Towards Smart Power Grids: Integrating Internet of Things and Machine Learning for Predictive Maintenance and Reliability in Power Transmission Lines.
Abstract
Electrical energy is essential to modern life, powering everything from homes to industrial operations. Ensuring reliable and efficient power transmission is critical to maintaining uninterrupted and safe energy delivery. However, transmission lines face numerous challenges, including environmental hazards, increasing demand, and complex maintenance requirements. To address these issues, this paper presents a smart, edge-enabled system that integrates Internet of Things (IoT) sensors with Machine Learning (ML) for predictive maintenance and reliability monitoring of electrical power transmission lines. The system utilizes a diverse array of sensors, including GPS, current, electromagnetic radiation, vibration, and temperature sensors to collect real-time data from transmission infrastructure. A lightweight Artificial Neural Network (ANN) model is deployed on the Arduino Nano 33 BLE microcontroller using TensorFlow Lite for Microcontrollers. This model analyzes sensor inputs directly at the edge to predict reliability scores and detect anomalies without relying on continuous cloud connectivity. The system also includes a mobile and web-based interface for data visualization and remote control of emergency actuators such as circuit breakers, fans, and alarms. Experimental results demonstrate the ANN model’s high predictive accuracy, achieving an R-squared (R²) of 0.9962, Mean Absolute Error (MAE) of 0.8640, and Root Mean Squared Error (RMSE) of 1.4016. The system exhibits low power consumption, compact hardware size, minimal data transmission, and enhanced privacy through on-device inference. These findings validate the feasibility of deploying ANN-based ML solutions for scalable, real-time reliability enhancement in smart power grids. unlike existing cloud-based system, the proposed edge-based approach has faster response-time, consumes less power, and it can work off-line in case of limited internet connectivity.