Research On Driving Fatigue Level Classification Based On EEG

Driving fatigue is one of the main causes of road traffic accidents.At present,both domestic and overseas researchers have conducted a lot of research on the causes and detection methods of driving fatigue,and have obtained a series of research results.In the research on the formation mechanism of driving fatigue,due to the close relationship between the driver's fatigue level and the cause of traffic accidents,it plays an important role in the further improvement of the reliability and real-time performance of the detection system.And the system can also take different warnings or active interventions for different levels of fatigue,so it has gradually attracted people's attention in recent years.However,since the assessment of the level of driving fatigue is greatly influenced by subjective factors,there is no quantitative unified grading standard.This paper has carried out research working on the detection and classification of fatigue state based on EEG signals.Based on the analysis of the changes of EEG features,the HMM model of driving fatigue level grading was constructed,and the model was verified based on the simulated driving experiment.The main research contents of this paper are as follows:The establishment of the driver's fatigue status database.In order to obtain EEG data under different fatigue conditions,this paper designed a complete experimental scheme based on the existing simulation driving platform and carried out the collection of relevant basic data.The experimental data was calibrated by the combination of driver's EEG data,facial video image and subjective scoring,thus the basic database used in this study was constructed.The analyzation of the change of EEG characteristic indicators and the optimization of indicators.Since a single characteristic indicator cannot respond to the driver's fatigue state with high accuracy at any time,this paper selects the energy index,fuzzy entropy and 90% edge spectrum value as the candidate feature indicators based on the existing research.By paired sample t-test of data samples before and after the occurrence of fatigue state,a corresponding set of dominant indicators is formed.The construction of the fatigue grading models based on HMM.In order to reduce the complexity of the model,the fuzzy C-means clustering algorithm is used to cluster the preferred indicators,and the optimal cluster number is used as the observation state of HMM.Based on the fatigue state database,the fatigue assessment model corresponding to different states is established.Determination of the result of fatigue level classification based on genetic algorithm.Considering the accuracy of model prediction and the real-time requirements in practical applications,the optimization objective function is constructed with accuracy and response time as elements.The multi-objective optimization problem was solved by NSGA-II algorithm,and the best classification result of fatigue level is determined.Focusing on the key issues in the study of driving fatigue,the research on the determination of EEG characteristic indicators and the quantitative of fatigue levels has been carried out.The difficulty of breaking through is as follows: The optimal index of fatigue evaluation is determined as the energy index and the fuzzy entropy index by analyzing the regularity of the candidate feature index and combining the paired sample t-test method;Furthermore,considering the model prediction accuracy and time efficiency,the best classification result of fatigue level is determined based on genetic algorithm.