基于新型柔性触觉传感器的摩尔斯电码分类识别研究OA北大核心CSTPCD

Tactile perception is an important way for people to connect with the outside world.it is important to give the robot touch per-ception ability in the human-machine interaction system.Based on the piezoelectric characteristics of polyvinylidene fluoride(PVDF),a new piezoelectric flexible touch sensor with high sensitivity(15.02 mV/N)is designed and prepared by using 3D printing technology.By tapping the Morse code of the"0-9"number on the surface of the sensor,the 4 000-dimensional voltage time series signals from 3 000 sets of sensors are collected as the experimental sample set.On this basis,a fusion network model of Convolutional Neural Net-work(CNN)and Gated Recurrent Unit(GRU)with strong feature extraction ability is constructed,which is used to recognize the Morse code of the number"0～9",and the average recognition accuracy is 98.17%.The results show that the new flexible tactile sensor studied can accurately perceive and distinguish the knock modes of different Morse codes,CNN-GRU can be well applied to the classification and recognition of contact modes of flexible tactile sensors because of its time series memory ability.