| 袁娜,刘沛.基于PLS特征筛选和改进结构PNN的步态识别[J].唐山学院学报,2016,29(3):47-52 |
| 基于PLS特征筛选和改进结构PNN的步态识别 |
| Gait Recognition Based on Partial Least Squares Method and Modified Probabilistic Neural Network |
| |
| DOI：10.16160/j.cnki.tsxyxb.2016.03.014 |
| 中文关键词: 步态识别 偏最小二乘法 概率神经网络 先验变量 |
| 英文关键词: gait recognition partial least squares method probabilistic neural network priori variable |
| 基金项目: |
|
| 摘要点击次数: 7956 |
| 全文下载次数: 6680 |
| 中文摘要: |
| 采集单侧大腿截肢患者髋关节有代表性的加速度信号(双轴)和角速度信号,并同时采集足底压力信号,进行多运动模式识别,经信号预处理提取出特征参数,利用偏最小二乘法(PLS)进行特征筛选,最后利用改进的概率神经网络(PNN)步态模式识别和分类,实现对下肢假肢在行走、上下坡、上下台阶的不同运动模式的有效识别。 |
| 英文摘要: |
| The authors of this paper propose a method of identifying the different motion patterns of artificial legs walking on the level ground, down or up stairs and slopes by first collecting the typical acceleration signal(dual-axis) and the angular velocity signal of the hip joint of the patients with one leg amputated, and the plantar pressure signal to recognize the motion patterns, then pre-processing the signals to extract the feature parameters and selecting them with partial least squares (PLS) method, and finally finding and classifying the gait pattern through improved probabilistic neural network (PNN). |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
|
|
|