Sequential pattern recognition employing recurrent fuzzy systems

Sequential pattern-recognition systems check whether data strings, e.g., time series, exhibit certain pattern primitives in a specified order. As in the case of most other pattern-recognition methods, either conventional methods or fuzzy systems may be used here. This paper presents a sequential pattern-recognition system employing recurrent fuzzy systems that is employed as a monitoring system on continuous-casting systems in the steel industry worldwide. Taking that application as a starting point, a general method for sequential pattern recognition in time series that uses recurrent fuzzy systems is described.