Research On Human Mental Fatigue Detection Method Based On Pulse Wave Multi Feature Fusion

The high rhythm of the new era,the fierce competition in the workplace,the gradual increase of life pressure,and the shorter and shorter rest time,all of which frequently cause mental fatigue in the human body.Mental fatigue not only leads to a decrease in personal work efficiency,but also may cause safety accidents,and even induce various chronic diseases,mental depression,and even cancer.More dangerously,people tend to ignore their own state of fatigue.Traditional medical detection equipment based on bioelectrical signals(EEG,EEG,ECG,EMG)can help people recognize,predict,and intervene in potential mental fatigue states.However,the acquisition devices of these devices are complex and bulky.,is not conducive to the promotion and application.In order to research and develop a real-time and effective portable mental fatigue detection device,this paper explores a new fatigue detection method based on photoplethysmography(PPG)signal.The main tasks are as follows:Firstly,a PPG acquisition device based on STM32F103 is built,which includes a pulse wave acquisition terminal composed of SFH7050 photoelectric sensor and AFE4404 analog front-end.Taking college students as the test objects,the pulse waves of the subjects in three different time periods a day are collected.Data;take these data as the original big data for data preprocessing and fatigue state pre-analysis.The preprocessing proves that the evaluation value obtained by feature detection with a single pulse wave is less accurate.In order to overcome this problem,this paper then adopts the multi-feature fusion method to detect human mental fatigue.The multi-dimensional pulse wave features are extracted from the time domain and frequency domain of the preprocessed pulse wave signal by using the time-frequency method.In order to reduce the dimensional disaster caused by multi-dimensional feature processing,an improved random forest algorithm based on bat algorithm(BA-RFC)is proposed for feature selection and fatigue detection.By optimally selecting parameters such as the number of decision trees L and the split feature parameter m,the pulse wave features with higher fatigue level representation ability are screened.Using the filtered features for fatigue recognition not only optimizes the parameters,but also reduces the dimensional disaster,improves the fatigue detection accuracy,and increases the recognition accuracy to96.67%.In order to verify the accuracy of the BA-RFC algorithm,this paper conducts comparative experiments with other three classification algorithms(Support Vector Machine,BP Neural Network,KNN Nearest Neighbor).The experimental results show that among the four comparison schemes,the BA-RFC algorithm proposed in this paper has the highest fatigue detection accuracy.Finally,this paper develops a mobile APP with a smartphone as the platform and designs a practical and portable human mental fatigue detection system with the PPG hardware as the signal acquisition device.In the system,the local data management module is used to analyze and process the collected PPG signal data and perform fatigue detection.The preprocessed pulse wave signal and fatigue detection results are displayed in real time on the mobile APP.