Gear Remaining Useful Life Prediction Based on Grey Neural Network

The condition monitoring data of gears is asymmetric distributed, moreover, sampling is usually conducted discontinuously in practice. Thus makes it difficult to predict gear remaining useful life accurately considering the two reasons above. In this paper, a fusion method is proposed using Elman Neural Network to modify residual series of grey model since Elman Neural Network performs better on feeding back and accuracy than BP network. The proposed method takes the advantages of both GM (1, 1) for data mining and Elman neural network for feedback. Experimental data is used to validate the proposed method. The results illustrate that the integrated method has a high prediction capability compared with GM model. In addition the proposed method is a promising approach for life prediction in the case of small sample, incomplete and discontinuous sampling data.