基于TSA-SVM的老人跌倒识别算法

Elderly Fall Recognition Algorithm Based on TSA-SVM

针对老人跌倒检测易受环境影响以及检测不够精确易出现误判的问题,提出了一种基于人体动作传感器的老人跌倒识别检测算法,采用被囊群算法(TSA)优化支持向量机(SVM)模型进行跌倒识别.针对人体动作传感器采集的数据,首先进行特征提取、降维等预处理,然后将预处理后的数据输入SVM模型进行训练,同时利用TSA算法寻找SVM最优参数,得到最优的跌倒识别模型,利用该模型即可进行跌倒识别.实验结果表明,本文所提算法的跌倒识别检测正确率可达96%以上,具有一定的优越性.

For the elderly fall detection is easy to be affected by the environment and the detection is not accurate enough to cause misjudgment,in this paper,a new method of detecting and identifying elderly falls based on human motion sensor is proposed.The algorithm uses the Tunicate Swarm Algorithm(TSA)Support Vector Machine(SVM)model for fall recognition.For human action sensor data,firstly,feature extraction,dimension reduction and other preprocessing are carried out,and then the preprocessing data is input into the SVM model for training.At the same time,TSA algorithm is used to find the optimal parameters of SVM to obtain the optimal fall recognition model,which can be used for fall recognition.The proposed algorithm’s accuracy surpasses can 96%,a remarkable superiority,as evidenced by the experimental results.