基于支持向量机的摩擦学系统状态判别方法研究

Research on State Discriminant Method of Tribology System Based on Support Vector Machine

传统的摩擦学系统状态的判别采用逐步判别分析法,该方法以油液监测历史数据为依据,通过从不同的油液监测方法中获取信息,构造出状态分类判别数学模型来进行状态判别,具有建模样本量大,建模时间长,对建模人员要求较高的缺点。通过引入支持向量机分类方法,缩短了对摩擦学系统进行状态判别的时间,提高了分类效率,最后通过实例说明了该方法的有效性、实用性和良好的推广应用前景。

The traditional state discriminant of tribology system uses stepwise discriminant analysis method. The method utilizes the information obtained from the history oil monitoring data to construct state classification mathematic model to distinguish the system state, it is complicated and need large quantity of samples and a long time to model. Support vector machine method was introduced to distinguish the tribology system state. The discriminant result of an example shows that support vector machine method shortens the discriminant time and enhances classification efficiency.