Machine learning-based maximum power point tracking

Abstract

Origin of the solar cell dated back to 1883 when the first solar power generation was invented. A solar system converts sunlight into electrical energy through photovoltaic panels and known as solar power generation. At present solar power generation shows highest development of among all renewable resources. Maximum Power Point Tracking is a method of optimizing the solar panel in order to obtain maximum power output from the solar panel. Comprehensive literature review of MPPT systems, tracing their historical development and examining conventional and modern MPPT methods were carried out and observed existing MPPT methods often fail to capture the relationships between voltage, current, irradiance, and temperature parameters which directly affect the optimal operating point of a photovoltaic panel without disturbing the energy harvesting process, to deliver the maximum power output of the photovoltaic system. As a result, the overall efficiency and performance of the photovoltaic system may be compromised. The proposed solution involves leveraging the capabilities of machine learning approaches to learn the complex relations between the voltage, current, irradiance and temperature paraeters using random forest regression and the proposed approach aims to overcome the limitations of traditional MPPT to obtain optimized output power from photo-voltic system. Developed MPPT model is tested using real-world data collected from PV installations under diverse environmental conditions. The tests evaluate the model's accuracy, adaptability, and performance compared to conventional MPPT techniques. Results demonstrate the effectiveness of the machine learning-based approach in improving energy harvesting efficiency and overall system performance. The findings suggest that machine learning-based VCIT-MPPT offers a promising solution for optimizing PV system performance and maximizing energy yields. In conclusion, this thesis presents a novel approach to MPPT leveraging machine learning techniques named as VCIT-MPPT, showing significant improvement in efficiency of energy harvesting and MPPT system performance. The findings contribute to the advancement of renewable energy technologies and pave the way for more efficient utilization of solar power resources.