The application of artificial intelligence in grinding operation using sensor fusion

Abstract

The application of multi-sensor systems for the monitoring of machining processes is becoming more commonplace to improve productivity, automation and reliability. In order to enhance knowledge in this area of applications, this study proposes a novel approach for the continuous on-line condition monitoring of grinding operation using low cost infrared and visual imager alongside with more commonly used sensors i.e. AE sensor, accelerometer and dynamometer. To achieve this aim a multi-sensor system is developed and installed for the monitoring of grinding operation. The signals acquired and analyzed by the system include visual, thermal, force, vibration and AE under different grinding conditions. Image processing techniques are used to establish that an increase in sparks within grinding zone results in rise of grinding zone temperature, which in turn results in increased surface roughness. Signal processing techniques are used to establish that dressing of wheel is most influential factor for surface roughness of workpiece. Artificial intelligence is then used successfully on both infrared and visual data to establish an automated continuous on-line monitoring system for grinding operation with an accuracy of 95 percent.