Artificial Neural Networks for Tool Wear Prediction Based on Sensor Fusion Monitoring of CFRP/CFRP Stack Drilling

Abstract

An intelligent sensor monitoring procedure was implemented to monitor the drilling of carbon fiber reinforced plastic (CFRP)/CFRP stacks used in the assembly of aircraft fuselage panels; the signals from these sensors were then used to develop an artificial neural network-based cognitive paradigm to predict tool wear, which would allow on-line decision making regarding tool replacement. A multiple sensor system, capable of acquiring signals relative to thrust force, torque, and acoustic emission RMS, was employed during experimental drilling tests, under different rotational speed and feed conditions. Advanced sensor signal processing techniques, including signal conditioning and segmentation, as well as statistical feature extraction and data fusion, were implemented on the acquired signals. Selected statistical features extracted from the multiple sensor signals in the time domain were combined via sensor fusion techniques to construct sensor fusion pattern vectors. These were then fed to artificial neural networks for pattern recognition, with the goal of finding correlations which would allow the prediction of the corresponding tool wear. The tool wear prediction performed by the artificial neural network can be utilized to support decision making at the appropriate time for worn tool replacement, which is extremely useful for drilling automation, as well as for estimating the quality of the drilled holes.