ANN-LSTM Based Tool For Photovoltaic Power Forecasting.

Abstract

Renewable energy resources are currently being utilized in different sectors in order to meet the demands of energy, reduce environmental pollution, decrease the usage of fossil fuels, and achieve social and economic benefits for sustainable development. As the integration of renewable energy resources into electrical energy systems is expanding. Meanwhile, The development of applications for forecasting renewable energy output has also been progressing rapidly. For example, forecasting the power dispatch from the solar photovoltaic (PV) system. Therefore, an accurate forecast of the main factors of power generation became highly important in the case of large-scale power systems for intelligent match the power generation with the power demand for efficient energy management. Many applications for forecasting have been performed using artificial intelligence techniques. This paper proposed a framework for predicting PV power output using artificial neural network (ANN) and long short-term memory (LSTM) models-based tools. In this study, the meteorological and power parameters are applied as input variables. In addition, two levels of deep learning are applied to allow a deeper and more comprehensive exploration of the learning mechanism. The proposed framework is evaluated using a centralized unit for a real-time hardware-in-the-Ioop (HIL) test platform. The objective of this research, as indicated by the findings, is to enhance the practical engineering application of deep learning-based intelligent models for predicting PV power generation.