13170

12018

كارشناسي ارشد

كنترل

اصفهان: دانشگاه صنعتي اصفهان، دانشكده برق و كامپيوتر

۱۳۹۶

سيزده، ۹۴ص.: مصور، جدول، نمودار

ايمان ايزدي، جعفر قيصري

احسان يزديان

جواد عسگري، محسن مجيري

1396/11/09

كتابنامه

برق و كامپيوتر

مهندسي برق و كامپيوتر

ID12018

Detecting Variance Change in Alarm Systems Using Square and Absulute Value Filters and Delay timer Zahra Sharifi Teshnizi z sharifi@ec iut ac ir December 24 2017 Department of Electrical and Computer Engineering Isfahan University of Technology Isfahan 84156 83111 Iran Degree M Sc Language Farsi Supervisors Iman Izadi and Jafar GhaisariAbstract Today alarm detection and management systems are an integral part of industrial systems and play an im portant role in good performance of industrial systems Alarm mechanism is widely used in modern monitoring systems An alarm system is a set of hardware and software that works on the basis of some standards for alarms to detect errorsin the various parts of an industrial unit Large industrial units are composed of many components that error may occur Hence identifying and responding to an error is a critical issue The main task of alarm systems is to inform the operatorof the unusual conditions in the process This abnormal state in industry is often determined by comparing the value of avariable with a predetermined optimal value during operation When the value of a variable exceeds its normal value theoperator must detect this state by an activated alarm and then solve the problem Typically the information of the variousunits of an industrial system is controlled centrally and through monitoring Ideally the operator should receive one andonly one alarm for any abnormal state This almost never happens With the complexity of industrial system the numberof alarms is increased and sometime making alarm floods It becomes almost impossible to detect the essential alarmsby the user Hence the design of an appropriate alarm system that can manage these alarms and respond only to essentialalarms is important Therefore in recent years the design and analysis of alarm systems have attracted the attention ofresearchers Some of the most important factors in measuring the performance of an alarm system are probability of falsealarm probability of missed alarm and detection delay Designing alarm systems such as deadband and delay timer willreduce false and missed alarms However these methods cause some delay in alarm In designing the alarm system atrade off between the performance and the latency of the alarms system should be considered In this thesis the method ofdata processing is used Filtering is one of the data processing methods used in the industry Two commonly used filtersin industry are moving average filter and exponentially weighted moving average filter These types of filters are usefulin detecting changes in the data s average while some faults appear to change the variance of data Nonlinear filters andgeneralized delay timer are used to detect these faults An algorithm to design a delay timer to achieve the value of theparameters to reach to a desirable alarm system is proposed Key Words Alarm systems Monitoring systems Alarms flood False alarm rate Missed alarm rate Filtering Generalized delay timer

ايمان ايزدي، جعفر قيصري

احسان يزديان

جواد عسگري، محسن مجيري