Pilots' Multiple Physiological Signals Based Workload Research

Recently,the research on aviation human factor has been popular.However,it is hard to find the effective method to avoid the flight accidents caused by human factor.After analyzing these flight accidents,we found most of them were caused by the high level of pilots' workload.So obtaining a reliable method to evaluate pilots' workload has become a difficulty.Therefore,pilots' workload has been a part in airworthiness certification.In the recent decades,researchers have paid more attention to multiple physiological signals to evaluate the pilots' workload.According to the requirement of China Civil Aviation Regulations,this topic proposes a method to rate the pilots' workload under different flight tasks,which is based on the different workload factors.This method attempts to provide some novel ideas and approaches for airworthiness certification of Chinese aircrafts.This topic analyze the workload of flight crews in CCAR firstly,and obtain the relationship between the workload factors and physiological signals.Then,according to the characters of physiological signals and the experiment condition,this topic chooses heart rate,respiratory rate and pupil diameter.At the same time,we experiment in the CRJ-200 cockpit simulator to do the four flight tasks with five crews and pre-process the obtained multiple physiological signals using different ways.Analyze multiple physiological signals in time and frequency domains and extract features.Using sequential floating forward selection method,10 features such as time domain features of heart rate and respiratory rate signals,frequency and entropy features of pupil diameter are chose to combine with Bedford value as data set.After that,the proposed evaluation method based on improved affinity propagation is used to classify the 10-dimensional data set.Compared with former method,the proposed method obtains precise pilot workload evaluation.Finally,this method is applied to analyze the workload trends of the 4 different tasks and the trends are discussed.