原子能科学技术2011，Vol.45Issue(8)：972-976,5.
基于支持向量回归的设备故障趋势预测
Fault Trend Prediction of Device Based on Support Vector Regression
宋梅村 1蔡琦1
作者信息
- 1. 海军工程大学船舶与动力学院,湖北武汉430033
- 折叠
摘要
Abstract
The research condition of fault trend prediction and the basic theory of support vector regression (SVR) were introduced. SVR was applied to the fault trend prediction of roller bearing, and compared with other methods (BP neural network, gray model, and gray-AR model). The results show that BP neural network tends to overlearn and gets into local minimum so that the predictive result is unstable. It also shows that the predictive result of SVR is stabilization, and SVR is superior to BP neural network, gray model and gray-AR model in predictive precision. SVR is a kind of effective method of fault trend prediction.关键词
支持向量回归/BP神经网络/灰色模型/灰色-AR模型/故障趋势预测Key words
support vector regression/ BP neural network/ gray model/ gray-AR model/ fault trend prediction分类
计算机与自动化引用本文复制引用
宋梅村,蔡琦..基于支持向量回归的设备故障趋势预测[J].原子能科学技术,2011,45(8):972-976,5.
