Dressing Tool Condition Monitoring through Impedance-Based Sensors: Part 2-Neural Networks and K-Nearest Neighbor...

Area of Science:

• Mechanical Engineering
• Materials Science
• Sensor Technology

Background:

• Grinding operations require precise dressing tool condition monitoring for efficiency and quality.
• Existing methods for monitoring dressing tool wear can be limited in accuracy and automation potential.
• The electromechanical impedance (EMI) technique offers a promising avenue for real-time sensor-based monitoring.

Purpose of the Study:

• To develop an optimal excitation frequency band selection method for EMI-based sensor monitoring of dressing tool condition.
• To utilize multi-layer neural networks (MLNN) and k-nearest neighbor (k-NN) classification for intelligent feature selection.
• To validate the proposed approach using experimental data from industrial dressing tests.

Main Methods:

• Implementation of the electromechanical impedance (EMI) technique for sensor monitoring.
• Application of multi-layer neural networks (MLNN) for data processing and feature selection.
• Utilizing k-nearest neighbor (k-NN) classifier for optimal frequency band selection.
• Analysis of impedance signatures and damage indices from experimental dressing tests.

Main Results:

• Successful identification of damage-sensitive features through optimal frequency band selection.
• Validation of the approach using industrial stationary single-point dressing tools.
• Achieved a general overall error rate lower than 2% in damage detection.
• Demonstrated robust contribution to the automation of grinding and dressing operations.

Conclusions:

• The proposed EMI-based sensor monitoring approach effectively diagnoses faults in dressing operations.
• The intelligent system for optimal frequency band selection enhances the accuracy of tool condition monitoring.
• This method significantly contributes to the efficient automation of industrial grinding processes.