DROP: Deep Reinforcement Learning Based Optimal Perturbation for MPPT in Wind Energy
DROP: Deep Reinforcement Learning Based Optimal Perturbation for MPPT in Wind Energy
复制标题
DOI：
10.1109/naps56150.2022.10012250
复制
发表时间：
2022-10
期刊：
影响因子：
--
通讯作者：
S. S. Shuvo-S.;Md Maidul Islam;Yasin Yılmaz
中科院分区：
文献类型：
--
作者：
S. S. Shuvo-S.;Md Maidul Islam;Yasin Yılmaz
风能的波动特性激发了研究人员寻找一种快速、有效的最大功率点跟踪（MPPT）算法。MPPT方法旨在通过调整转子转速来利用不同风速下的最大功率。我们对风能MPPT任务的贡献是双重的。首先，我们使用预测模型将水轮机转速和输出功率的当前工作点映射到最优工作点（即最大输出功率的最优水轮机转速）。其次，我们提出了一种基于深度强化学习的解决方案，该解决方案提供自适应速度控制，以快速准确地达到MPP。实验结果表明，与现有技术相比，该方法性能优越。
The fluctuating nature of wind energy has inspired researchers to look for a fast, efficient Maximum Power Point Tracking (MPPT) algorithm. The MPPT method aims to harness maximum power at varying wind speeds by adjusting rotor speed. Our contribution to the wind MPPT task is twofold. First, we use a predictive model to map the current operating point of the turbine speed and output power to the optimal operating point (i.e., optimal turbine speed for maximum output power). Second, we propose a Deep Reinforcement Learning based solution that provides adaptive speed control to reach the MPP fast and precisely. Our experimental results demonstrate the superior performance of our method compared with the existing techniques.