Related Experiment Video
Updated: Jun 27, 2026
07:58
Automatic Laser-based Geometry Capture for Finite Element Analysis of Weld Beads
Published on: July 25, 2025
Sensor-Derived Mechanism-Informed Prediction of Section-Level Residual Profile Error in Robotic Blade-Edge Finishing
Zhuohang Gao1, Xi Zeng1, Zhenyu Cai1
1College of Mechanical Engineering, Zhejiang University of Technology, 288 Liuhe Road, Xihu District, Hangzhou 310023, China.
Sensors (Basel, Switzerland)
|June 26, 2026
Summary
This study introduces a new framework to predict turbine blade edge errors using sensor data and machine learning. The developed descriptors accurately predict residual profile errors, improving quality assessment in robotic belt finishing.
Area of Science:
- Manufacturing Engineering
- Robotics
- Materials Science
Background:
- Robotic belt finishing of turbine blades faces challenges in controlling residual profile errors.
- Local edge radius, contact compliance, and incoming profile state are key factors influencing final error.
Purpose of the Study:
- To develop a sensor-derived, mechanism-informed framework for predicting section-level root-mean-square (RMS) residual profile error.
- To establish interpretable section-level descriptors for improved prediction accuracy.
Main Methods:
- Utilized online force measurements, robot/process records, CAD geometry, and CMM profiles.
- Introduced three coupled descriptors: load-to-radius ratio, force-radius-mismatch interaction, and normalized radius mismatch.
- Evaluated four Gaussian process regression (GPR) configurations, a training-mean predictor, and a ridge-regression baseline.
Main Results:
- The proposed descriptors demonstrated clear predictive value in blade-wise evaluation.
- Ridge-B3 achieved the best deterministic accuracy (RMSE = 1.0285 µm, R² = 0.7759).
- GPR-B3 provided predictive intervals and descriptor-attribution information, crucial for uncertainty-aware assessment.
Conclusions:
- Descriptor construction is paramount for deterministic accuracy in predicting residual profile errors.
- Gaussian process regression offers an uncertainty-aware modeling layer for risk-aware quality assessment in robotic finishing.
