Data-driven Alarm Flood Analysis Method Based On Improved Prefixspan

The alarm system monitors industrial plants in real-time to ensure safe operation.With the emergence of innovative new technologies in sensor,computer software and communication technology,advanced control system has been more and more widely used in modern factories.On the one hand,it reduces the cost of alarm design and configuration,but on the other hand,it increases the difficulty of alarm rationalization,resulting in the improper design of alarm system.Improper alarm system design give rise to a substantial increase in the frequency and intensity of alarm flood issue.When the alarm flood occurs,the operators lack adequate time to analyze and can't make an effective response to the alarm,which will cause or aggravate the anomalies and failures in the system,and even lead to disasters.Aiming at the problem of alarm flooding,this paper proposes a data-driven alarm flood analysis method based on improved prefixspan,and analyzes and solves the problem of alarm flood hierarchically.Firstly,for the chattering alarm,this paper combines the response time of the operator and the off-delay timer technology to effectively suppress the univariate chattering alarm.Then,select different ways to extract the alarm flood sequence according to the data in the alarm log.Finally,this paper proposes the improved prefixspan algorithm with tolerance to short-term order ambiguity to mining alarm flood patterns,which not only avoids the problem that sequence alignment algorithm ignores the secondary mode in the clustering process,but also solves the problem that the traditional PrefixSpan algorithm can't deal with the order confusion caused by strong correlation variables,and finds more practical alarm flood patterns.The effectiveness of the proposed method is tested in different cases of TE process.Finally,the experimental results show that the proposed method can significantly reduce the number of alarms and extract more practical alarm flood patterns.