USING CONVOLUTIONAL NEURAL NETWORKS TO PREDICT ELECTRICITY CONSUMPTION
DOI:
https://doi.org/10.31891/2307-5732-2024-337-3-25Keywords:
convolutional neural network, recurrent neural network, electricity consumption, machine learningAbstract
Predicting energy consumption is critical for efficient energy management and cost reduction. While recurrent neural networks (RNNs) have traditionally been used for time series processing due to their ability to model time dependencies, convolutional neural networks (CNN) can offer significant performance advantages due to parallel data processing and the ability to detect local patterns in sequences. In this paper, we analyze the performance of these two types of neural networks, evaluating their efficiency and accuracy in the context of electricity consumption forecasting using ANNs, temporal convolutional networks (TCN), and long-term short-term memory (LTSM). As a result, it was found that convolutional neural networks are 30% more accurate and 25% faster.
