A Multi-Population Genetic Backpropagation Neural Network Approach for Perceived Workload Evaluation of Pilots

The perceived workload of pilots reflects their current state, and understanding the pilot's state is crucial for system design, accident prevention, and mission planning. Monitoring the perceived workload of pilots during missions is always challenging; however, previous studies demonstrate that objective data, such as eye movements and performance, are sensitive indicators of the perceived workload. Neural networks are effective tools for addressing such nonlinear and complex problems. In this study, we employ the multi-population genetic backpropagation neural network (MPGA-BP) algorithm to establish an efficient perceptual workload evaluation model. The model takes reaction time and gaze entropy rate as inputs and provides perceived workload values based on the NASA-TLX scale as outputs. The results show that the MPGA-BP model performs well, providing a reliable method for monitoring perceived workload.