Predictive Maintenance Framework for Hard Disk Media Production

Manufacturing yield control for hard disk media is getting tougher. This is proportional to the tremendous increases of its bit/inch storage capacity. With the significantly difficult lithography process will be involved and drastically increase in number of product total output volume that will be required in near future, conventional maintenance type is no longer feasible. This paper proposes a novel framework for the implementation of predictive maintenance in hard disk media production. A novel technique to visualize the temporal data into pattern that can be trained with machine learning algorithm is introduced. Predictive models were produced after dealing with imbalance datasets issue, ensemble datasets and data cleaning. Experimental results have indicated that the proposed framework is successful.