基于MEMS三轴加速度传感器的摔倒检测

MEMS Tri-Axial Accelerometer Based Fall Detection

摔倒作为人体活动的一部分,是影响人体健康的一大因素,尤其对病人和老年人而言,摔倒检测至关重要。基于MEMS三轴加速度传感器采集的人体活动加速度信号,提出了一种基于固定阈值的信号幅度向量滑动平均法SVMSA。该方法根据人体活动时的加速度信号特征,利用预先设定的阈值对加速度信号幅度向量SVM的滑动平均SVMSA进行判决,同时使用差分信号幅度域DSMA区分快速跑步等剧烈运动,准确实现了人体的摔倒检测。主要优势在于分析并区别了人体快速跑步等剧烈运动对摔倒检测的影响。通过对8位实验者的测试,该算法实现了94.4%的精确度。实验表明该算法能够较为准确地实现人体的摔倒检测。

As a part of human activities,fall is one of the key factors affecting human health,especially for patients and elders,fall detection is of much importance. This paper presents a method of Signal Magnitude Vector Sliding Average（SVMSA）with Fixed Threshold,based on the acceleration signals of human activity acquired from a MEMS triaxial accelerometer. By extracting the characteristics of human activity acceleration signals, this algorithm accurately achieves human＆#39;s fall detection,using the prefixed threshold to judge the SVMSA and the differential signal magnitude area（ DSMA） to distinguish fast running. The major advance lies in the attempt to analyze and distinguish human＇s fall and other intense activities,like running fast. By testing eight participants,we get 94. 4% accuracy. Experimental results indicate the proposed algorithm can realize human＇s fall detection with much accuracy.