Research On Hysteresis Compensation And Encrypted Classification Methods For Flexible Strain Sensor Signal

With the rising standard of living of the people,People's concerns have shifted from external material life to physical health,and smart wearable devices have emerged.Smart wearables are devices that combine sensors into a part of the human body to monitor physical conditions in real time,providing personalized,exclusive service to everyone.A large part of medical wearable devices are made up of polymer flexible strain sensors.Such sensors have large hysteresis during use,which affects the accuracy of the sensor and the accuracy of subsequent data processing.Aiming at the hysteresis nonlinear problem of the sensor,this paper combines a class of asymmetric model operators to establish the hysteresis inverse model of the flexible sensor.Based on this,the feedforward compensation of the sensor is studied.The experimental results show that the method can be applied to a flexible sensor based on carbon black/polyurethane composite material,which can compensate the hysteresis nonlinearity of the flexible sensor during long-term use.The root mean square error of the sensor output is reduced by 92.93%.Compared with the traditional construction model method,the root mean square error is reduced by 52.38%.In addition,based on the considerations of information security and user privacy,the method of encrypting the signals collected by the sensor is firstly studied,and the method of classification is implemented to realize the homomorphic encryption analysis processing of the collected information.The deep convolutional neural network is mainly used to classify the encrypted signals.Experiments show that the ECG signal is encrypted by the advanced encryption standard AES algorithm,and the 11-layer convolutional neural network is built to classify the encrypted ECG signals.Under the condition that the ECG signal samples are sufficient,the accuracy can be reached 88.7% or more.