Design of a Smart Distribution Panelboard Using IoT Connectivity and Machine Learning Techniques

Abstract

Electric load management through continuous monitoring and intelligent controlling has become a pressing requirement, particularly in light of rising electrical energy costs. The main purpose of this work is to realize a low-voltage electrical distribution panelboard that allows for real-time load monitoring and that provides a load forecasting feature at the household level. In this regard, we demonstrate the design and the implementation details of an IoT-enabled panelboard with smart features. An IoT dashboard was used to display the most significant information in terms of voltage, current, real power, reactive power, apparent power, power factor, and energy consumption. Additionally, the panel system offers visualization capabilities that were integrated into a cloud-based machine learning modeling. Among several algorithms used, the Gaussian SVM regression exhibited the best training and validation results for the load forecasting feature. It is possible for the proposed design to be simply developed to add more smart features such as fault detection and identification. This assists in an efficient management of energy demand at the consumer level.