Artificial Intelligence to Enhance the Drilling of Composites

Abstract

Advances in the study of fibre reinforced polymers have led to a huge interest in applying them to multiple fields as an alternative to more costly materials such as their metallic counterparts. However, if the machining of fibre reinforced polymers is done incorrectly this will lead to many defects. Such problems might lead to the underutilization of the fibre reinforced polymers; therefore, optimizing the drilling process is necessary to eliminate the defects. Drilled composite panels must be free of defects for them to succeed in their structural applications. Therefore, the objective of this study is to enhance the drilling process of composites by developing a machine learning mathematical model which will be able to predict the failure behaviour considering the delamination area and fibre pullout area as the response variables in terms of a set of process parameters. The proposed methodology consists of several steps to assess the quality of the drilled hole. Firstly, the composite material selection discusses the process of selecting a specific composite material taking into consideration the material’s properties. Secondly, the experimental setup describes how the experiments were conducted and what machines and tools were used in the process. Thirdly, different inspection techniques are proposed to monitor the quality of a drilled hole during the drilling process and after. Lastly, the modelling of the response variable in terms of the process parameters and the process monitoring variable. Based on a specific sample thickness and tool diameter for the composite panel the machine learning model developed was able to provide the optimum feed rate and spindle speed values needed to attain the minimum delamination area and fibre pullout area. In addition, the in-process monitoring identified a threshold value for the delamination area in terms of the force exertion.