CNN-Bi-LSTM Based Household Energy Consumption Prediction

Driven by technological advances, there is a increase in electricity-based equipments and this leads to excessive energy consumption (EC) and demand for power every day. To enhance power management and collaboration between electricity used in a building and the smart grid, the EC must be predicted. Forecasting techniques used for prediction of the energy accurately are limited due to challenges like dynamic behaviour of residents and climatic condition. So, to conquer such challenges we proposed a deep learning based methodology. The proposed methodology uses hybrid model consisting of CNN and Bi-LSTM for predicting EC. The performance of the proposed methodology is tested using publically available real dataset. Test results shows that the proposed methodology are able to predict the consumption with very small error. The proposed methodology helps in management for producing optimum amount of power.