GMM-BinSeg: A Data Segmentation Method for CFRP/Aluminum Stacks Drilling– Countersinking Monitoring Signal

Due to their exceptional physical properties and reduced weight, carbon fiber-reinforced plastic (CFRP)/ aluminum stacks are extensively utilized in the aviation industry as essential structural components for bearing high loads. These stacks must undergo single-shot drilling and countersinking for fastener installation. However, drilling–countersinking process could introduce defects in the CFRP structure, such as delamination, which could jeopardize the reliability of the aircraft structure. To avoid defects, sensors are commonly employed in the aviation industry to monitor the drilling–countersinking process. However, these sensors may collect excessive amounts of noninformative data, prolonging signal processing time and degrading the monitor precision. To resolve this problem, this article proposes an approach named Gaussian mixture model (GMM)-BinSeg for identifying informative data acquired during material removal. First, a constraint parameter for primary change points (PCPs) searching is automatically estimated using a designed Butterworth filter and GMM. Then, the PCPs for drilling signal segment are searched using a built binary segmentation (BinSeg) algorithm under the constraint of the obtained parameter. Finally, the countersinking signal segments are recognized based on the PCPs and machining parameters of the cutting tool. Experiments with ten distinct signals illustrate that GMM-BinSeg can precisely separate out CFRP drilling signals, aluminum drilling signals, and countersink signals from raw data.