Review of fault detection techniques in power converters: Fault analysis and diagnostic methodologies

Power converters are crucial components in modern electrical systems, and their reliable operation is paramount for maintaining system performance and safety. However, they are susceptible to various faults, including overcurrents, voltage sags, sensor failures, and asymmetrical voltage in the grid, which can adversely affect the converter's performance and the entire system. Hence, detecting and diagnosing faults in power converters promptly is crucial. Open circuit (OC) and short circuit (SC) faults are common in power electronics, especially in transistors, and can be caused by overvoltage, overcurrent, elevated temperatures, and other factors. An OC in a transistor means it cannot conduct current, resulting in a voltage drop in the circuit and power reduction. Conversely, a SC in a transistor means that current flows through it unrestricted, potentially damaging the circuit and power supply. Sensor failures can also cause control problems, leading to the above issues. Additionally, grid or external load failures can cause control problems and damage the converter's internal components. These faults can be particularly dangerous in Voltage Source Converters (VSC), affecting their ability to maintain synchronization of the output signal with the power supply. Additionally, SC faults can damage other VSC components, potentially resulting in total system failure. Therefore, it is crucial to have measurement and fault detection strategies that can, through digital signal processing, quickly detect and locate problems in transistors, enabling a faster and more effective response to prevent damage to other system components and ensure the safety of the equipment and the power grid. This work aims to evaluate various techniques used in recent years for detecting faults in power converters, particularly in VSC transistors. The focus is on assessing the speed of detection, computational complexity, robustness, and efficiency, as well as the difficulty of tuning, calibration, and adjustment of these techniques. By evaluating these factors, this study seeks to provide insights into the strengths and weaknesses of different measurement and fault detection strategies, which can be valuable for researchers and practitioners in the field of power electronics instrumentation, which is critical for extending the lifetime of VSC. Research indicates that signal processing techniques excel in rapid fault identification but are accompanied by notable tuning complexities. In contrast, although slightly slower, model-based techniques are recognized for their efficient and straightforward tuning, which has led to extensive research in this field. However, data-driven approaches exhibit slower failure detection rates than the aforementioned techniques. The complexity of fault location algorithms escalates with implementing multilevel converters; nonetheless, their resilience and reconfiguration capabilities surpass those of two or three-level converters. Integrating multilevel techniques with PWM modulation in oversized three-level cells eliminates expensive inductive filters. It enhances tolerance to VSC failures, resulting in minimal power reduction during in fault thus