没有合适的资源?快使用搜索试试~ 我知道了~
人工智能-数据挖掘-基于数据挖掘技术的空气质量指数预测研究.pdf
1.该资源内容由用户上传,如若侵权请联系客服进行举报
2.虚拟产品一经售出概不退款(资源遇到问题,请及时私信上传者)
2.虚拟产品一经售出概不退款(资源遇到问题,请及时私信上传者)
版权申诉
0 下载量 125 浏览量
2022-06-28
11:36:15
上传
评论
收藏 1.97MB PDF 举报
温馨提示
试读
70页
人工智能-数据挖掘-基于数据挖掘技术的空气质量指数预测研究.pdf
资源推荐
资源详情
资源评论
基于数据挖掘技术的空气质量指数预测研究
摘 要
近年来空气污染状况日益严重,频繁出现的重污染天气对人们的日常
生活乃至生命健康都造成了严重影响。空气质量指数(AQI)的预报作为公
众知晓未来空气状况最直接的途径,不仅能为环保部门开展空气环境治理
工作提供指导,同时也能够提醒公众合理规避重度污染天气。但由于影响
空气质量的因素较多,气象环境的复杂性、污染物之间的非线性关系都为
空气质量指数的预报造成了困难。且传统的潜势预报、数值预报也并未做
到对海量历史数据的充分利用,因此在预报中依然存在着准确度不高、实
效性不强等局限性。
针对这些问题,论文以西安市
2014
年
1
月
28
日
-2016
年
8
月
29
日间
空气质量监测站点每日所采集的数据为研究对象,使用数据挖掘技术中的
灰色系统理论以及神经网络技术,建立了两种空气质量指数预测模型。并
通过对模型的进一步优化,提高预测准确度以及结果的可接受度。论文的
主要工作如下:
(1)选择合适的指标数据参与空气质量指数预测模型的建立。指标的
选取应从环境要素以及气象条件两方面考虑。根据环境空气质量标准,选
取 SO
2
、PM
2.5
、PM
10
、CO、O
3
、NO
2
、AQI 等 7 项污染指标,以及风力级
别、平均湿度、最高温度、最低温度、平均温度等 5 项气象因素共计 12 个
指标作为此次预测建模的主要研究对象。
(2)对参与预测建模的数据进行预处理。首先,需要对监测站点采集
的数据进行初步筛选,在剔除无效和缺失数据后,保留 940 条数据。其次,
考虑到不同影响因子的数据取值范围以及单位存在明显差异,会对预测结
果产生影响,因此采用 mapminmax 函数对数据进行归一化处理,以消除各
类数据之间的量纲差别。
(
3
)建立基于
GM(1,1)
的空气质量指数预测模型。根据灰色预测方法
的适用特性,以空气质量指数的历史数据作为研究对象,建立基于 GM(1,1)
预测模型。通过对预测结果进行分析,评估灰色系统理论在空气质量预测
上的效果。
(4)建立基于 BP 神经网络的空气质量指数预测模型。根据参与建模
的数据调整 BP 神经网络的参数,确定合理的网络结构,并在 MATLAB 平
台下编写完整预测程序,对空气质量指数进行预测。计算结果的平均绝对
万方数据
百分比误差、可接受度以及空气质量等级预测的正确率,以此来评价预测
模型。
(5)对 BP 神经网络的预测模型进行优化。为进一步提高预测的精确
性,分别使用主成分分析法和遗传算法对 BP 神经网络的预测模型进行优
化。一方面通过降低输入变量的维度来消除训练过程的复杂度,而另一方
面则是通过优化 BP 神经网络的初始参数来提高模型的预测能力。最后,将
优化后的 PCA-BP、GA-BP 神经网络的预测模型与单一 BP 神经网络的预测
模型进行对比。
结果表明,在有效信息较少的情况下,灰色预测模型对于空气质量指
数的预测有一定意义。而在数据较丰富的情况下,单一 BP 神经网络预测模
型的平均绝对百分比误差为
21.96%
。在对
BP
神经网络进行主成分分析以
及遗传算法优化后,误差分别减小了 2.23%和 3.67%,且结果的可接受度以
及空气质量等级预测的准确率也均有提高。优化后的模型虽然不能完全准
确对空气质量指数进行预测,但其结果仍具有一定的参考价值。
关键词:空气质量指数,灰色系统理论,
BP
神经网络,主成分分析,遗传
算法
万方数据
RESEARCH ON AIR QUALITY INDEX
FORECASTING BASED ON DATA
MINING TECHNOLOGY
ABSTRACT
In recent years, air pollution is getting worse, and frequent heavy pollution
weather has seriously affected people's daily life and health. The air quality
index (AQI) forecast which is the most direct way for the public to know the
future air quality, not only provides guidance for the environmental protection
department to carry out air pollution management, but also reminds the public to
avoid the severe pollution weather. However, there are so many factors which
make the AQI forecast become difficult, such as complexity and variability of
meteorological environment, the nonlinear relationship between pollutants, etc.
And the traditional potential forecast and numerical forecast do not make full
use of the massive historical data. Therefore, there are still some limitations in
AQI forecast, such as low accuracy and low efficiency.
Aiming at these issues, this thesis acquired the data by Xi'an air quality
monitoring station from 2014/1/28 to 2016/8/29, and took the data as research
object. We established two models of AQI by using the grey system theory and
neural network technology which belong to data mining. Through the further
optimization of the model, the accuracy of prediction and the acceptability of
the results can be improved. The main work of this paper is as follows:
(1)We selected appropriate index data to establish AQI prediction model.
The selection of indicators should be considered from environmental factors and
meteorological conditions. According to the air quality standards, including 7
pollution indicators such as SO
2
, PM
2.5
, PM
10
, CO, O
3
, NO
2
, AQI, as well as 5
meteorological factors such as the wind level, mean humidity, highest, lowest,
mean temperature, as the main research objects of the forecast model.
(2)We preprocessed the data involved in predictive model. Firstly, the data
collected from the monitoring site was initially screened. After eliminating
invalid and missing data, 940 data was retained. Next, in view of the different
influence factors whose data range and unit has obvious differences will have
万方数据
impacts on the prediction results. Thus, we used the mapminmax function to
normalize the data to eliminate the dimensional difference between various
kinds of data.
(3)We established an AQI prediction model based on GM(1,1). According
to the characteristics of grey prediction method, putting the historical data of
AQI as the research object, we established the prediction model based on
GM(1,1). The prediction results were analyzed to evaluate the effect of grey
system theory on air quality prediction.
(4)We established an AQI prediction model based on BP neural network.
According to the data of the model to adjust the parameters of the BP neural
network, we determined a reasonable network structure, and wrote a complete
prediction program on MATLAB to make the AQI forecast come true. In order
to evaluate the prediction model, we calculated the average absolute percentage
error, acceptability and the correct rate of air quality grade.
(5)We optimized the BP neural network prediction model. In order to
improve the accuracy of prediction, principal component analysis and genetic
algorithm were used to optimize the prediction model of BP neural network. On
the one hand, the complexity of the training process is reduced by decreasing
the dimension of the input variables. On the other hand, we improved the
predication ability of the model by optimizing the initial parameters of BP
neural network. Finally, the prediction model of the PCA-BP and GA-BP neural
network were compared with the single BP neural network prediction model.
The results show that the grey forecast model is of great significance to the
prediction of air quality index in the case of less effective information. In the
situation of abundant data, the mean absolute percentage error of a single BP
neural network prediction model is 21.96%. After the optimizing of the BP
neural network by principal component analysis and genetic algorithm, the error
is reduced by 2.23% and 3.67% respectively, and the acceptability of the results
and the accuracy of air quality forecast are also improved. Even though the
optimized model cannot completely accurately predict the air quality index, the
results still have some reference value.
KEY WORDS: Air quality index, grey system theory, BP neural network,
principal component analysis, genetic algorithm
万方数据
I
目 录
摘 要............................................................................................................................................ I
ABSTRACT...............................................................................................................................III
目 录
............................................................................................................................................ I
1 绪论..........................................................................................................................................1
1.1
研究的目的和意义
...................................................................................................... 1
1.2
国内外应用研究现状
.................................................................................................. 3
1.2.1 国外研究现状................................................................................................... 3
1.2.2
国内研究现状
................................................................................................... 4
1.3
本文的主要工作及内容安排
...................................................................................... 5
2 相关理论与技术基础............................................................................................................. 7
2.1
数据挖掘与人工智能
.................................................................................................. 7
2.1.1
数据挖掘的概述与应用
...................................................................................7
2.1.2 人工智能的概述与应用...................................................................................7
2.2
人工神经网络
.............................................................................................................. 8
2.2.1
人工神经网络概述
........................................................................................... 8
2.2.2 人工神经元模型............................................................................................... 9
2.2.3
人工神经网络的分类
..................................................................................... 11
2.2.4
人工神经网络的学习规则
.............................................................................12
2.3 灰色系统理论............................................................................................................ 13
2.3.1
灰色系统理论概述
......................................................................................... 13
2.3.2
灰色预测建模及其特点
.................................................................................13
2.4 主成分分析法............................................................................................................ 14
2.5
相关软件介绍
............................................................................................................ 14
2.5.1 SPSS
软件
....................................................................................................... 14
2.5.2 MATLAB 软件................................................................................................15
2.6
本章小结
.................................................................................................................... 16
3
空气质量指数的预测原理及数据的处理
...........................................................................17
3.1 空气质量指数(AQI)预测原理.............................................................................17
3.2
数据介绍以及预处理
................................................................................................ 18
3.2.1
数据来源
......................................................................................................... 18
3.2.2 数据介绍......................................................................................................... 19
万方数据
剩余69页未读,继续阅读
资源评论
programyp
- 粉丝: 88
- 资源: 1万+
上传资源 快速赚钱
- 我的内容管理 展开
- 我的资源 快来上传第一个资源
- 我的收益 登录查看自己的收益
- 我的积分 登录查看自己的积分
- 我的C币 登录后查看C币余额
- 我的收藏
- 我的下载
- 下载帮助
最新资源
- 批量word文件内容替换工具1.0 (批量实现多个 Word 文档文件文字替换利器).exe
- Cartoon GUI Pack 1.2.zip
- 【数据集和代码】基于加速度传感器的步态识别行人分类实验(可做步态识别)
- 我分享个魔兽内存修改器
- Python毕业设计基于Django的网易云数据分析可视化大屏系统的设计与实现+使用说明+全部资料(优秀项目).zip
- mp3 idv2,idv1,frame分析工具
- Python毕业设计基于Django的网易云数据分析可视化大屏系统的设计与实现+使用说明+全部资料(高分项目).zip
- 人工兔优化算法ARO MATLAB源码, 应用案例为函数极值求解以及优化svm进行分类,代码注释详细,可结合自身需求进行应用
- 人才网站设计-asp.net+sql-(系统源码)
- asp.net+sql人才网站设计-含系统源码
资源上传下载、课程学习等过程中有任何疑问或建议,欢迎提出宝贵意见哦~我们会及时处理!
点击此处反馈
安全验证
文档复制为VIP权益,开通VIP直接复制
信息提交成功