In HDD manufacturing processes, a real time monitoring tool used to detect failed HDD components has become an important issue. Once we can monitor the failed components, we can remove them from the production process. This could help the HDD manufacturers to reduce the production time and improve productivity. The focus of this paper is to construct a classification model to detect failed HDD components and also to study and choose potential learned models. The learned models in this paper include the C5.0 decision tree algorithm, CHAID, QUEST, neural network, and the committee machine. Experimental results reveal that the committee machine of C5.0 has the most robust performance and very practical computation time.