Research On Fatigue Detection System Based On ECG

With the development of society,driving has become the main mode of transportation for more and more people,followed by a sharp rise in the number of traffic safety accidents.According to the investigation,in the occurrence of traffic safety accidents,the drivers fatigue driving is one of the very important factors.Therefore,a system that can effectively detect whether the driver is in fatigue driving is of great significance to alleviate the occurrence of traffic accidents.A fatigue detection system based on ECG is developed in this paper.The ECG detection device is embedded in the steering wheel,and the ECG signal of the driver is collected through the drivers palm.This non-invasive collection method will not cause discomfort to the driver,and it is more scientific,convenient and more acceptable to the driver.In addition,we also designed the corresponding fatigue detection algorithm to ensure the accuracy and robustness of the fatigue detection system.The algorithm includes: ECG signal extraction RR interval data preprocessing method;Denoising model based on convolutional neural network and generate adversarial neural network;Combined with fuzzy neural network and convolutional neural network detection model.Compared with the original ECG signal,the RRs feature extracted by our pretreatment method has a lower dimension and can well reflect the fatigue information of human body,which enables our fatigue detection system to better analyze and process to capture fatigue features.Our denoising model is based on the design of convolutional neural network and adversarial neural network,which can simulate the RRs of the palm to generate the corresponding RRs of the chest.The RRs of the chest cavity usually has a very low noise level,which solves the problem that the ECG signal obtained by the drivers palm usually has a lot of noise,and improves the accuracy and robustness of the fatigue detection system.The detection model is a fuzzy convolutional neural network proposed by combining fuzzy neural network and convolutional neural network.This network combines the advantages of translation invariance of convolutional neural network and fuzzy processing of fuzzy neural network,which can better process ECG and other time series signals,thus can have higher classification accuracy,and can make the results of fatigue detection system more accurateAt the end of this paper,a series of comparative experiments are set up.The experimental results prove the effectiveness and superiority of the denoising module and classification model of the fatigue detection system.Comparing our method with the best available methods,our fatigue detection model has better robustness and prediction accuracy.