Application Research Of Nano-gold Flexible Sensor In Gesture Classification

With the work rapidly progressed in developing forms of artificial intelligence(AI),biomedical engineering with pattern recognition technology is widely used in many fields such as human-computer interaction,disease diagnosis,and prosthetic control.Among them,the development of functionally intelligent prosthetic limbs makes the classification of gestures and recognition become one of the hot spots.So far,decoding surface electromyogram(sEMG)has become the main research method of gesture recognition.However,sEMG is a weak physiological electrical signal,which is susceptible to electromagnetic interference and muscle fatigue.Generally traditional EMG acquisition equipment mostly uses metal electrodes.It's usually necessary to reduce the gap between the electrode and the skin by applying the conductive paste.It's not suitable for people who are allergic to conductive paste and easy to cause patient discomfort,and not convenient for long-term monitoring of electrophysiological signals.Gestures need to be coordinated between muscle groups,and the muscle contraction level of each gesture is different.For the shortcomings of sEMG,the paper chooses a nano-gold flexible sensor.Apply its excellent electrical conductivity in the stretching process to the classification of gestures.The nano-gold flexible sensor uses the difference of the forearm muscle contraction level to apply the deformation signal of the sensor after stretching to the gesture classification and have good classification performance,which provide a new technical method for prosthetic limb control applications based on physiological signals.The research content of the paper mainly includes the following aspects:(1)A preliminary study is made on the nano-gold flexible sensor and its acquisition platform.The study designs and prepares a nano-gold flexible sensor,and discusses its different packaging forms.According to the frequency of daily gesture actions,selected eight as the target gestures,and explored the signal data processing method of the deformed signal,selected the optimal characteristics and the classifier,which lays a foundation for the classification of subsequent gestures.(2)The effects of classification performance of gestures by the position of the nanogold flexible sensor and the posture of the arm were experimentally studied,respectively.The position of the nano-gold flexible sensor on the forearm was clarified,and analyzed its classification performance.Other experiment verified the two arm posture states of the arm's natural vertical and arm bending 90 °,and showed that there was no significant difference.(3)For the deformation signal,the effects of two different training modes,random training mode and repetitive training mode,on the classification accuracy of gestures were studied,respectively.The experiment compares the difference between the two training modes in terms of time-domain waveform,classification accuracy of each gesture and the average classification accuracy of each subject,and it is clear that the two gesture movement training modes are not significant difference.(4)Based on decoding sEMG signals,which is one of the common methods of gesture classification.Designed the experiments to collect sEMG and nano-gold flexible sensor deformation signals at the same time,and classify gestures from time domain waveforms,gesture accuracy of the same channel number,gestures of different channel numbers.Comparative analysis of gesture classification accuracy and feature vector scatter plots.Through experiments,it was further verified that the nano-gold flexible sensor,like the electrodes of traditional EMG acquisition devices,can achieve reliable gesture recognition.(5)Based on the classification performance of the deformation signal in the gesture action classification,this paper selected flexible stretchable material and prepared as a nano-gold flexible sensor.For deformation signals,designed experiment verifies the classification performance of the gesture and different strengths.The classification performance of the nano-gold flexible sensor for different strengths improves the exploration of the influencing factors of the deformation signal in the classification of gestures.