Deep-Learning Algorithmic-Based Improved Maximum Power Point-Tracking Algorithms Using Irradiance Forecast

Abstract

Renewable energy is a key technology for achieving carbon-free energy transitions, and solar power systems are one of the most reliable resources for achieving this. Solar power systems have a simple structure and are inexpensive. However, depending on the input irradiance, the existing maximum output control algorithm (P&O) has disadvantages due to its slow transient response and steady-state vibration. Therefore, in this paper, we propose a maximum output control algorithm based on a deep learning algorithm that can predict the input irradiance. This can achieve a quick transient response and steady-state stability. The proposed method predicts the irradiance based on the output voltage/current and power of the photovoltaic (PV) system and calculates the duty ratio that can accurately follow the maximum output point according to the irradiance. The deep learning model applied in this study was trained based on the experimental results using a 100 W PV panel, and the performance of the proposed algorithm was verified by comparing its performance with that of the conventional algorithm under various input irradiance conditions. The proposed algorithm exhibits a maximum efficiency increase of 11.24% under the same input conditions as those of the existing algorithms.