Long Short-Term Memory (LSTM) Deep Neural Networks in Energy Appliances Prediction

The application of Long Short-Term Memory (LSTM) Deep Neural Networks has been increased the last years. This paper proposes a novel methodology based on a hybrid model using the Long Short-Term Memory (LSTM) Networks and the Particle Swarm Optimization (PSO) in energy appliances prediction in a low-energy house. The Particle Swarm Optimization was implemented in order to evaluate the feature importance of the energy related factors in the input vector and the LSTM Networks to perform the time series forecasting. The results have illustrated an improved accuracy compared to other machine learning techniques such as Support Vector Machines and Feedforward Neural Networks.