Increasing levels of complexity of industrial automation systems in the next decade as part of Industry 4.0 call for scalable, real-time Fault Monitoring and Diagnostic (FMD) to guarantee the resilience and continuity of the operation of such complex systems. This paper proposes an AI-Driven Real-Time Fault Monitoring and Diagnostic Framework, which incorporates complex deep learning architectures (hybrid CNN-LSTM and Transformer based models) to classify multi-domain faults well on mechanical, electrical and hybrid systems. Compared with the impossible implementation that applies centralized data processing as in traditional approaches, the developed framework capitalizes on edge computing that serves as a technology to conduct ultra-low latency fault detections domainly where data is generated, leading to much less data transmission and response time. The adaptive CTS self-learning whatever module constantly improves diagnostic precision through any newly arisen fault pattern generation, with embedded XAI (explainable AI) components increasing transparency and interpretability for operators and regulation. Moreover, the framework includes a closed-loop fault handling scheme to activate real-time fault correction response, significantly shortening the down-time and minimizing operation risks. The system is based on a scalable system architecture that can support huge industrial environment, which includes huge heterogeneous data feed, and ensure that it operates reliably even under challenging conditions. Extensive validation using real-world industrial case studies demonstrates the framework’s superior performance, addressing key gaps identified in prior research and establishing a comprehensive solution for intelligent, real time fault management in modern industrial automation.