Intelligent Real Time Fault Diagnosis Framework for Cyber Physical Electronic Systems Using Adaptive Learning Techniques

The growing sophistication of cyber-physical electronic systems demands intelligent real-time fault diagnosis with the ability to cope with dynamic settings at low latency and high reliability. Traditional methods like deep learning and symbolic reasoning are limited by lack of interpretability, flexibility, or addressing uncertainty in time-varying signals. To address these issues, this study introduces an adaptive learningbased fault diagnosis architecture named AR-FDNet (Adaptive Reinforcement-Fuzzy Diagnosis Network) that combines deep frequency-domain feature extraction and reinforcement-based classification. The model learns in real-time from changing fault patterns dynamically, supporting correct classification even in noisy and complex operational conditions. Experiments on the Electrical System Fault Diagnosis Dataset verify the efficacy of the framework, with 95% recall, 97% accuracy, F1-score 95% and 96 % precision. The proposed approach brings enhanced generalization and fast responsiveness, which makes it extremely suitable for real-world industrial process control applications, smart grids, electric vehicle applications, and embedded safety-critical systems where continuous monitoring and adaptive diagnosis are critical.

Ҳали таржима қилинмаган