A Neural Network Model for Human Workload Simulation in Complex Human-Machine System
Abstract
The overall goal of this study is to develop neural network models for analysis of electroencephalogram (EEG) data and use the results obtained to classify the level of mental workload experienced by humans during task processing. The study uses EEG data on piloting tasks from the STORM (Simulator for Tactical Operations Research and Measurement) experiments performed at the Cognitive Assessment Laboratory of the Human Effectiveness Directorate at Wright-Patterson AFB. Comparisons of classical backpropagation neural networks (CBNN) and resilient backpropagation neural networks (RBNN) were conducted. The RBNN performed 50% faster in deriving a model for cognitive load with a marginal decrease in classification accuracy over the CBNN. The results indicate that the neural network model can successfully classify mental workload states at an average rate of 83%. The results obtained indicate that neural network models can be used to automate the classification of human mental workload based on EEG signal data.
Document Details
- Document Type
- Technical Report
- Publication Date
- Dec 01, 1999
- Accession Number
- ADA387791
Entities
People
- Celestine A. Ntuen
- Robert Li
Organizations
- North Carolina Agricultural and Technical State University