MODELING OF RELIABILITY PREDICTION FOR LUBRICANT LAYER IN HARD DISK DRIVE

Abstract

The artificial neuron network model with back propagation algorithm was developed successfully for reliability prediction of lubricant layer. The input variables are overwrite (OW) value by radius that collected from the hard disk drives which testing for 6 weeks. The result is indicated that the optimized artificial neural network model which is suitable for reliability prediction of lubricant layer is 41,45,1 that consistent of 41 input variables, 45 nodes in hidden layer and 1 output. The predicted results of this model provide the mean square error is 1.68. From the testing model results that compared the experiment output and predicted output observed the model can achieved the accuracy performance 91.74%.