MONITORING PROCESS IN TURNING OPERATIONS FOR CRACKED MATERIAL ALLOY USING STRAIN AND VIBRATION SENSOR WITH NEURAL NETWORK CLASSIFICATION

Abstract

Surface finish and monitoring tool wear is essential for optimization of machining parameters
and performing automated manufacturing systems. There is a very close relationship between tool wear and machining material parameters as surface roughness, shrinkage, cracks, hard particle ... etc. Monitoring of manufacturing processes plays a very important role to avoid dawn time of the machine, or prevent unwanted conditions such as chatter, excessive tool wear or breakage. Feature extraction and decision making is a matter of considerable interest for condition monitoring of complex phenomena with multiple sensors.  In this work, the implementation of a monitoring system utilizing simultaneous vibration and strain measurements on the tool tip is investigated for the shrinkage and crack of cast iron work piece. Machining parameters taken into consideration are cutting speed (116.5 and 136.6) m/min, feed rate (0.17 and 0.23)rev/min respectively and depth of cut (1) mm. Data from the machining processes were recorded with one piezoelectric strain sensor type (PCB 740B02) and an accelerometer type (4370), each coupled to the data acquisition card type (9111 DR). There were 22 features indicative of crack were extracted from the original signal. These include features from the time domain (mean, STD, crest factor, RMS, kurtosis, variance), frequency domains (power spectral density), time-series model coefficient (AR) and four packet features extracted from wavelet packet analysis (RMS, STD, kurtosis, crest factor). The (2x1) self organizing map neural network was employed to identify the crack and shrinkage effect on the tool state. The program used with this process is MATLAB V.6.5. As a result of the present work, we have an SOM model can classifying the crack with minimal error.