Contextual information-based human fatigue prediction for integrated traffic control

Air traffic controllers are often confronted with a high possibility of human fatigue due to high mental workload from processing the intensive amount of information and making timely decisions, especially under dynamic working conditions such as heavy traffic density and adverse weather conditions. This study proposes a hybrid approach to make a proactive prediction of human fatigue based on contextual information of the dynamic working conditions. The proposed hybrid approach which combines the Artificial Immune System (AIS) and the eXtreme Gradient Boosting (XGBoost) technique has several innovations. Firstly, it can take into consideration causal factors, such as work-related factors and weather-related factors, and adapt to dynamic traffic conditions. Secondly, it attempts to analyse individual differences and utilize the results obtained to facilitate human fatigue prediction. Finally, Artificial Immune System is used to pre-process the causal factors of human fatigue and the Extreme Gradient Boosting technique is applied to adaptively predict states of human fatigue.