Short-Term Photovoltaic Prediction Technology Based on Combined Artificial Intelligence Technology
Hitherto, there has been a need for more energy. In recent years, their need has exponentially increased due to the influx of energy-dependent systems. Mark you, this might have caused an unexpected turn of events whose domino effects might have been trivial to the livelihood of many if not the entire human race. Fortunately, the introduction of renewable energy sources came to aid, which in turn led to the invention of renewable energy systems. The expansion of these systems, their stability, availability the quality of the smart grids, which work hand in hand with these systems have become tremendously important. A barrage of literature has been written on the same explaining their importance. These systems, commonly going by the name, photovoltaic (PV) are rapidly increasing owing to the variable energy resource (VER) increase f solar penetrative power. Notwithstanding the merits of this PV technology, their weather-dependent nature makes them vulnerable to unpredictability and variations in power output which might, in turn, increase their operation and management costs. Developing a reliable and stable method for forecasting photovoltaic power is thus imminent and should be considered. By having forecasting systems, the problems of variabilities and unpredictability can be countered which may, in turn, increase their stability. Artificial intelligence on the other hand is a very lucrative discipline, averaging the fields of neurology, engineering, and technology. It has been used in solving many problems due to its great versatility. It might come in handy in forecasting PV power. The objective of this study is to employ the use of artificial intelligence (AI) in short-term photovoltaic prediction technology. For this cause, we will be employing the feedforward neural network approach, in particular, convolutional neural networks in the prediction process. PVCNet model is employed in this situation due to its ability to generate both deterministic and probabilistic forecasting for a duration of 24hrs. Its accuracy is determined by computing both the root mean square error (RMSE) and mean square error (MSE) for which the results indicate values 164.01 and 108.95 respectively. The values are indicators that this is the best model for PV prediction.