A hierarchical recurrent neuro-fuzzy model for system identification Original Research Article

Neuro-fuzzy systems are by now well established in data analysis and system control. They are well suited for the development of interactive data analysis tools, which enable the extraction of rule-based knowledge from data and the introduction of a priori knowledge in the process of data analysis and system identification. Despite the successful application of feed-forward models in diverse areas, its recurrent variants are still rarely used. However, recurrent models are able to store information of prior system states internally and could be therefore more appropriate for the analysis of dynamic systems. In this paper a hierarchical recurrent neuro-fuzzy model is presented which was developed for application in time series prediction and analysis of dynamic systems. It has been implemented in a tool for the interactive design of hierarchical recurrent fuzzy systems.