Artificial Intelligence Driven Control and Fault Diagnosis in Converter Dominated Power Systems

The increasing penetration of power electronic converters in modern power systems has significantly reduced system inertia and introduced complex nonlinear dynamics, making conventional control and fault diagnosis techniques inadequate. This paper presents an artificial intelligence–driven control and fault diagnosis framework for converter-dominated power systems aimed at enhancing dynamic performance, stability, and reliability. An adaptive artificial neural network (ANN)–based current controller is developed to replace conventional linear regulators, enabling accurate current tracking under varying operating conditions and disturbances. The controller is trained using supervised learning and incorporates online weight adaptation to ensure convergence and robustness. In parallel, a model-based residual generation scheme is integrated for real-time fault detection and diagnosis. Residual signals derived from measured and estimated system responses are processed to identify and classify converter and sensor faults with minimal delay. Time-domain simulations demonstrate that the proposed approach achieves fast transient response, near-zero steady-state error, and smooth control action. Fault scenarios confirm reliable residual separation, rapid fault detection within 0.01 s, and effective post-fault recovery. Quantitative results validate the superior tracking accuracy, learning stability, and diagnostic speed of the proposed framework. Overall, the study highlights the potential of artificial intelligence techniques to provide resilient control and intelligent fault monitoring for future converter-dominated and renewable-rich power systems.