Deep Learning Based Renewable Energy Forecasting in Microgrids

• Published in: 2023 4th International Conference on Smart Electronics and Communication (ICOSEC) pp. 774 - 780
• Main Author: Choudhary, Sunila
• Format: Conference Proceeding
• Language: English
• Published: IEEE 20.09.2023
• Subjects: deep learning (DL); hybrid threshold-adaptive Harris Hawks optimized convolutional long short-term memory (HTHHO-DCLSTM); Microgrid; Microgrids; Prediction algorithms; Production; Reliability; Renewable energy sources; renewable energy sources (RESs); Smart grids; Technological innovation
• DOI: 10.1109/ICOSEC58147.2023.10275943

Microgrids (MG) have become an essential part of smart grids that also contain advanced technologies for storing energy, load control strategies, and dispersed renewable energy sources (RESs). For an intelligent dc microgrid, the intermittent nature of RESs presents a variety of difficulties, including reliability, power source, and to provide balance. So, predicting RES energy production is increasingly crucial for achieving the most significant possible usage RESs and the practical and continuous operation of the power grid. Energy demand projections are included in intelligent microgrids to aid in planning power output and power exchanges with the utility grid. DL-based algorithms are promising for predicting customer requirements and energy output from RESs. This article presents a unique hybrid threshold-adaptive Harris Hawks optimized deep convolutional long short-term memory (HTHHO- DCLSTM) to forecast the renewable energy in microgrids. First, we use the min-max normalization method to pre-process the gathered raw RES data. In this instance, the HTHHO approach identifies the essential features from the processed data set. After learning these properties, the suggested DCLSTM approach is used to generate an overall prediction of energy production. Several metrics are used to evaluate the efficacy of the recommended strategy, and a comparison between the proposed and existing approaches is also given. Our suggested method outperforms the current practices in forecasting the RE in microgrids, according to the findings of our study.