基于惯性测量单元的远程跌倒报警系统  

Remote Fall Alarm System Based on Inertial Measurement Unit

作　　者：于玮[1] 许彬[2] 

机构地区：[1]南通大学工程训练中心,南通226019 [2]南通大学护理学院

出　　处：《单片机与嵌入式系统应用》2017年第4期55-59,共5页Microcontrollers & Embedded Systems

基　　金：2014年南通大学自然科学类科研基金一般项目(14Z006)

摘　　要：设计了一种基于九轴惯性测量单元和无线通信网的远程跌倒报警系统。从数学模型入手,将跌倒过程划分为身体失衡、碰撞冲击、倒地静止这三个阶段,研究了加速度与角速度向量模值、瞬时姿态角等参数;采用卡尔曼滤波算法解决了陀螺仪静态测量时的零点漂移问题;利用佩戴于腰间的惯性测量传感器对运动参数进行测量,并由微处理器作出基于多阈值的最终判决;当检测到有跌倒发生时,系统立刻发出多重报警,最大程度地降低跌倒引发的伤害。实验结果显示,该系统对跌倒识别的正确率超过90%,可用于老年人的一般医疗监护场合。A remote fall alarm system based on nine-axis inertial measurement unit and wireless communication network is proposed. From the mathematical model, the fall process is divided into three stages including imbalance, crashing and lying on the ground. The vector modulus of acceleration and angular velocity,the instantaneous attitude angle and other parameters are studied. The Kalman filtering algorithm is used to cancel the zero drift of the gyro in static measurement. The motion parameters are measured by the inertial measurement sensors tied to the waist. And the final judgment is made by microprocessor based on multi-threshold. When a fall occurs,the system sends the alarm information in the different ways immediately to reduce the damage to the greatest extent. The experiment results show that the accuracy of fall recognition is more than 90 %. It can be used in general situation for elderly health care.

关 键 词：跌倒检测 远程监护 姿态解算 四元数算法 卡尔曼滤波 

分 类 号：TP277[自动化与计算机技术—检测技术与自动化装置]