Granular Computing Based Neuro-Fuzzy System Modeling For Forecast Of Operator Functional State

The core problem of Operator Functional State (OFS) in human-machine systems is extracting knowledge from the collected operator electrophysiological signals and performance data. This paper adopts granular computing method for knowledge acquisition based on electrophysiological signals and the operator performance data, proposed for granular computing neuro-fuzzy system (GrC-NF) modeling method and applied to the problem of modeling and forecasting OFS. GrC-NF Methods and adaptive neuro-fuzzy inference system (ANFIS) approach on the issue of the OFS were compared. The results show that, GrC-NF method for improving the accuracy of the model also has better explanatory, its assessment is valid for the OFS. Then neuro-fuzzy system based multi-granularity level modeling results show that Multi-GrC-NF system modeling method has better generalization ability and robustness.