摘要
基于支持向量机开发的航空发动机磨损趋势预测技术运用结构风险最小化准则,可通过内积函数将低维空间的非线性问题转化为高维空间的线性问题,在发动机滑油光谱监控中十分有用。阐述了支持向量机的原理和数学模型,建立了适用于航空发动机磨损趋势预测的支持向量机回归模型和自回归模型,并对支持向量的核函数模型参数进行了讨论。对实际发动机的润滑油光谱监控数据趋势预测结果表明,基于支持向量机回归模型的趋势预测技术具有很高的预测精度和很强的实用性,可有效提高通过润滑油光谱监控技术预报航空发动机磨损类故障的预测能力。
Support vector machine(SVM) modeling method used in spectral oil analysis,which is based on principle of structural risk minimization,can transform non-linear model in lower-dimensional space into linear model in higher-dimensional space.The theory of SVM and its mathematical models were introduced,and the models of regression and self-regression based on SVM were established to forecast abnormal wear in aviation engine.The parameters of SVM kernel functions were discussed to fit actual application in aviat...
出处
《润滑与密封》
CAS
CSCD
北大核心
2008年第5期84-87,共4页
Lubrication Engineering
关键词
航空发动机
支持向量机
润滑油光谱分析
磨损
趋势预测
aviation engine
support vector machine（SVM）
spectral oil analysis
wear
trend forecast
