基于机器学习的城市环境空气质量评价研究

发布时间：2018-06-23 09:47

本文选题：机器学习 + 随机森林　；参考：《上海应用技术大学》2017年硕士论文

【摘要】：近年来,城市空气污染问题愈发突出,大量的汽车尾气、工业废气、扬尘等直接排放到城市环境中,远远超过了城市环境的自净能力,导致空气质量下降,直接危及到城市居民的健康和安全,城市空气质量问题已经引起了中国社会各界的高度重视。为了有效地控制空气污染,提高城市的空气质量,首先一定要对城市环境空气质量进行科学合理的评价,使城市居民以及环境保护部门更加客观地了解城市环境的空气质量,做出合理的生活安排及科学的防治措施。所以城市环境空气质量评价对于防治空气污染发挥着重要的作用。随着大数据时代的来临以及人工智能的兴起,传统的评价方法已经不能满足对海量传感器数据智能化处理的需求,大量学者开始基于大数据,利用智能方法来评价城市环境的空气质量。机器学习是人工智能的分支,本文将机器学习引入到城市环境空气质量评价中,利用机器学习中的随机森林算法来评价城市环境的空气质量,通过对随机森林模型训练,找到多种空气污染物与空气质量等级之间的内在映射关系,建立随机森林评价模型,提高评价科学性和鲁棒性。本文在仿真实验时,将随机森林评价模型与支持向量机,朴素贝叶斯和K最邻近模型进行对比,仿真采用的数据为上海市2013年到2015年的部分空气质量真实数据,仿真实验得到了较好的效果,实验结果表明评价方法效果最好,准确性最高可达99.69%,同时本文对随机森林模型的性能进行了深入分析,进一步验证了该评价方法的适用性与稳定性,从分析可以看出本文模型的泛化误差对特征变量个数不是很敏感,并且在准确性与时间复杂度之间有较好折衷,可以用于准确有效的评价城市环境的空气质量。
[Abstract]:In recent years, the problem of urban air pollution has become more and more prominent. A large number of automobile exhaust, industrial waste gas and dust are discharged directly into the urban environment, which far exceeds the self purification capacity of the urban environment, which leads to the decline of air quality, which directly endangers the health and safety of urban residents. The problem of urban air quality has already caused the high social circles in China. In order to effectively control air pollution and improve the air quality of the city, first of all, we must make a scientific and rational evaluation of the air quality of the urban environment, so that the urban residents and the environmental protection departments can understand the air quality of the urban environment more objectively and make reasonable living arrangements and scientific prevention measures. Air quality assessment plays an important role in preventing air pollution. With the advent of the era of large data and the rise of artificial intelligence, the traditional evaluation method can not meet the demand for intelligent processing of mass sensor data. A large number of scholars begin to use intelligent methods to evaluate the air of the urban environment based on large data. Quality. Machine learning is the branch of artificial intelligence. This paper introduces machine learning into urban environmental air quality evaluation, uses random forest algorithm in machine learning to evaluate the air quality of urban environment, and through the training of random forest model, the internal mapping relationship between air pollution and air quality is found. The stochastic forest evaluation model is established to improve the scientific and robust evaluation. In the simulation experiment, the stochastic forest evaluation model is compared with the support vector machine, the naive Bias and the K nearest model. The simulation data are the real data of the air quality in Shanghai from 2013 to 2015, and the simulation experiment is better. The results show that the evaluation method has the best effect and the maximum accuracy can reach 99.69%. At the same time, the performance of the random forest model is analyzed in this paper, and the applicability and stability of the evaluation method are further verified. There is a good trade-off between accuracy and time complexity, which can be used to evaluate the air quality of urban environment accurately and effectively.
【学位授予单位】：上海应用技术大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：X823

【参考文献】