基于稀疏故障演化判别分析的故障根源追溯

Sparse Fault Degradation Oriented Fisher Discriminant Analysis Based Fault Trace

火力发电过程规模庞大,过程变量众多.在故障发生时,部分变量将受故障扰动影响偏离正常运行状态,分析这些故障变量之间的故障传递关系并找到根源故障变量,这对于定位故障位置以及排除故障具有重大意义.因此采用稀疏演化判别分析方法(FDFDA)隔离火电过程中的故障变量,随后对隔离到的故障变量进行格兰杰因果分析,追溯故障根源.

The thermal power processes contain many variables, while only a part of variables will be influenced when the fault occurs. It is meaningful to analyze the fault causalities, which may help track root fault reasons and locate abnormal components. Therefore, for the fault processes, this paper isolates the faulty variables on basis of sparse fault degradation oriented fisher discriminant analysis (FDFDA) and then analyzes the causalities between different variables by Granger Causality analysis for identifying root faulty reasons.