AI-Augmented Fault-Tolerant Control Architecture for High-Reliability Electric Drive Systems in Smart Manufacturing

Abstract

Electric drive systems are considered vital elements in the contemporary intelligent manufacturing systems where uninterrupted production, exceptional accuracy, and a minimal time of downtime are vital in upholding production and operational efficiencies. Nevertheless, these systems are prone to multiple failures that lie in sensors, power electronic transformer, motors and control subsystems that may severely affect system reliability and result in the costly production outage. To overcome these obstacles, this paper suggests an artificial intelligence (AI)-enhanced fault-tolerant control (FTC) architecture that could enhance the reliability, robustness, and resiliency of electric drive systems that can be implemented in the Industry 4.0 contexts. According to the proposed architecture, machine learning-based fault diagnosis, predictive analytics, and adaptive control mechanisms are combined with traditional control strategies and allow early fault detection, proper classifying, and automating controller reconfiguration. Within the framework suggested, the data of the operation of several sensors, including current, voltage, temperature, and vibration, is processed with the help of high-level signal processing and feature extraction algorithms. The real-time monitoring and anomaly detection are then carried out on the deep learning models in order to allow the system to pick off future faults at their initial appearance. After a fault has been identified, a decision-making process triggers an adaptive control course of action using model predictive control (MPC) to continue stabilising the system and maintain its operation even with partial failure of the system. The suggested AI-based architecture provides self diagnostics and self-recovery features that promote the continuity of operations at smart factories. Evaluations on a simulation level reveal that the proposed method is much more competent in terms of fault detection accuracy, minimization of system downtime, and the overall reliability of the operation in relation to the traditional fault-tolerant control over the system. As such, the framework proposed will be a potential solution to building intelligent, resilient, and autonomous electric drive systems to be used in the next generation of smart manufacturing processes.