Year: 2025
Type: article
Abstract: Industrial settings increasingly employ real-time monitoring systems to ensure operational safety and efficiency. Traditional centralized fault detection schemes are of high latency, energy loss, and limited scalability in WSNs. Addressing these challenges, this paper proposes a novel hybrid framework, EFD-IoT (Edge-Cloud Fault Detection Model), for efficient, accurate, and scalable fault monitoring in industrial IoT settings. The architecture enables the combination of light decision tree models in edge nodes and deep CNN-LSTM models in the cloud for rapid local decision-making and intensive centralized analysis. The Industrial IoT Fault Monitoring Dataset (IIFMD) was developed using real-time sensory data from operating factories (temperature, vibration, current). Accuracy, precision, recall, and F1-score for the proposed model were 98.1%, 97.3%, 96.9%, and 97.1%, respectively. It also exhibited a 40.6% reduction in latency, 30.0% less in false alarms, and more than 40% energy efficiency improvement. The system also accomplished a 94.7% model update success ratio and exhibited stable multi-sensor fusion properties. All these outcomes confirm the possibility of EFD-IoT for industrial WSN applications in predictive maintenance and real-time fault diagnosis. The paper concludes with its potential for deployment at large scales while considering energy in industrial settings.
Authors: Prabhakara Rao T, Vikas B, Venkatesh Sharma K
Language: English
Subfield: Control and Systems Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W4413904112