12718

11642

كارشناسي ارشد

كنترل

اصفهان: دانشگاه صنعتي اصفهان، دانشكده برق و كامپيوتر

1396

يازده، 69ص.: مصور، جدول، نمودار(رنگي)

ص.ع. به فارسي و انگليسي

ايمان ايزدي

احسان يزديان

1396/05/31

كتابنامه

برق و كامپيوتر

مهندسي برق و كامپيوتر

ID11642

Analisys and Design of Moving Variance Filter For Alarm Systems Saeed Amiri Pebdani saeed amiri1@ec iut ac ir June 19 2017 Department of Electrical and Computer Engineering Isfahan University of Technology Isfahan 84156 83111 Iran Degree M Sc Language Farsi Supervisor Prof Iman Izadi iman izadi@cc iut ac ir Abstract Today with the advent of technology and the use of digital systems in industry almost all process variables are measuredby sensors With modernization of digital systems and sensors the use of alarm systems to monitor the performance ofprocesses in the industry was considered Concerning the abundance of variables and the definition of alarms for eachvariable the probability of weakness of the system of alarms is significant and hence the improvement of the efficiencyof alarm systems is important The task of the alarm system is alerting the occurrence of any fault in the process In thissituation when the value of a variable exceeds the threshold defined for it the alarms are triggered visually or audibly to theoperator and should be taken to identify and correct the alarms According to the standard the operator must receive up to 6alarms per hour but due to environmental factors incorrect design of alarm systems and sensor wear the operator receivesmore alarms per hour It is desirable to receive one and only one alarm for each fault but there are usually many alarms for thesame fault that are known as nuisance alarms The presence of false and nuisance alarms makes the operation of the alarmssystem difficult The most important criteria for assessing the performance of alarm systems are the false alarm rate themissed alarm rate and the delay in detecting alarms Methods that have a significant impact on improving the performanceof alarms include multivariate monitoring threshold level design model based monitoring and data processing In dataprocessing methods the use of an efficient filter improves the quality of alarms Commonly used filters include movingaverage and moving variance filters In this thesis the effect of these filters on the performance criteria of alarm systems arestudied Then by using quantitative methods and Mont Carlo simulation the optimal parameters for a moving variance filterare calculated and designed based on a compromise between delays in detection and optimal performance of the system Key Words Alarm Systems Alarms Assessment Criteria linear Filters Nonlinear Filters MovingAverage Filters Moving Variance Filters

ايمان ايزدي

احسان يزديان