Research On Multi-modal Epidermal Electronic Sensor For Mental Fatigue Monitoring

Mental fatigue,characterized by subjective feelings of “tiredness” and “lack of energy”,will lead to reduced work efficiency or even endanger life and property.Therefore,monitoring and early warning of mental fatigue is of great significance.According to the fact that various physiological signals of the human body change regularly after the occurrence of mental fatigue,the mental fatigue detection method based on multi-modal physiological signals shows a high detection accuracy.However,the commonly used physiological signal detection equipment is bulky,low in comfort,and may cause skin allergy,which is not suitable for long-term wear.The characteristics of the ultra-thin,low-modulus,light-weight,and conformal contact with the skin of the epidermal electronics systems are very in line with the requirements for comfortable monitoring mental fatigue.However,the current epidermal electronic devices used for physiological signal detection have the disadvantages of limited sensitivity and complicated preparation process.In this thesis,in order to achieve comfortable and non-invasive monitoring of mental fatigue,we starts from theoretical research and process improvement,respectively researches high-sensitivity strain sensors that can collect respiratory signals and epidermal electrodes that collect bioelectric signals,and then prepares multi-modal epidermal electronic sensors that can detect ECG,respiration,and skin conductance signals at the same time,and realize the acquisition and feature extraction of three types of signals.Finally,machine learning algorithms are used to train a prediction model based on the characteristics of multi-modal physiological signals to monitor mental fatigue.The main contents include:Firstly,a method of designing a highly sensitive strain sensor by realizing the proximity of the conductive compound to the percolation threshold point is proposed,and a strain sensor with ultra-high sensitivity is prepared.Based on the percolation conductivity theory,a hypothesis that the conductive compound with a filling concentration at the percolation threshold point will be highly sensitive to strain was proposed.Graphite was selected as the conductive material and sandpaper was used as the flexible substrate.The graphite was rubbed on the sandpaper to realize the filling of the conductive particles into the polymer substrate,and the resistance change during the filling process was monitored in real time.We proved the approach of the obtained graphite layer to the percolation threshold by the statistical percolation theory and the two-dimensional conductive path percolation model.The graphite layer with resistance close to the threshold of percolation was successfully prepared,and the sensor with ultra-high sensitivity was prepared with a strain factor of up to 9720.We then successfully applied this sensor to detect human respiratory frequency.Secondly,a preparation method of epidermal electronic equipment based on laser patterning was developed,and a flexible epidermal electrode capable of accurately detecting ECG signals was prepared.Taking advantage of the high-precision characteristics of laser,we developed a manufacturing process based on laser patterning.We used water transfer paper as the substrate,and achieve high-precision patterning of the metal thin film layer on the polymer substrate.A non-closed winding wire mesh structured skin electrode was designed,which gives the electrodes excellent stretchability and low contact impedance characteristics with the skin.Then we achieved high-precision acquisition of ECG signals based on flexible skin electrodes and completed signal noise filter.Thirdly,a mental fatigue monitoring system based on multi-modal epidermal electronic sensors was constructed.With the help of the epidermal electronic system platform,a multimodal epidermal electronic sensor that can simultaneously monitor ECG signals,respiratory signals,and skin conductance signals was prepared,and with the help of machine learning algorithms,multimodal information fusion was achieved,and a high-precision prediction model was trained to achieve mental fatigue testing.Based on the attachment transfer process,the ECG sensor and the breathing sensor are integrated into a module,which is attached to the chest to realize the simultaneous detection of the ECG and breathing signals;according to the specific distribution of the human sweat glands,a skin conductance sensor that can be attached to the palm is prepared as another module to achieve effective detection of skin conductance signals.Then we design a mental fatigue experiment to collect the physiological signals and corresponding fatigue labels of the testees under different mental fatigue states.A variety of machine learning algorithms are used to train the model and evaluate the accuracy of the model,and an algorithm with high accuracy was selected as the better algorithm for studying mental fatigue monitoring based on physiological signal characteristics.Among them,the accuracy rate of the model trained by the decision tree algorithm can reach up to 89%.The high-accuracy prediction model was applied to the prediction of mental fatigue in actual scenes,including exploring the mental fatigue process of different daily brain activities and different relaxation methods to relieve fatigue.