Adaptive filtering and multi-scale MixStyle network for cross-domain fault diagnosis of marine electric thruster bearings

Key Points

• Improved fault diagnosis was observed with an advanced multi-scale MixStyle network, leading to enhanced accuracy.
• The study demonstrated a reduction in classification errors by over 25% using adaptive filtering techniques.
• Analysis employed a novel adaptive filtering approach across various marine environments and fault scenarios.
• These findings underscore the potential for cross-domain applications in marine systems, though real-world validation is needed.