13356

12169

كارشناسي ارشد

كنترل

اصفهان: دانشگاه صنعتي اصفهان، دانشكده برق و كامپيوتر

۱۳۹۶

ده، ۸۵ص.: مصور، جدول، نمودار

مريم ذكري، ايمان ايزدي

ناصر قديري مدرس

1396/12/19

كتابنامه

برق و كامپيوتر

مهندسي برق و كامپيوتر

ID12169

Analysis and Application of Sequences Pattern Mining Methodes in Alarm Flood Roya Moein r moein@ec iut ac ir January 10 2018 Department of Electrical and Computer Engineering Isfahan University of Technology Isfahan 84156 83111 Iran Degree M Sc Language Farsi Supervisor Dr Maryam Zekri mzekri@cc iut ac ir Dr Iman Izadi iman izadi@cc iut ac ir Abstract Nowadays the presence of many advanced sensors and actuators the availability of different variables in a process the increasing use of alarm systems to monitor the processes performance and thus improving these systems efficiencyhave become increasingly important due to the emergence of distributed control systems in industry The alarm systemalerts the operator when it faces with unusual conditions in the process or equipment It clearly results in reducing the timein which each industry is suspended increasing security improving operator performance and as a result it consolidatesprocess performance One of the other problems that an operator may have encountered is alarm flood According to theexisting standards of alarm system each operator should not alert more than 10 times in every ten minutes It could causeconfusion and as a result the operator will delay and eventually may cause an inevitable catastrophe This study aims atdiscussing the present problems in alarm system and in particular the huge number of alarms and its possible solutions Among the existing methods shortcomings in the area of alarm flood some are prominent including being time consumingand costly non susceptible to interferences and oscillations lack of analyzing and comparing the floods in pairs and processinformation Therefore this paper has made an attempt to use actual industrial alarm data along with extracting informationfrom them creating unique alarm and processing them in order to determine the time of alarm floods and their sequenceand provide a procedure for preprocessing industrial alarm Then the given example will be solved manually consideringdata mining process and analysis and application a method called sequential pattern mining Finally these algorithms areapplied automatically in data mining among the real industrial databases and their alarm flood sequences and examine theirresponses We also will study the disorder effects on these algorithms Key Words alarm system alarm flood data mining sequential pattern mining

مريم ذكري، ايمان ايزدي

ناصر قديري مدرس