Energy consumption prediction has become an integral part of a smart and sustainable environment. With future demand forecasts, energy production and distribution can be optimized to meet the needs of the growing population. However, forecasting the demand of individual households is a challenging task due to the diversity of energy consumption patterns. Recently, it has become popular with artificial intelligence-based smart energy-saving designs, smart grid planning and social Internet of Things (IoT) based smart homes. Despite existing approaches for energy demand forecast, predominantly, such systems are based on one-step forecasting and have a short forecasting period. For resolving this issue and obtain high prediction accuracy, this study follows the prediction of household appliances' power in two phases. In the first phase, a long short-term memory (LSTM) based model is used to predict total generative active power for the coming 500 hours. The second phase employs a hybrid deep learning model that combines convolutional characteristics of neural network with LSTM for household electrical energy consumption forecasting of the week ahead utilizing Social IoT-based smart meter readings. Experimental results reveal that the proposed convolutional LSTM (ConvLSTM) architecture outperforms other models with the lowest root mean square error value of 367 kilowatts for weekly household power consumption.(c) 2022 Elsevier Inc. All rights reserved.

Predicting Household Electric Power Consumption Using Multi-step Time Series with Convolutional LSTM

Cascone, L;
2023-01-01

Abstract

Energy consumption prediction has become an integral part of a smart and sustainable environment. With future demand forecasts, energy production and distribution can be optimized to meet the needs of the growing population. However, forecasting the demand of individual households is a challenging task due to the diversity of energy consumption patterns. Recently, it has become popular with artificial intelligence-based smart energy-saving designs, smart grid planning and social Internet of Things (IoT) based smart homes. Despite existing approaches for energy demand forecast, predominantly, such systems are based on one-step forecasting and have a short forecasting period. For resolving this issue and obtain high prediction accuracy, this study follows the prediction of household appliances' power in two phases. In the first phase, a long short-term memory (LSTM) based model is used to predict total generative active power for the coming 500 hours. The second phase employs a hybrid deep learning model that combines convolutional characteristics of neural network with LSTM for household electrical energy consumption forecasting of the week ahead utilizing Social IoT-based smart meter readings. Experimental results reveal that the proposed convolutional LSTM (ConvLSTM) architecture outperforms other models with the lowest root mean square error value of 367 kilowatts for weekly household power consumption.(c) 2022 Elsevier Inc. All rights reserved.
2023
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11386/4846377
 Attenzione

Attenzione! I dati visualizzati non sono stati sottoposti a validazione da parte dell'ateneo

Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 32
  • ???jsp.display-item.citation.isi??? 22
social impact