【作者】 王琳；

【导师】 刘卓夫；

【作者基本信息】 哈尔滨理工大学 ， 检测技术与自动化装置， 2016， 硕士

【摘要】 在日常工作和生活中,由于科技水平的进步、生活方式的改善以及工作强度的提升,人们对座椅热舒适度的要求越来越高。舒适的座椅有助于缓解身体疲劳、提高工作效率。温度是影响热舒适度关键的因素,因此研究人体与座椅接触面温度场对提高座椅热舒适度具有重要意义。因此,本文主要针对国内外的尚不成熟的研究现状,利用温度场采集装置对人体-座椅接触面的温度场进行数据获取,并通过EMD算法抑制噪声、三维图像成像等数据处理方法,结合AR模型,客观的预测和评估温度场的变化规律,进而研究和讨论预测结果和实际问题的解决。本文进行了人体-座椅面传感器类型的选择以及数量和布局的分析,设计温度场采集装置,应用接触式测量方式,采集接触面的温度数据并予以保存。根据实验数据的特点,设计EMD滤波器进行数据的预处理,使之平滑,以便温度数据适用于后续的训练分析。在此数据处理基础上,主要应用AR自回归预测模型进行数据的分析和训练,实现的方法主要依靠决策树算法,用已知部分预测未知部分,应用MATLAB进行仿真和分析。对于使用者来说,方便提前进行作出决策、及时调整坐姿、更换座椅等操作,达到最佳的护理效果。根据实验设定要求,在分析过程中主要对采集的数据,采取基于决策树算法的AR模型预测,并进行组内和组间均方根误差分析,从而验证模型及预测方法的可行性和准确性。实验结果表明,组内对未来15分钟温度变化进行预测的5次重复实验,误差结果显示平均RMSE(均方根误差)小于0.4℃,证明该自回归预测数据模型可以提供可接受精度的预测能力。组间交叉实验结果同样证实,该预测结果最大偏差为0.47±0.06℃。更多还原

【Abstract】 With the advancement of technology level, the improvement of life styles and the promotion of work intensity, the demand for seat thermal comfort is becoming increasingly high in daily work and life. It is beneficial to alleviate body fatigue and improve the efficiency of work while taking a comfortable chair.It is quite significant to study temperature field between human body and seat contact surface for improving the seat thermal comfort, as the temperature is the key factor that affects the thermal comfort.According to the previously investigative literature, temperature field data between human body and seat contact surface were collected by using the temperature acquisition equipment in this paper. Through data processing such as smoothing filter of EMD and 3D imaging, and certain predictive analysis such as Decision Tree, an objective evaluation and prediction approach based on the variation of temperature field was applied. In addition, the results of analyzing temperature field and solving practical problems were discussed.In this paper, the number and arrangement of sensors array between human body and seat contact surface was analyzed. A temperature field acquisition system was designed which measures the temperature data and saves them to PC’s hard drive. According to the characteristics of experimental data, EMD filter for data preprocessing was designed. The aim is to smooth raw data and apply processed data to subsequent analysis. Based on this pretreatment, the auto regression prediction model was employed to analyze and train the data. Based on the Decision Tree algorithm, the unknown data were predicted by the known part of the section. Finally, MATLAB was used to simulate and analyze. From the perspective of chair users, it is easy to make decision in advance, timely to adjust the sitting posture, and to replace the seat for achieving the best caringeffect. According to experimental sets and requirements, collected data were mainly analyzed and predicted based on Decision Tree about Autoregressive predict model. Root mean square errors of within group and between groups were discussed. Furthermore, the feasibility and accuracy analysis of prediction method and AR model were evaluated and verified in terms of solving the temperature field prediction problem.The experimental results show that, the average RMSE is less than 0.4 ℃within the group, which proves that the AR model can provide the predictive ability of the acceptable accuracy, through the repeated predicted experiment of future 15 minutes. The cross experiment result also confirmed that the prediction accuracy is 0.47±0.06℃.更多还原