Workload Analysis Of High Speed Railway Dispatchers Based On Eeg

The work load of high-speed railway dispatcher is an important factor affecting the safety of high-speed railway operation.If the high-speed railway traffic dispatcher can accurately identify whether he is under excessive work load and give a warning or prompt,the railway traffic safety accident caused by the dispatcher’s error can be avoided to a great extent.It is a key technical problem to accurately identify the workload of high-speed railway traffic dispatcher to improve work efficiency and ensure the safety of train operation.Aiming at this problem,this paper designs a method to identify the workload of high-speed railway dispatchers according to EEG signals.Combined with the work content and environment of high-speed railway dispatchers,through the background information of high-speed railway dispatchers’ working mechanism and working mode,it is concluded that the main influencing factors of high-speed railway dispatchers’ workload are work difficulty and unit time workload.In this paper,a high-speed railway dispatcher workload classification experiment is designed.The workload is taken as the independent variable of this study,and the workload level is controlled by work difficulty and unit time workload.Subjective scale method and main task performance method were used to measure the subjective and objective values of workload.Design the experiment according to the required index parameters,design the experimental task according to the basic work instructions of the high-speed rail dispatcher,collect the required data,and extract and process the EEG data and workload measurement data.The EEG signals were preprocessed by Butterworth filtering and independent component analysis.Perform time-frequency conversion on EEG signals and extract frequency domain features closely related to workload,includingδ(0.5～4Hz),θ(4～8Hz),α(8～13Hz),β(13～30hz)class 4;In addition,three composite indicators are introducedα/β、(α+θ)/(β+θ)、(α+θ)/β,used to amplify the difference between data.Normal distribution test and sample paired t-test were used to verify the effectiveness of the experiment on workload classification.Using Kruskal Wallis nonparametric test and Pearson correlation analysis,59 EEG indicators with significant differences in different levels of workload were obtained.The BP neural network model commonly used in the field of EEG pattern recognition is used to train the collected EEG signal related indicators and workload level,so that the model can judge the workload level of the subjects,and obtain the corresponding accuracy,sensitivity and specificity values,so as to verify the impact of different indicators on workload level recognition.The results show that,αand(α+θ)/β frequency band characteristic index has a good effect in workload identification;Frontal area is more sensitive to the workload identification of high-speed rail dispatchers,which proves that the main workload of high-speed rail dispatchers comes from mental cognitive channels;The principal component analysis method is used to fuse and reduce the dimension of all indicators,which can eliminate redundant information and improve the recognition accuracy;In addition,by changing the activation function into radial basis function,the accuracy and operation speed of the model can be improved.