Journal of Frontiers of Computer Science and Technology ›› 2022, Vol. 16 ›› Issue (12): 2832-2840.DOI: 10.3778/j.issn.1673-9418.2104086

• Artificial Intelligence • Previous Articles     Next Articles

Research on Prediction Model of Physical Activity Energy Expenditure with Wearable Sensors

WANG Lin1, SUN Qian1, MA Xiaona1, GAO Yongyan2, LIU Yi1, MA Hongwei1, YANG Dongqiang1,+()   

  1. 1. College of Computer Science and Technology, Shandong Jianzhu University, Jinan 250101, China
    2. Department of Sports, Shandong Jianzhu University, Jinan 250101, China
  • Received:2021-04-13 Revised:2021-06-01 Online:2022-12-01 Published:2021-06-08
  • About author:WANG Lin, born in 1975, M.S. candidate, lecturer. Her research interests include machine learning, intelligent city, etc.
    SUN Qian, born in 1983, M.S. candidate, lecturer. Her research interest is intelligent city.
    MA Xiaona, born in 1977, M.S. candidate, associate professor. Her research interests include software testing, machine learning, etc.
    GAO Yongyan, born in 1980, M.S. candidate, associate professor. Her research interests include physical education teaching and training, physical culture, etc.
    LIU Yi, born in 1978, M.S. candidate, lecturer. His research interests include machine vision, artificial intelligence, etc.
    MA Hongwei, born in 1969, Ph.D. candidate, professor, M.S. supervisor. His research interests include computer system architecture, computer network, machine learning, etc.
    YANG Dongqiang, born in 1970, Ph.D. candidate, associate professor, M.S. supervisor. His research interests include natural language processing, artificial intelligence, etc.
  • Supported by:
    Natural Science Foundation of Shandong Province(ZR2018MF012);Natural Science Foundation of Shandong Province(ZR2020QF029);Ph.D. Foundation of Shandong Jianzhu University(XNBS1811)

可穿戴传感器的人体活动能量预测模型研究

王琳1, 孙倩1, 马晓娜1, 高永艳2, 刘毅1, 马宏伟1, 杨东强1,+()   

  1. 1.山东建筑大学 计算机科学与技术学院,济南 250101
    2.山东建筑大学 体育部,济南 250101
  • 通讯作者: +E-mail: ydq@sdjzu.edu.cn
  • 作者简介:王琳(1975—),女,山东泗水人,硕士研究生,讲师,主要研究方向为机器学习、智慧城市等。
    孙倩(1983—),女,山东临沂人,硕士研究生,讲师,主要研究方向为智慧城市。
    马晓娜(1977—),女,河北衡水人,硕士研究生,副教授,主要研究方向为软件测试、机器学习等。
    高永艳(1980—),女,潍坊昌邑人,硕士研究生,副教授,主要研究方向为体育教学与训练、体育文化等。
    刘毅(1978—),男,山东威海人,硕士研究生,讲师,主要研究方向为机器视觉、人工智能等
    马宏伟(1969—),男,山东济阳人,博士研究生,教授,硕士生导师,主要研究方向为计算机系统结构、计算机网络、机器学习等。
    杨东强(1970—),男,山东招远人,博士研究生,副教授,硕士生导师,主要研究方向为自然语言处理、人工智能等。
  • 基金资助:
    山东省自然科学基金(ZR2018MF012);山东省自然科学基金(ZR2020QF029);山东建筑大学博士基金(XNBS1811)

Abstract:

To solve the contradiction between multiple wearable sensor features and the limited computing power and storage capacity of embedded devices, feature engineering is used to select the best features for predicting physical activity energy expenditure (PAEE) on the basis of data fusion of multiple sensors (accelerometer and gyroscope sensors). In the data preprocessing stage, time-domain and frequency-domain features of the sensor are extracted by using sliding window technology, and sinusoidal curve fitting is used for dataset at three velocity levels, finally hypothesis testing is carried out to check data outliers. A WEKA experimental platform is constructed based on filtering, warpper and embedded feature selection algorithms and machine learning prediction models such as multiple linear regression, regression tree, support vector machine and neural network. Finally, the optimal model is selected by evaluating the correlation coefficient and mean absolute error of each model during the decision level fusion. The dataset with jitter is used as the test data, which shows that feature selection can mitigate model overfitting and improve the model’s generalization ability and robustness. Embedded feature selection adopts classical elastic network algorithm. Experimental results show that the features extracted from accelerometer sensors play a more decisive role than those from gyroscope sensors in PAEE and the neural network model of multi-sensor feature fusion based on correlation coefficient method is the optimal model.

Key words: physical activity energy expenditure, feature selection, neural network

摘要:

为了解决可穿戴传感器特征过多与嵌入式设备计算能力和存储能力有限的矛盾问题,在多传感器(加速度传感器、陀螺仪传感器)数据融合的基础上,采用特征工程的方法选出人体运动能量消耗预测(PAEE)的最优特征。在数据预处理阶段,使用滑动窗口技术提取传感器的时域、频域特征,对三个速度水平的数据集使用正弦曲线拟合,并通过显著性差异检验分析选出有效数据。构建了过滤式、封装式和嵌入式特征选择算法与多线性回归、回归树、支持向量机和神经网络等机器学习预测模型结合的WEKA实验平台。最后决策级融合时,通过评估每个模型的相关系数和平均绝对误差选择出最优模型。模型训练时采用带抖动的数据集作为测试集,避免出现模型的过拟合现象,提高模型的泛化能力和鲁棒性。嵌入式特征选择采用经典的弹性网络算法。实验结果表明,在PAEE中加速度计传感器的特征比陀螺仪传感器的特征更具有决定性的作用,基于相关系数方法的多传感器特征融合的神经网络模型是最优模型。

关键词: 人体活动能量消耗, 特征选择, 神经网络

CLC Number: