基于震动信号的异常步态识别

Recognition of Abnormal Gait Based on Vibration Signal

异常步态的识别对老年人的健康看护有很大帮助.现有的相关研究主要通过图像采集设备或穿戴设备获取相关特征信息进行识别,这些方法大多都具有侵入性或对用户有较高的操作要求.本文研究实现了一种基于对脚步震动信号为识别源进行异常步态和跌倒检测的系统原型,该系统通过本文设计的应用于大范围数据采集的多传感器协同信号采集方法采集信号,从中分割出有效部分作为活动元,去噪后再使用改进的动态时间规整算法(Dynamic Time Warping,DTW)计算出代表活动元之间差异性的异常指数,接着由K最近邻(K-Nearest Neighbor,KNN)算法分类异常指数,得到初步表征用户步态的推测值,最后由隐马尔科夫模型(Hidden Markov Model,HMM)进一步处理推测值,识别出用户的步态.实验结果表明,本文提出的方法能够有效在不同的步态模式下识别异常步态,在稳定的环境中识别准确率达到96%,在具有不稳定地板的环境中准确率为94%.

The recognition of abnormal gait is of great help to the health care of the elderly.Existing related research mainly uses image acquisition equipment or wearable equipment to obtain relevant feature information for identification.Most of these methods are invasive or have high operational requirements for users.This paper studies and realizes a sys⁃tem prototype based on the detection of abnormal gait and fall based on foot vibration signals as the source of identifica⁃tion.This paper first designs a multi-sensor cooperative signal acquisition method to achieve a large range of signal acquisi⁃tion,and separate the effective part from it as the active element.After the collected active elements are denoised,an im⁃proved dynamic time warping algorithm(DTW)is used to calculate the abnormal index representing the difference between active elements,then the abnormal index is classified by the K nearest neighbor(KNN)algorithm,and the inferred value that initially characterizes the user's gait is obtained.The inferred value is further processed by hidden Markov model(HMM)to identify the user's gait.The experimental results show that the method proposed in this paper can effectively identify abnormal gait in different gait modes,with an accuracy of 96%in a stable environment,and an accuracy of 94%in an environment with unstable floors.