Increasing Accuracy of Power Consumption Using Artificial Neural Network

This paper proposed a smart lighting system using Artificial Neural Network (ANN) algorithm. Power saving is one of the concerns of researchers to continue to be improved. The addition of predictions will increase the ability to save power on the use of smart home devices in the future. Through prediction, the system can decide when the lights are used and when the lights are not used. Therefore, an IoT-based smart light system that can predict the state of the lamp based on sensor data is needed. In this paper ANN algorithm is used to predict the state of the lamp. The purpose of this paper is to analyze the performance of the ANN by entering data from the light system based on the presence of lecturers using magnetic door and infrared sensors. The accuracy of the ANN application is influenced by the lamp state pattern. The result shows that an accuracy of 58.17% for training data and 52.54% for test data in predicting the state of the lamp. The significant power saving is calculated using Wilcoxon Method. It shows that this system provides significance for power saving of 31.75%.