Study On Risk Classification Of Driving Fatigue

The research data show that nearly 20% of traffic accidents have a direct or indirect relationship with the drivers,and the proportion of the heavy traffic accidents caused by fatigue driving is up to 40%.With the development of automobile intelligent technology,driver fatigue detection has become one of the hot topics in the field of vehicle engineering.At present,most of the warning strategies used in the safety auxiliary driving system are largely determined by the judgment of the fatigue state of the driver,and the classification of the fatigue level is rarely involved.However,reasonable fatigue level plays an important role in improving the real-time performance of the monitoring system and predicting the dangerous situation.Based on the analysis of the related factors of driving fatigue and the related theory of hidden Markov,this paper carries out relevant research work on the method of driving fatigue level classification.The main research contents are as follows:(1)In most of the previous studies,the parameters of eye opening and closing PERCLOS,eye blink speed AECS and the mouth opening parameter PERLVO are usually clustered Individually by clustering methods,then it is input into the driver fatigue assessment model to judge the fatigue state of the driver.Research and analysis found that the observed data of driver fatigue characteristics are consistent,so the whole clustering method is adopted.Therefore,this paper adopts an overall clustering method to analyze the above indicators based on the FCM algorithm,and uses the F statistic to determine the optimal number of clustering observations to simplify the data processing process and model construction complexity.(2)EEG is considered to be the most effective indicator of driver fatigue.In EEG,there are many waveforms that can reflect the fatigue state of drivers.Based on the detailed analysis of related parameters,this paper constructs a hybrid statistical feature index parameter T to evaluate the driver's fatigue status.On this basis,the FCM clustering method was used to determine the number of driver fatigue states suitable for this study.(3)In order to verify the rationality and accuracy of classifying driver fatigue status,a three-state driver fatigue assessment model based on HMM and a two-state driver fatigue assessment model based on HMM were established respectively,and then evaluated and analyzed.The results show that the three-state driver fatigue assessment model has a higher recognition accuracy and is also higher than the two-state driver fatigue assessment model,thus verifying the rationality and accuracy of classifying the driver fatigue state into three categories.(4)Based on the research objectives of the project,in light of the proposed driver fatigue characterization indicator system structure,based on the detailed formulation and analysis of the experimental program,the relevant feature parameters were collected using the eye tracker,EEG and other related experimental equipment.Combined with the experimental data,the fatigue fatigue identification model was verified and analyzed.