| In process industry,due to the widely use of distributed control system(DCS)and the rapid development of sensor technology,information technology and software technology,monitoring of process variables during manufacturing are becoming easier than before,since there is no limit to add a new monitoring variable in distributed control system.When designing the alarm system,rationality and effectiveness have not been considered thoroughly.Usually,threshold alarm is used,which increases the number of alarms.Meanwhile,there is a lot of coupling in process industry.It will cause an alarm when there is an alarm in the coupled node or related node.This will probably cause alarm flood,which is a very difficult problem for operators to deal with.In order to deal with alarm flooding,the correlation between different alarm flood sequences can be analyzed by alarm flood sequence alignment.Proper procedures can be given in time in the early stage of alarm flood to prevent it from causing trouble to the normal operating.Alarm flood sequence alignment is studied by considering algorithms proposed to deal sequence alignment in other research fields.A modified algorithm based on Levenshtein distance and BLAST is proposed to align alarm flood sequences.Two alarm sequences are aligned by seeding,matching and extending.Meanwhile,Levenshtein distance is used to set a boundary for extending.The accuracy and reliability of the proposed algorithm are tested by align several different alarm flood sequences generated by quadruple tank experiment and Tennessee Eastman process experiment. |
