摘要
采用大型设备正常运行的传感器数据,通过基于多元统计的主元分析方法建立训练矩阵,利用预测误差进行故障分析工作。不同程度的故障,来分析主元分析方法的检测效率可行性。结果表明,主元分析方法分析故障检测明显,不同传感器的不同故障会对故障诊断存在一定的差异,可行性非常高。
The use of large equipment uptime sensor data through PCA based on multivariate statistical analysis method to establish training matrix, using the prediction error for fault analysis. Varying degrees of failure, principal component analysis method to analyze the efficiency of detecting viability. The results showed that the main element method analysis of fault detection significantly, different sensors be different faults Fault diagnosis there are some differences, feasibility is very high.
出处
《电脑知识与技术》
2013年第2X期1340-1342,共3页
Computer Knowledge and Technology
