一种新的模糊聚类-分级算法及其在流程报警管理中的应用

1 INTRODUCTION Normally, operating management experience of in- dustrial process indicates that the probability one ig- nores alarm is 10-4 after handling one alarm signal and the probability one takes no direct action is 10-5 after hearing one alarm. Experiences also indicate that the probability one ignores alarm will increase to 10-3 from 10-4 if one alarm isn't handled within one minute1]. Therefore, when the second alarm is appearing the probability which it can't be identified is 10…

A New Fuzzy Clustering-Ranking Algorithm and Its Application in Process Alarm Management

Overmany alarms of modern chemical process give the operators many difficulties to decision and diagnosis. In order to ensure safe production and process operating, management and optimization of alarm information are challenge work that must be confronted. A new process alarm management method based on fuzzy clusteringranking algorithm is proposed. The fuzzy clustering algorithm is used to cluster rationally the process variables, and difference driving decision algorithm ranks different clusters and process parameters in every cluster. The alarm signal of higher rank is handled preferentially to manage effectively alarms and avoid blind operation. The validity of proposed algorithm and solution is verified by the practical application of ethylene cracking furnace system. It is an effective and dependable alarm management method to improve operating safety in industrial process.