Research on a Self-Powering Solution for Cable Online Monitoring Devices Based on Reinforcement Learning Collaborative Evolutionary Algorithm

To address the energy supply challenges of online monitoring devices for power cables, a self-powering solution based on magnetic induction energy harvesting device is proposed. Electromagnetic field simulations are analyzed on the areas of straight and curved cable installation to determine the optimal installation position of the magnetic induction energy harvesting device, identifying regions with higher magnetic flux density. To enhance the output voltage of the magnetic induction energy harvesting device, an Actor-Critic reinforcement learning-based evolutionary algorithm is introduced to optimize the magnetic core structure. This approach overcomes the issues of selection pressure imbalance and the reduction in population diversity typically encountered in traditional evolutionary algorithms. Experimental results demonstrate that the proposed self-powering solution effectively identifies the optimal installation position and coil structure. Compared with the genetic algorithm and Advantage Actor-Critic algorithms, the coil structure optimized using the Actor-Critic reinforcement learning-based evolutionary algorithm achieves a 28.9% and 12.2% increase in output-induced voltage, respectively.