Application of Big Data Analytics and Machine Learning in Solar Power Plant Management

To slow down the ongoing climate changes and relieve the catastrophic events caused by them, our society is taking action to limit the use of fossil fuels to generate power. However, due to the increasing demand for electricity as a result of population growth and technological improvements, it is exigent to expand and optimize the ways through which sustainable energy can be exploited. One of the promising forms of sustainable energy is solar energy, but due to its unstable nature, its deployment on a large scale, both locally and globally, has stagnated. To expand access to solar energy, make solar power plants work more efficiently, and support the grid-connected solar power systems, we need to be able to identify malfunctioning solar panels and predict electricity generation. With the innovative technical solution, using a data-driven decision engine to monitor power plants became possible. This article discussed two important use cases of advanced data analytics and machine learning in solar power plant management. By analyzing the daily electricity yielded, total electricity yielded, DC power, AC power, ambient temperature, module temperature, and irradiation of solar panels, we discovered their regular patterns, which can help with the quick identification of faulty panels. Discrepancies from these trends may suggest improper functions and need for maintenance or replacement, and deviations in different categories could imply different types of problems. Training the Time Series model and testing its robustness, we propose a reliable approach to predict electricity generation. These two results have profound applications including suggesting needs for adjustments for solar panels and managing the grids connecting solar plants and individual units in advance to distribute just sufficient electricity, thus promoting the grid system, and contributing to the transition to renewable energy and sustainable lifestyle.