The Implementation of Building Intelligent Smart Energy using LSTM Neural Network

Abstract

Internet of Things (IoT) makes many devices getting smarter and more connected in the 4.0 industrial revolution. One of the implementations of the Internet of Things is smart energy. It allows communication between humans or between things that make a building smarter. This paper proposes the implementation of the MQTT-based smart meter. The smart meter is used to make it easier for users to monitor and manage the energy consumption of buildings in real-time. It is considered as the main component of a smart network to make efficient and manage energy consumption remotely. Taking into account the increasing demand for electricity in Indonesia, smart meters can reduce overall energy use and reduce global warming by optimizing energy utilization through the internet of things and artificial intelligence. This paper proposes the implementation of the MQTT-based smart meter. This smart meter can measure energy consumption, transmit information related to the energy used, and provide an early warning system to stakeholders through the website in real-time analytics with predictive data on the following month and what days are most used to support energy consumption efficiency planning. This study conducted LTSM and ARIMA to determine forecasting energy consumption with 59 epochs, 8 batch sizes, 64 hidden layers with the results of MSE Error, RMSE Error, Mean Accuracy 0.14,0.373, and 95.16%, respectively. This result is better than ARIMA with MSE error results of 0.812 and 0.66 and RMSE error.