threefold. 1. We introduce a novel two-level autonomous control system architecture for WECs, specifically designed to address the challenges posed by the dynamic and immeasurable ocean environment. In this architecture, the high-level aims to actively learn the ocean condition and then generate the PTO profile accordingly. It consists of a particle filter for key parameter estimation and an intelligent search controller designed using DCEE, operating at a large time