Real-time Grinding Wheel Condition Monitoring Using Linear Imaging Sensor

Abstract

Grinding is a widely used process in precision finishing of machined part surfaces. Grinding wheels, the key component of the grinding process, consist of a mixture of abrasive grains and bonding materials with or without a metal core. During the grinding process, the grits are either removed or broken on the wheel surface due to multiple mechanisms. Such tool wear on grinding wheel determines the quality of the ground surface as well as the part dimension accuracy and machining efficiency. This work presents a new method to monitor the health status of the grinding wheel using linear Charge-coupled Device (CCD) sensor, which captures one-dimensional grayscale images of the grinding wheel surface. The statistical features were extracted from the sensor data to estimate the present tool wear. Compared to general camera imaging method, the proposed approach is able to achieve high speed sampling with the CCD sensor to scan across the width of wheel in milliseconds, which enables the method a potential solution for monitoring the wheel condition in real-time. The method was tested by capturing surface images of a silicon carbide wheel on a commercial grinding machine. Statistical features such as standard deviation, kurtosis, and entropy were extracted from the grayscale color intensity of the image data and compared to the measured wheel life represented by the counted grinding cycles. The statistical features were fused by an Artificial Neural Network (ANN) model to estimate the life of the grinding wheel. Experimental results show a good match between the estimated and true wheel life with an average error less than 5%.