Comparison of the Smith-Waterman and Needleman-Wunsch algorithms for online similarity analysis of industrial alarm floods

Alarm floods are considered to be the major obstacles that prevent the smooth process operations of large-scale industrial facilities. During an alarm flood situation, industrial operators often get confused by too many alarms and thus have difficulties in observing and handling critical alarms. In recent years, sequence alignment based similarity analysis has emerged as an effective way to handle alarm floods. Alarm floods caused by the same fault are very likely to consist of the same group of alarms in a certain sequential order. Conducting realtime sequence alignment of industrial alarm floods can help operators to quickly recall the root cause and make prompt corrective actions. This paper proposes the online similarity analysis of alarm floods based on the Smith-Waterman and Needleman-Wunsch algorithms, and compares their differences and application conditions. Case studies are provided to illustrate the proposed online similarity analysis methods and the differences of the two sequence alignment algorithms.