Research On Human Body Pressure Signal Based On PVDF Flexible Piezoelectric Sensor

Respiration,pulses,voice and other human activities could generate different pressure signals on the surface of the human body,which contain a lot of information and have potential application values.Therefore,it is especially necessary to create a wearable device which acquires the signals effectively.The device could collect the signals and explore potential application values and application methods through reasonable analysis.In this paper,a poly(vinylidene fluoride)(PVDF)flexible piezoelectric sensor was constructed and made for collecting pressure signals of human hands(pulses,fingers and wrists),breathing signals,and voice vibration signals(human neck skin vibration signals when speaking).Data conversion,machine learning and other data processing methods were used to explore their application values.The main contents of this paper are as follows:1.A PVDF flexible piezoelectric sensor with sandwich structure was constructed,which PVDF was took as the sensitive material and polydimethylsiloxane(PDMS)was regarded as the flexible substrate material.The sensor has a safe human skin fitness,high softness,good response repeatability,good linear sensitivity in the 0.2-0.72 N pressure range,and long-term durability,which has no obvious drift in the output voltage during 10,000 reciprocating bends.2.The hand pressure signals from the pulses,fingers and wrists,and the pressure signals from breathing and blowing were explored.The pulse signal includes the vital signs information of the superposition of the systolic pressure of the heart and the reflection pressure generated by the palm of the hand,the reflection pressure generated by the systolic pressure and the body parts below the hand,and the diastolic ventricular pressure were extracted.The pressure signals of the bent fingers based on subjective evaluation index for the subjects were detected,and we selected its distinctive characteristic values.Regular wrist pressure signals derived from dynamic gestures were detected.Breathing signals of subjects were detected to realize real-time monitoring for human breathing frequency.The blowing pressure signals based on Morse code were collected and the decoder were compiled to realize the man-machine communication.3.The PVDF flexible piezoelectric sensor was used to collect the voice vibration signals of26 English letters from 6 subjects.Then,this work analyzed the waveform features of the voice vibration signals.We removed the noises of the voice vibration signals,extracted the timefrequency domain features and nonlinear features of the voice vibration signals,and used the Grid Search-Support Vector Machine(GS-SVM)to perform pattern recognition on the features of the voice vibration signals for 26 letters.When the ratio between the number of training samples and that of testing samples was 50:100,the recognition accuracy rates for 6subjects were 94.62%,89.23%,89.50%,90.12%,89.65%,and 90.19%,respectively.In order to realize more information recognition based on the voice vibration signals,voice vibration signals of the letters were arranged and combined into voice vibration signals for the words.The principal component analysis(PCA)method was introduced for feature mining,and the recognition accuracy rates of the 6 subjects were 100%,99.75%,100%,100%,100% and99.75%,separately.The voice vibration signals of all English letters and words for the 6subjects were clustered,the corresponding accuracy rates of voice vibration signals were87.26% and 100% respectively.