In multiple operation forging processes, missing parts in some dies during production is a critical problem. The objective of this paper is to develop an effective missing part detection method through automatic classification of continuous production data. In the paper, a new feature extraction and sequential classification decision rule is developed, which aims to enhance the detection sensitivity and robustness. In the methodology development, the data segmentation is conducted at the first step based on an offline station-by-station test in a forging process. Then, PCA (Principal Component Analysis) is used as the data transform for the selected data segment of the training data sets under different missing part conditions. The effectiveness of the selected features is justified to minimize the misclassification probabilities among different classes. Finally, a decision rule is proposed for online classification of different missing parts conditions. A case study using a real-world forging process is provided to demonstrate the analysis procedures and effectiveness of the proposed methodology.

This content is only available via PDF.
You do not currently have access to this content.