Open-Circuit Fault Diagnosis of Modular DC-DC Converter Based on Multi-Layer Perception

The analysis of fault diagnosis has become crucial in improving the reliable performance of DC-DC converters utilized in power micro-grid (MG) systems, electric vehicles (EVs), and photovoltaic (PV) systems. In recent times, novel data-driven-based algorithms have been introduced for fault detection, surpassing the capabilities of signal-processing-based methods. These algorithms excel at continuously monitoring parameters and accurately predicting faults before they occur. This paper presents a neural network model, specifically a multi-layer perception (MLP), for the detection of open-circuit faults (OCFs) in MOSFETs within a modular DC-DC converter. The methodology of MLP approach is described in detail, and a comparative analysis is conducted with alternative models. The proposed model attains a remarkable prediction accuracy of $\mathbf{9 9 \%}$ under noise-free conditions, exhibiting its prowess. Furthermore, it demonstrates an accuracy exceeding 90% when exposed to varying levels of noise in the input signal. The outcomes of this research contribute to advancements in fault diagnosis methodologies, enhancing the operational reliability of DC-DC converters in diverse applications.