摘要
文章描述和评估了一个基于智能手机的加速度传感器进行活动识别的系统,它可以识别用户的物理活动。为了实施该系统,从10个用户的日常活动中采集加速度数据,像手机分离、空闲、步行、奔跑和跳跃,然后汇集这个时间序列数据汇总成5 min的时间间隔,最后使用这个训练结果数据去预测未知的活动。测试实验表明,使用该模型可以精确的对上述5种活动进行分类和预测。
A behavior identity system developed with the acceleration sensor of a smart phone was designed. It can identify the physical activities users. To design the system, daily behaviors of ten users were tested when they were separated from the phone, idle, walk, run and jump. The obtained time series data were collected into 5 - minute intervals. Finally, unknown behaviors could be forecast with this training data. The test results indicate that using this model can accurately predict and classify the five kinds of behaviors above.
作者
王星峰
WANG Xing - feng(School of Information Engineering, Eastern Liaoning University, Dandong 118003, China)
出处
《辽东学院学报（自然科学版）》
CAS
2017年第1期64-71,共8页
Journal of Eastern Liaoning University:Natural Science Edition
关键词
加速度传感器
活动分类
数据挖掘
智能手机
acceleration sensor
activity classification
data mining
smart phone
