A performance degradation assessment method for complex electromechanical systems based on adaptive evidential...

Area of Science:

• Engineering
• Artificial Intelligence
• Data Science

Background:

• Traditional evidence reasoning (ER) rules struggle with dynamic systems where indicator importance shifts over time and under varying conditions.
• Existing ER methods lack adaptability for tasks like performance degradation assessment in complex electromechanical systems.

Purpose of the Study:

• To propose an adaptive evidence reasoning (AER) rule capable of dynamically adjusting indicator weights based on time and working conditions.
• To address the limitations of traditional ER rules in handling time-varying and condition-dependent data.

Main Methods:

• Developed an adaptive evidence reasoning (AER) rule with adaptive weight operations for time and working-condition divisions.
• Utilized the CMA-ES algorithm for optimizing AER model parameters.
• Validated the AER rule through two case studies: computer numerical control (CNC) experiment and turbofan aeroengine simulation.

Main Results:

• The AER rule demonstrated effective dynamic weight adjustment for performance degradation assessment.
• Case studies confirmed the superiority of the AER rule over traditional methods in handling changing system dynamics.
• The method proved to be effective and practical for real-world applications.

Conclusions:

• The proposed adaptive evidence reasoning (AER) rule offers a robust solution for analyzing dynamic systems with uncertain and time-varying data.
• AER significantly enhances the accuracy and reliability of performance degradation assessment in complex electromechanical systems.
• The AER rule provides a practical and effective approach for dynamic weight adjustment in evidence-based decision-making.