A Performance Comparison Study of Digital Twin-Based CBM+ Algorithms for Diesel Engines Utilizing Physical Contradiction Data Synthesis

Key Points

• This study aims to develop a method for synthesizing multivariate virtual fault data using digital twin technology for diesel engines.
• Combines test results and digital twin to synthesize fault data under various scenarios.
• Utilizes FMECA for fault type identification and applies degradation logic for dynamic time-series data.
• Compares predictive performance of deep learning algorithms with other machine learning models.
• Proposed model demonstrates superior early warning performance in fault detection compared to other models.
• Effectiveness confirmed through analysis of predictive performance metrics and lead time for fault warnings.