基于SVM的旋转机械转子故障预示研究

Research on Fault Prediction of Rotating Machine Based on SVM

支持向量机包括支持向量回归机和支持向量分类机。本文提出了一种用于旋转机械转子故障预示的方法,通过支持向量分类机(SVC)对旋转机械转子故障进行分类并建立故障分类器,利用支持向量回归机(SVR)对转子运行状态趋势进行预示,并将预示结果输入到SVC以判断预示结果的属性。对支持向量回归机进行了仿真研究。将支持向量机与神经网络算法从理论和实验研究两个方面进行了对比研究,结果表明,该方法具有较好的故障预示能力。

Support vector machine is composed of support vector regression and support vector classification.A method for prediting the fault of rotating machine was presented in the paper.By using support vector classification（SVC）to classify the fault of rotating machine,a fault classification was built.It used support vector regression（SVR）to predict the condition trend of rotating machine,input the results of prediction to support vector classification for judging attribute of the results.It made simulating research about support vector regression.It conducted the contrast research between SVM and BP neural network algorithm from the theory and the experimental study.The results show the method has a good capacity for fault prediction.