摘要
基于深度学习的神经网络,对步态识别算法进行了优化研究。利用粒子群优化BP神经网络阈值、权值,在神经网络中代入优化后的初始值进行训练,避免陷入局部最优。通过Vicon MX系统对角度特征值进行采集,利用基于粒子群优化BP神经网络进行识别,验证其识别步态的可行性;筛选出传感器系统特征值,在对其优化改进时选取粒子群优化BP神经网络。与传统神经网络法、粒子群优化法相比,基于粒子群优化BP神经网络法的识别方式,识别时间短且识别率高。
Based on the deep learning neural network,this paper studies the optimization of gait recognition algorithm.The threshold and weight of BP neural network are optimized by particle swarm optimization,and the initial value after optimization is substituted into the neural network for training to avoid falling into local optimization. The angle eigenvalues are collected by vicon MX system and identified by BP neural network based on particle swarm optimization to verify the feasibility of gait recognition;the eigenvalues of sensor system are screened out and the BP neural network based on particle swarm optimization is selected for optimization and improvement of sensor system.Compared with the traditional neural network method and particle swarm optimization method,the recognition method of BP neural network based on particle swarm optimization is characterized by short recognition time and high recognition rate.
作者
张馨心
姚爱琴
孙运强
白桂峰
郭瑞琦
ZHANG Xin-xin;YAO Ai-qin;SUN Yun-qiang;BAI Gui-feng;GUO Rui-qi(School of Communication and Information Engineering,North University of China,Taiyuan 030051,China)
出处
《自动化与仪表》
2020年第4期70-74,共5页
Automation & Instrumentation
基金
山西省研究生教育创新项目(2019BY107)
国家自然基金项目(61774137)。
关键词
步态识别
BP神经网络
粒子群
深度学习
识别率
gait recognition
BP neural network
particle swarm optimization
deep learning
recognition rate
