Fault Diagnosis and Tolerant Strategy for Triple-Port Hydrogen Converter Using SSA-Optimized Random Forest Algorithm

Multi-port hydrogen converters are crucial for hydrogen production power sources, requiring high reliability. However, current fault diagnosis and fault-tolerant solutions for such converters are inadequate, relying on numerous sensors, offering limited diagnostic accuracy, and struggling with multiple switch fault detection. Additionally, existing fault-tolerant strategies cannot ensure stable power for electrolyzers or enable power-sharing between ports. To address these challenges, this paper presents a method for diagnosing open-circuit faults and implementing fault tolerance in triple-active bridge (TAB) hydrogen converters. The method uses a sparrow search algorithm (SSA) optimized random forest (RF) algorithm to achieve high accuracy, diagnosing faults in multiple switches using fewer sensors. For fault-tolerant operation, auxiliary switches maintain stable system output and enable power-sharing between ports after a fault. This ensures the hydrogen electrolyzer operates efficiently and reliably. Simulations validate the solution's effectiveness, demonstrating its economic viability and reliability, providing a foundation for safe and efficient green hydrogen production.