Design Of Flexible Nanometer Heart Sound Sensor Based On PVDF Copolymer Film And Study Of Heart Sound Signal Classification Algorithm

Heart sound is a kind of physiological sound produced during the process of systolic and diastolic.It contains a variety of physiological and pathological information and is a reflection of the mechanical movement of the heart and blood vessels.This paper proposes a design scheme of heart sound recognition system,which includes the methods of sensing,collecting,denoising and machine classification of heart sound signals,which provides a new diagnostic method for heart sound auscultation of patients with suspected heart disease.First of all,a nanoflexible piezoelectric thin film heart sound sensor design solution based on polyvinylidene fluoride(PVDF)-zinc oxide(Zn O)copolymer is proposed to address the disadvantages of traditional heart sound auscultation.The authors set up a high-voltage electrostatic spinning experimental platform.Based on this experimental platform,the authors prepared and compared the piezoelectric characteristics of three kinds of flexible nano-piezoelectric copolymer films of pure PVDF,PVDF/Zn O and PVDF/graphene(GR).Observed by scanning electron microscope,the fiber orientation of the prepared piezoelectric film is relatively consistent,and the film has good mechanical flexibility.Secondly,a flexible heart sound sensor is fabricated,which is a "sandwich" type packaging protection for PVDF/Zn O copolymer piezoelectric films with good piezoelectricity.The author designed and implemented a conditioning circuit for a PVDF / Zn O copolymer flexible heart sound sensor.The circuit includes a charge amplifier circuit,a voltage amplifier circuit,and a bias circuit.The conditioning circuit amplifies the weak heart sound signal obtained by the flexible heart sound sensor and sends it to the USB-6008 data acquisition card.The author programed the acquisition,filtering and storage of heart sound signals on the Lab VIEW platform.Based on the theoretical basis of wavelet transform,the collected heart sound signals were subjected to threshold wavelet denoising processing on the MATLAB platform.Daubechies6(abbreviated as db6)wavelet was selected as the base wavelet,and the number of decomposition layers selected was 5 layers.Experimental results show that this method has a good noise reduction effect.Finally,this paper proposes a heart sound classification algorithm based on a feedforward neural network.This algorithm screens the quality of heart sound signals collected after noise reduction,and classifies normal and abnormal heart sound signals.The algorithm extracts a total of 18 feature variables and uses cross entropy as a loss function for training.The training set consists of 3,580 heart sound data from the Physio Net open source heart sound database.Randomly select 480 cases of heart sound data in the heart sound database and 20 cases of heart sound data measured by the new PVDF copolymer flexible nano heart sound sensor to form a verification set for classification test.The test results obtained the average sensitivity of the heart sound classification algorithm proposed in this paper is 83.09%,with an average The specificity was 83.27%.The experiment proves that the system proposed in this paper has certain feasibility and advancedness,and the heart sound recognition system with automatic machine classification has certain application prospects.