Today, alarm detection and management systems are an integral part of industrial systems and play an important 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 errors in 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 operator of the unusual conditions in the process. This abnormal state in industry is often determined by comparing the value of a variable with a predetermined optimal value during operation. W hen the value of a variable exceeds its normal value, the operator must detect this state by an activated alarm and then solve the problem. Typically, the information of the various units of an industrial system is controlled centrally and through monitoring. Ideally, the operator should receive one and only one alarm for any abnormal state. This almost never happens. With the complexity of industrial system, the number of alarms is increased and, sometime, making alarm floods. It becomes almost impossible to detect the essential alarms by the use r. Hence, the design of an appropriate alarm system that can manage these alarms and respond only to essential alarms is important. Therefore, in recent years, the design and analysis of alarm systems have attracted the attention of researchers. S ome of the most important factors in measuring the performance of an alarm system are probability of false alarm, probability of missed alarm and detection delay. Designing alarm systems such as deadband and delay timer will reduce false and missed alarms. However, these methods cause some delay in alarm. In designing the alarm system, a trade- off between the performance and the latency of the alarms system should be considered. In this thesis, the method of data processing is used. Filtering is one of the data processing methods used in the industry. Two commonly used filters in industry are moving average filter and exponentially weighted moving average filter. These types of filters are useful in detecting changes in the data’s average, while some faults appear to change the variance of data. Nonlinear filters and generalized delay timer are used to detect these faults. An algorithm to design a delay timer to achieve the value of the parameters 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.
