Research on sensor fault diagnosis in digital intelligence monitoring system of process production safety
To ensure the accuracy of process production safety monitoring data,a fault diagnosis method that combined kernel principal component analysis and cumulative residual contribution rate method was proposed.A multi-level digital intelligence monitoring system architecture based on the"perception-aggregation-decision"paradigm was put forward.For the sensors in the perception layer,a fault detection model was constructed based on kernel principal component analysis,and the fault sen-sors were located by the cumulative residual contribution rate method.The continuous casting operation area in the DYTG converter plant was selected as the case analysis.The results show that the average detection rate and average false detection rate of the proposed fault diagnosis approach on the SPE index are 95.28％and 2.61％,respectively,while those on the T2 index are 84.36％and 1.71％,respectively.Furthermore,it can accurately locate the fault sensor for four kinds of fault forms.The research results are conducive to reduce the maintenance cost of the monitoring system,and improve the control level of process production safety.
process productionsensorfault diagnosiskernel principal component analysiscumulative residual
万方数据
维普
