Pilot Fatigue State Detection Based On EEG

In recent years,due to the accelerating pace of people’s life,the aviation industry has been developing rapidly,and the pressure faced by pilots while driving in aviation is huge,which will cause significant casualties in case of accidents.If the pilot’s mental condition can be found in time,the occurrence of tragedy will be reduced,so the assessment of the pilot’s fatigue status during driving is of great significance.The EEG signal is traditionally considered as one of the "gold standards" for fatigue detection,which has the advantages of accuracy and objectivity.In this paper,we analyze the fatigue state of pilots based on the EEG signal detection method,and the main contents are as follows:EEG signals are complex signals composed of multiple signals.Improper processing will seriously affect the subsequent model establishment and result analysis.To improve the accuracy of pilot fatigue driving detection,The experiment is based on pilot EEG signals collected through the A320 simulation cockpit.In this paper,FIR band-pass filtering and independent component analysis are used to eliminate the artifacts such as EOG signals and industrial frequency interference in EEG signals,and waveforms in different frequency bands of EEG signals are extracted by wavelet transform.At the same time,the time domain,frequency domain,time-frequency domain and nonlinear domain are combined to extract features of many aspects of EEG signals.Considering the problems of large EEG signal data set and many features,in order to improve the efficiency of fatigue driving detection,a feature dimension reduction method is proposed in this paper,which is transformed into a low-dimensional feature selection problem by evaluating and selecting the solution quality through the feature dimension reduction method in the initialization stage of the artificial bee colony algorithm and keeping the generated features with the highest probability for further optimization.The results show that compared with the traditional artificial bee colony algorithm,the number of selected features is reduced by 6,reducing by nearly 50%,the average time index is reduced by 32.5%,and the final classification accuracy is increased by 2.0%.The proposed algorithm gets better results both in classification accuracy and optimization efficiency,indicating that the proposed method is useful in the detection of tired driving.