基于惯性传感的跌倒检测系统的设计与实现

发布时间：2018-06-14 14:30

本文选题：跌倒检测 + 惯性传感　；参考：《电子科技大学》2017年硕士论文

【摘要】：人口老龄化是一个世界性的问题。根据世界卫生组织的定义和我国第六次全国人口普查数据,我国已经步入老龄化社会。在日常生活中,老年人极易出现跌倒等意外情况。据央视报道,在中国每年约有4000万65岁以上的老人意外跌倒。当老人跌倒时,若未得到及时救助,会造成不可挽回的后果,因此跌倒检测技术有很强的实际应用价值。本文的具体工作内容为:1)开展关于惯性传感节点佩戴位置对于数据影响的实验,选定腰部作为佩戴位置。将真实场景下跌倒样本与实验采集获得的跌倒样本进行对比,分析了二者的特点。对系统进行需求分析并确定设计目标,提出系统的技术方案。2)采集跌倒和日常行为活动的惯性数据样本,构建算法设计所需的数据集。对所获得的原始数据集进行预处理;使用Relief算法进行特征选择,选取了分类器设计所需的特征集合;兼顾系统开发,对比算法特异度和敏感度,选择决策树作为分类模型。3)实现包含云端服务器、安卓应用和惯性传感节点的跌倒检测系统。当检测到跌倒时,惯性传感节点向智能手机发送检测结果,由手机进行定位并向云端服务器发送求助信息,而云端服务器则通知社区或医院的监护人员对跌倒者进行救助。4)对整个系统的功能进行验证,验证结果表明系统功能均正确实现。对系统跌倒检测算法的敏感度和特异度进行测试,测试实验结果表明该跌倒检测算法敏感度为95.56%、特异度为98.00%,说明系统能够较为准确地检测跌倒且误报较少。本文通过分析跌倒检测的国内外研究现状,对比不同跌倒检测技术,选择基于惯性传感信号的检测技术进行方案设计,结合现在成熟的云端服务器技术和安卓智能手机,实现了一套基于惯性传感的跌倒检测系统。系统功能完善,具有较强的实用价值。
[Abstract]:The aging of the population is a worldwide problem. According to the definition of the WHO and the data of the sixth national population census, China has entered an aging society. In daily life, the elderly are extremely prone to fall and other accidents. According to CCTV, it is reported that some people over 40 million and 65 years of age in China fall unexpectedly every year. When people fall, it will cause irreparable consequences if they are not saved in time. Therefore, the fall detection technology has a strong practical application value. The specific contents of this paper are as follows: 1) to carry out the experiment on the influence of the position of the inertial sensor node on the data, select the waist as the position to wear. The characteristics of the two fall samples are analyzed, and the characteristics of the two are analyzed. The system needs analysis and the design target is determined. The system's technical scheme is proposed to collect the inertial data samples from the fall and daily activities, and the data set required for the algorithm design is constructed. The original data set is preprocessed by the Relief calculation. The feature selection is carried out by the method, and the feature set required by the classifier is selected. The system is developed, the specificity and sensitivity of the algorithm are compared, and the decision tree is selected as the classification model.3) to implement the fall detection system containing cloud server, Android application and inertial sensor nodes. When the fall is detected, the inertial sensor node sends to the smart phone. Send the detection results, locate the mobile phone and send the help information to the cloud server, and the cloud server notifies the community or hospital guardians to help the fall person to help.4) to verify the function of the whole system. The results show that the system functions are realized correctly. The sensitivity and specificity of the system fall detection algorithm are advanced. The test results show that the sensitivity of the fall detection algorithm is 95.56% and the specificity is 98%. It shows that the system can detect the fall and less misinformation more accurately. In this paper, the status of the fall detection at home and abroad is analyzed, and the detection technology based on the inertial sensing signal is selected to choose the detection technology based on the inertial sensing signal. Design, combined with the mature cloud server technology and Android smart phone, a set of fall detection system based on inertial sensing is realized. The system has perfect function and has strong practical value.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP274;TP212.9

【参考文献】