基于SVM的柴油机油液光谱预测模型研究  

Study on Prediction Model of Oil Spectrum Based on Support Vector Machines

作　　者：范红波[1] 张英堂[1] 任国全[1] 罗鸿飞[1] 

机构地区：[1]军械工程学院,河北石家庄050003

出　　处：《润滑与密封》2006年第11期148-150,共3页Lubrication Engineering

摘　　要：针对传统油液光谱预测模型存在的缺陷,利用支持向量机(SVM)的回归算法建立光谱预测模型。支持向量机的非线性回归方法得到的回归函数对光谱历史数据有良好的逼近效果,通过SVM回归函数的外推计算建立的SVM回归预测模型能有效消除换油和采样间隔不一致对模型预测效果的影响,具有更广泛的应用范围。Aimed to the faults in traditional prediction model of oil spectrum, a new model was established by means of regression algorithm of support vector machines （SVM）. The extrapolated calculation of the regression function of SVM was used to establish prediction model. The regression function that got from the nonlinear regression method of SVM can fit estimated value to historical data very well. The model can effectively avoid the influence of oil change and sample interval disaccord. This new model can be applied in a larger range.

关 键 词：油液光谱分析 SVM 柴油机 预测模型 

分 类 号：TH165.3[机械工程—机械制造及自动化]