Improving efficiency in smart grid monitoring using hybrid classification and dimensionality reduction

With the rapid expansion of smart grids and renewable energy integration, efficient monitoring and data communication have become critical tasks. Traditional techniques lack adaptability, generating inefficiencies in path selection, increased communication overhead and high latency. This research aims to develop a Smart Grid Monitoring System (SGMS) that improves collection of data, processing and transmission efficiency in Internet of Things (IoT) based smart grids. The proposed SGMS employs sensors to collect data from photovoltaic (PV) systems, wind systems, the grid and battery systems. The gathered data is processed and stored using an ESP8266 NodeMCU microcontroller, facilitating monitoring and analysis. The monitored data undergoes preprocessing steps, including one-hot encoding and dimensionality reduction through Linear Discriminant Analysis (LDA). One-hot encoding ensures categorical data compatibility for machine learning algorithms, while LDA optimally reduces the data dimensionality preserving class discriminability. To improve the efficiency of data transmission and minimize latency, a Hybrid Updated Gazelle-Random Forest (HUG-RF) classifier is proposed for identifying the shortest path from the NodeMCU to an IoT webpage for data visualization. The processed data is thus transmitted via the identified shortest path, enhancing the system's responsiveness and scalability. The proposed SGMS offers comprehensive monitoring of critical parameters like voltage, current and state-of-charge in the smart grid ecosystem, efficient data processing and transmission along with reduced computational complexity. The simulation outcomes establish the effectiveness of the developed approach in achieving 96.8% accuracy and 0.36% of energy consumption. The study thus presents a robust and intelligent monitoring system with IoT platform for remote accessibility and control.