Secure AI-Enabled Cyber-Physical Framework for Integrated Fluid, Mechanical, and Electrical Systems

Abstract

The growing use of fluid, mechanical, and electrical elements in modern industrial systems creates a complex cyber-physical ecosystem in which an anomaly in one area can trigger subsequent issues throughout the other areas. Traditional monitoring practices typically operate in isolation and, as a result, may be ill-suited to detecting coupled faults, cyber-attacks, and incipient performance degradation. This article proposes a secure artificial intelligence (AI)-enabled cyber-physical framework for fluid, mechanical, and electrical systems. The framework leverages Internet of Things (IoT) sensing, secure edge/fog computing, cross-domain feature engineering, ensemble artificial intelligence, digital-twin-assisted monitoring, and coordinated supervisory control. Fluid pressure, flow rate, and temperature, mechanical speed, torque, and vibration, and electrical voltage, current, frequency, and power factor are continuously acquired and processed. Cross-domain residual indicators are proposed to capture physically inconsistent information and enhance the discrimination of equipment faults from malicious data tampering. A reproducible synthetic digital-twin dataset comprising 10,500 operating instances was constructed, representing normal conditions, fluid, mechanical, electrical faults, and cyber-attacks. Random Forest, Gradient Boosting, Support Vector Machine, Logistic Regression, and a soft-voting ensemble were evaluated. The ensemble achieved an overall accuracy of 91.43%, a weighted precision of 91.59%, a recall of 91.43%, and an F1-score of 91.37%. Cross-domain consistency features improved the overall ensemble accuracy from 89.08% to 91.43% and the recall of cyber-attack from 69.50% to 76.50%. The study reveals that the combination of AI-based condition assessment and secure fog processing with cross-domain consistency can form a tripartite foundation for building a resilient monitoring and intelligent control paradigm of multidisciplinary industrial systems.