基于HMM与SVM的运动跌倒预警模型研究——以安徽省老年群体为例

Research on Sports Fall Warning Model Based on HMM and SVM:Taking the Elderly Population in Anhui Province as an Example

人口老龄化问题日益加重,老年人的身体机能随着年龄的增加而降低.为提升运动跌倒预警的检测率和延长预留时间,提出基于HMM与SVM的运动跌倒预警模型,通过人体行为特征分析和判别算法设计实现运动跌倒的成功预测,并以安徽省老年群体为背景,从中采集实验数据.实验结果表明,SVM模型的检测率为69.5%,基于HMM与SVM的运动跌倒预警模型的检测率为91%,对比结果表明基于HMM与SVM的运动跌倒预警模型针对突发情况的误判率较低,具备良好的泛化能力.基于HMM与SVM的运动跌倒预警模型检测的平均预留时间是245 ms,满足充气机构反应耗时,结果有助于老年人跌倒预警以及人体保护.

The problem of aging population is attracting more attention,while the physical functions of elderly people decrease with age.In order to improve the detection rate and extend the reserved time of sports fall warning,a sports fall warning model based on HMM and SVM is proposed in the study.The successful prediction of running falls is achieved through human behavior feature analysis and discrimination algorithm design.Experimental data is collected from the elderly population in Anhui Province as the background.The experimental results show that the detection rate of the SVM model is 69.5%,and the detection rate of the motion fall warning model based on HMM and SVM is 91%.The comparison results show that the motion fall warning model based on HMM and SVM has a low misjudgment rate for sudden situations and has good pan Chinese ability.The average reserved time for detect‐ing the motion fall warning model based on HMM and SVM is 245 ms,which meets the reaction time of the inflation mechanism.The results are helpful for elderly fall warning and human protection.