【作者】 秦树凯； 杨劭伟； 杨英华； 刘晓志；

【Author】 QIN Shu-kai YANG Shao-wei YANG Ying-hua LIU Xiao-zhi College of Information Science and Engineering,Northeastern University,Shenyang 110004,China.

【机构】 东北大学 信息科学与工程学院；

【摘要】 针对实际生产过程中数据大多不服从高斯分布以及故障样本少的特点,提出了基于 ICA-SVM 的故障诊断方法．通过 ICA 建立在线过程监测模型,利用 SVM 训练多种故障分类器．对于监测到的故障,经过故障分类器进行进一步故障确认及故障识别,不但有效地降低了误报率,而且可以精确诊断故障原因．以3水箱系统为例进行实验研究,取得了很好的效果．更多还原

【Abstract】 In the practical production process,the data is not subject to normal probability distribution and the samples of faults are few.Due to this situation,a method of fault diagnosis based on ICA and SVM is presented.The on-line process monitoring model is built by using ICA and multi-fault classifiers are trained by using SVM.The faults monitored are confirmed and identified by the fault classifiers,so the false alarm rate is reduced validly and the failure cause is diagnosed accurately.The simulation results of three water tank system shows the effectiveness of this method.更多还原