A Hybrid Bee-Markov And Cat Kernel Vector Optimized Intelligent Fault Diagnosis Framework For Multilevel Inverter Systems

Abstract

Multilevel inverters are one of the widely used modern power electronic systems, which play an important role in converting DC power into controlled AC output suitable for high-power and industrial applications. Despite their advantages in high-power systems, reliable fault diagnosis and classification in multilevel inverter structures remain challenging due to complex switching operations and signal disturbances. These effects often overlap with fault-related signal characteristics, making reliable diagnosis very difficult. To overcome this issue, this study introduces a hybrid intelligent framework called the Hybrid Bee-Markov and Cat Kernel Vector Optimized Intelligent Fault Diagnosis Framework for Multilevel Inverter Systems. To regulate switching state distribution, the proposed approach applies an optimized Bee-based Markov modulation strategy.  This helps to reduce, suppress EMI effects, harmonic distortion and produce smoother output waveforms. The technique makes it possible to more clearly identify fault-specific signal components by enhancing the standard of the resultant inverter waveform. Additionally, the presence of several flaws may result in nonlinear interactions that make diagnosis more difficult. A kernel vector principal machine is used to reduce feature dimensionality and improve classification performance in order to capture these intricate patterns. The suggested framework achieves 99.5% precision, 98.5% recall, and 91.3% classification accuracy, according to experimental findings.