A novel hybrid explainable artificial intelligence modelling approach for smart manufacturing.

Area of Science:

• Manufacturing Engineering
• Artificial Intelligence
• Physics-Based Modelling

Background:

• Complex manufacturing processes face accuracy and explainability challenges.
• Physics-based models offer interpretability but lack accuracy.
• Data-driven models are accurate but lack transparency, hindering trust.

Purpose of the Study:

• To develop a novel hybrid modelling framework integrating physics-based models with explainable AI.
• To enhance accuracy and maintain transparency in manufacturing process modelling.
• To address limitations of existing hybrid approaches that retain black-box AI components.

Main Methods:

• Proposed a novel framework intrinsically integrating physics-based models with explainable AI.
• Utilized explainable AI to correct inaccuracies in physics-based models.
• Demonstrated effectiveness via a case study in ultra-precision diamond turning.

Main Results:

• Achieved high accuracy in predicting cutting tool positions.
• Ensured transparent and traceable decision-making.
• Successfully corrected modelling inaccuracies through the hybrid approach.

Conclusions:

• The proposed hybrid framework offers a solution for accurate and explainable manufacturing process modelling.
• This approach enhances trust and adoption of AI in critical manufacturing applications.
• Effective for real-time prediction tasks, such as in ultra-precision diamond turning.