摘要
主要研究了AdaBoost算法在液压系统故障诊断中的应用。为了解决"一对一"算法和"一对余"算法的分类速度随着训练样本数或类别数的增多而变慢的问题,提出了基于决策树的AdaBoost算法。利用CART算法构造决策树,建立AdaBoost分类器,并根据样本数据的分布情况,使得在决策树中每一个节点的最可分类别尽可能分开。将该算法应用于某型装甲车辆液压系统故障诊断,结果表明:该算法的性能优于其他两个算法,具有更高的通用性。
Application of AdaBoost algorithm on the hydraulic system fault diagnosis was researched.In order to solve the problems of classification speed of "one-versus-one" algorithm and "one-versus-rest" algorithm were slowed down,along with increasing quantity of training samples and classes,the AdaBoost algorithm based on decision tree was proposed.The decision tree was constructed by using CART algorithm,and AdaBoost classifiers were established.Based on the distribution of data samples,the most separable classes could be separated at each node of decision tree.This algorithm was applied to hydraulic system of some armored car for fault diagnosis.The results show that this method has better performance and higher generalization ability than other two methods.
出处
《机床与液压》
北大核心
2012年第9期154-157,共4页
Machine Tool & Hydraulics