摘要
为了提高复杂系统运行的有效性和可行性,避免系统发生故障时造成巨大的财产损失甚至灾难性的后果,提出了一种基于PCA(Principal Component Analysis,PCA)和HMM(Hidden Markov model,HMM)-支持向量机的故障诊断方法;首先获取故障征兆特征向量,然后采用PCA主成分分析法对特征向量进行降维以减少样本数据的复杂性,将降维后的训练样本数据输入HMM模型和支持向量机模型进行训练得到最终的HMM-支持向量机混合模型,最后将降维后的测试样本数据输入最终的HMM-支持向量机混合模型进行故障诊断;在Matlab仿真环境下进行故障诊断实验,结果证明文中故障诊断精度高达98.9%,与其它方法相比,不仅具有较少的诊断时间而且具有较高的诊断精度,具有很强的可行性。
In order to improve the effectiveness and feasibility of system operation, and avoiding the property loss and disasters due to the fault, a diagnosis method based on PCA （Principal Component Analysis） and HMM （Hidden Markov model） is proposed. Firstly, the fault diagnosis eigenvector was obtained, then the PCA was use to reduce the dimension of the sample data, and the trained sample data was input to HMM model and support vector machine to be trained. Finally, the test sample data with dimension reduced was input to the com- pound model by combing HMM model and SVM model. The simulation experiment was operated in the Matlab environment, and the result shows the method in this paper the fault diagnosis as high as 98. 9%, and compared with the other methods, it not only has less diagnosis time but also higher fault diagnosis accuracy. Therefore, it has strong feasibility.
出处
《计算机测量与控制》
北大核心
2014年第2期370-372,共3页
Computer Measurement &Control
基金
国家自然科学基金项目(61071217)
关键词
故障诊断
主成分分析
支持向量机
隐形马尔科夫
Fault diagnosis
Principal Component Analysis
Support Vector Machine
Hidden Markov model
