Short-Term Energy Forecasting Using an Ensemble Deep Learning Approach
P. Yogendra Prasad, Moola Ramu, Annavarapu Yasaswini, Mangala Gowthami, Putta Sai Harika, Chettipalli Abhishek
Abstract
Precise estimation of domestic electricity usage is essential for sustainable energy management, enabling effective energy allocation, and advancing the development of intelligent networks.To anticipate home electric power consumption from time series data, this research investigates usage of advanced machine learning models, including Recurrent neural networks.The dataset includes observations of a household's electricity use over a long period, together with details like usage patterns and time of day.To address anomalies, standardize the series, and organize the data for sequential learning, we preprocess it.The study assesses the performance of each model, finding that GRUs are better at spotting spatial-temporal patterns in the data, RNNs are better at sequential data prediction, and LSTMs are better at capturing long-term dependencies.To increase prediction accuracy, the comparison study lays the groundwork for future efforts to optimize model architectures and incorporate outside variables like weather and economic data.This study highlights how deep learning has the ability to change energy management procedures and open the door to more economical and environmentally friendly home energy use.