基于支持向量机的液压泵故障诊断

Study on Fault Diagnosis Based on SVM for Hydraulic Pump

支持向量机在解决小样本、非线性及高维模式识别问题中表现出许多特有的优势,用ν-支持向量机构造"一对一"多分类算法,应用于ZB40液压泵的故障诊断,取得了较好效果,较神经网络方法,它不必预先提取信号的特征量,只需要少量的故障样本训练分类器,实用性好。

Support vector machine（SVM ）is effective to solve problems of pattern recognition under the condition of finite samples and high dimensional space. Instead of common c - SVM, v - SVM was selected as binary classifier to construct multi - class SVMs, in which the meaning of parameter v was more obvious and could be determined more easily. As an application example, 4 kinds of real fault samples for ZB40 hydraulic pump were classified correctly using this algorithm.