面向可穿戴设备的空气质量监测系统的设计与实现

Design and implementation of air quality monitoring system based on wearable devices

实时监测的实现主要依赖于对空气质量数据的收集、处理、存储以及显示,而空气质量预测功能是项目开发的难点。这里所采用的预测算法,其核心是针对影响空气质量指数的污染因子的历史浓度做出分析,根据其变化规律建立预测模型,通过非线性最小二乘法拟合各个污染因子的变化曲线,求出其预测值,进而计算出空气质量预测值。所提出的预测算法是建立在统计计算的基础之上,对已有数据进行挖掘和分析,通过机器学习的方法来做出预测。在可穿戴设备的开发方面,需要考虑到其交互特点,结合Android开发技术开发出便捷的应用。经过测试,该算法的空气质量等级预测的准确率达到了47.77%,而首要污染物预测的准确率则达到了73.66%,满足基本需求。

The realization of real-time monitoring is mainly dependent on the collection,processing,storage and display of air quality data,but the air quality prediction function is the difficulty of the project development. The prediction algorithm is used to analyze the historical concentration of the pollution factors affecting on the air quality index,and establish a prediction model according to the concentration variation. The changing curve of each pollution factor is fitted by the nonlinear least square method to find the prediction value,and then the air quality prediction value can be figured out. The proposed algorithm is based on the statistical calculation to mine and analyze the existing data,and then the prediction is obtained by means of machine learning. In the aspect of wearable device development,it is necessary to take into account the interactive features and combine with the convenient applications developing by Android development technology. The test results show that the accuracy rate of the air quality grade prediction can reach up to 47.77%,and the accuracy of the primary pollutant prediction can reach up to 73.66%,which can satisfy the basic demands.