An Equipment Failure Prediction Accuracy Improvement Method Based on the Gray GM(1,1) Model

摘要

Failure prediction is an important and difficult research aspect. Firstly we introduce the concepts of Gray GM(1,1) model. In this paper, addressed to the existing problems of GM(1,1) in the prediction accuracy aspect, that is often affected by the smoothness of the sequence of collected failure datum. In order to improve the prediction accuracy, we introduce the concept of transforming trying to improve the smoothness of the original failure datum sequence. That is using GM(1,1) model to implement the transforming. And apply it to equipment prediction. Based on many collected failure datum, we use the proposed method. Then MATLAB simulation is applied to implement the residual test and posterior difference test. The results show that the above methods are valid and accuracy test meets the requirements. And the program proposed in this paper is shown to improve accuracy on the failure prediction.