Similarity Analysis of Sequential Alarms for Plant Alarm System Rationalization – A Case Study of Ethylene Plant –

A poor alarm system might cause sequential alarms, which are a collection of many alarms that almost always occur simultaneously with specific time lags within a short period of time. The sequential alarm reduces the ability of operators to cope with plant abnormalities because critical alarms are buried under many unnecessary ones. In the previous research, we proposed an analysis method of similarities between sequential alarms by using the levenshtein distance, which is a string metric for measuring the difference between two sequences defined as the minimum number of edit costs needed to transform a string into another with edit operations such as insertion, deletion, and substitution of a single character. Then, similarities between all combinations of any two sequential alarms in the plant operation data are evaluated using the levenshtein distance. The number of sequential alarms is effectively reduced by grouping them in accordance with the degree of similarities. The method was applied to the operation log data of the ethylene plant. Results showed our method is able to group a lot of sequential alarms into small number of groups in accordance with the degree of similarities.