摘要
提出了一种利用支持向量机进行可靠度时序预测的方法。通过重构相空间的饱和嵌入维数,确定支持向量机的最佳输入变量;利用支持向量机强大非线性映射能力、网络结构的自动最优化特性,实现时间序列的非线性预测。最后,应用于某型发动机涡轮增压器可靠度预测,结果证明该方法具有较高的预测精度和较强的推广能力,对于一般意义上的可靠度监测具有重要的价值。
A method of time series prediction of reliability based on support vector machines is proposed. The network＇s input variable number is determined through computing reconstruct phase space＇s saturated embedding dimension; It makes use of support vector machines＇ strongly nonlinear mapping ability, and network＇s structure is optimally auto-created. Application results in turbochargers of an engine show that the presented method possesses much better precision. The method is important for general reliability prediction.
出处
《火力与指挥控制》
CSCD
北大核心
2007年第11期118-120,共3页
Fire Control & Command Control
关键词
可靠度
时序预测
涡轮增压器
支持向量机
reliability time series prediction, turbocharger, support vector machines
