Long-Term Equilibrium Relationship of the Johansen Model Applied to an Island Zone for Photovoltaic Energy Prediction

Abstract

Solar photovoltaic (PV) plant systems transform solar energy into electric power. A main goal of the Mauritian government is to significantly increase from 13% to 60% of the share of renewables in the energy mix. This paper presents findings based on the Johansen vector error correction model (VECM) cointegration approach, drawing from the author's original and prior research, with a focus on short-term predictions (15-minute intervals) for PV power output in Mauritius. The innovation of this study lies in applying the long-term equilibrium relationship identified in earlier research from Reunion Island to the PV plants in Mauritius. The prediction model has been trained using data from 2019 to 2022, focusing on a random month and day, for both hourly and 15-minute intervals. The data used in this study are measured from the PV system installed on the rooftop of the University of Mauritius (UoM) which is located on the west coast of the island. Experimental results indicate that the performance metric R² values exceed 93%, demonstrating a strong positive correlation between the Johansen model and onsite measurements. This proposed model serves as a robust predictive tool, with the potential for enhanced accuracy when integrated with machine learning techniques such as LSTM that can adapt to varying datasets. Future works will focus on the best artificial intelligence to fit the model as sudden variability in PV power output due to sudden fluctuation in irradiance can have a serious impact on grid stability.