| Video-game mental fatigue has a great effect on the physical health of players and product's user experience.And EEG signal is the "gold standard" for mental fatigue evaluation,which can reflect the activities of the brain directly and objectively.For the lack of evaluation research on video-game mental fatigue,a series of evaluation indexes based on EEG signal is proposed with the feature extract methods of Wavelet Transform and Sample Entropy,along with subjective evaluation and performance evaluation.And the classification model for mental fatigue pattern recognition is made with excellent classification performance,based on Support Vector Machine(SVM).The specific content and findings include the following parts:(1)Designing the difficulty levels for video-game mission for fatigue experiment.In considerationof the effect of mental workload,the experiment uses the NASA-TLX scale and subjects'performance in the mission to analysis the differences in mental workload between 8 difficulty levels for video-game.And according to the result,the representative easy and hard missions are selected.(2)Building the Mental Fatigue Assessment Scale.Preliminary designed scale is based the conclusion of previous studies.After questionnaire survey and reliability and validity testing,11 items meet the requirement for the assessment scale.(3)Video-game fatigue experiment and data preprocessing.After set bad signals,part of EMG components and EOG components to zero,to obtain the clear EEG signals which are in the range of 0.5-32Hz,this thesis using two kinds of algorithms.First,using Butterworth Low-pass Filter to filter out the power frequency interference and most of the EMG signals whose frequency are higher than 32Hz.Then using Wavelet Transform to filter out the EOG signals,part of the EMG signals and background noise that mixed in EEG signal,and obtaining the clear signal in range of 0.5-32Hz.And the filter process is effective from analyzing the time-frequency graph of the signal.(4)Extraction of EEG sigal features and mental fatigue analysis.To differentiate the mental fatigue,the data of subjective assessment and task performance are analyzed.Then the EEG signal features under different mental fatigue statuses are studied considering the relative energy of rhythm wave,the ratio of energy and sample entropy.The results are obtained as following:? The mental fatigue of subjects keeps growing in the easy task,and there is a certain time excitement in the hard task,in which the fatigue accumulates later.? After analyzing the relative energy of four rhythm wave including Delta,Theta,Alpha and Beta,and the ratio of them including ?/??(?+?)(? and ?/(?+?),by Mean Value Analysis and ANOVA,along with observing the change of EEG topography,it is concluded that the energy of slow wave increase and the energy of fast wave decrease with the fatigue growing,especially in the change of Theta wave and Beta wave,and all three ratio parameters could used to evaluate video-game mental fatigue.? According to result of analyzing sample entropy by ANOVA,the sample entropy values decrease with the growing of fatigue,and sample entropy could become a feature index for video-game mental fatigue assessment.(4)Building classifier model based on SVM.This paper uses multidimensional feature indexes to build the two-class classifier for waking status and fatigue status and the four-class classifier for waking status,excited status,fatigue status and heavy fatigue status based on Support Vector Machine(SVM).The result shows that the two-class classifier has an excellent performance with classification accuracy above 80%,but the accuracy of the four-class classifier can only reach to 65%,which we need to improve in the future study.Study results of this paper provide theoretical basis and technical support for real-time monitoring video-game mental fatigue as well as complements for mental fatigue assessment theories,which has a significant application background. |
PDF Full Text Request