Sudden variability in solar photovoltaic (PV) power output to electrical grid can not only cause grid instability but can also affect power and frequency quality. Therefore, to study the balance of electrical grid or micro-grid power generated by PV systems in an upstream direction, predicting models can help. The power output conversion is directly proportional to the solar irradiance. Unlike time horizons predictions, many technics of irradiance forecasting have been proposed, long, medium and short term forecasting. For the Island of Mauritius in the Indian Ocean, and regards to key policy decisions, the government has outlined its intention to promote the PV technologies through the local electricity supplier but oversee the technical requirements of PV power output predicts for 1 hour to 15-minutes ahead. So, this paper is illustrating results of the Johansen vector error correction model (VECM) cointegration approach, from the author original and previous studies, but for a very short term prediction of 15-minutes to PV power output in Mauritius. The novelty of this study, is the long run equilibrium relationship of the Johansen model, that was initially determined in previous research works and from dataset in Reunion Island, is then applied to the PV plant in the Island of Mauritius. The proposed prediction model is trained for an hourly and 15-minutes period from year 2019 to year 2022 for a random month and a random day. The experimental results show that the performance metric R² values are more than 93% signifying that Johansen model is positively and strongly correlated to onsite measurements. This proposed model is a powerful predicting tool and more accuracy should be attained when associated to a machine learning method that can learn from datasets.