当前位置: X-MOL 学术Urban Clim. › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
The evaluation on artificial neural networks (ANN) and multiple linear regressions (MLR) models for predicting SO2 concentration
Urban Climate ( IF 6.4 ) Pub Date : 2021-04-13 , DOI: 10.1016/j.uclim.2021.100837
Seyedeh Reyhaneh Shams , Ali Jahani , Saba Kalantary , Mazaher Moeinaddini , Nematollah Khorasani

Nowadays air quality is the main issue in urban areas that have been affecting human health, the environment, and the ecosystem. So, governmental authorities, environmental and health agencies usually need the prediction of daily air pollutants. This prediction is often based on statistical relations between various conditions and air pollution. This study aims to compare the performance of Multiple Linear Regression (MLR) and Multi-layer perceptron (MLP) for predicting SO2 concentration in the air of the Tehran. Different parameters namely meteorological parameters, urban traffic data, urban green space information, and time parameters were chosen for the prediction of SO2 daily concentration. Considering result, the correlation coefficient (R2), and root means square error (RMSE) of the MLR model are 0.708, and 6.025, respectively while these values for the MLP equal 0.9 and 0.42. According to the result of sensitivity analysis, the value of the one-day time delay, park indicator, season/year, and the total area parks were the main factors influencing SO2 concentration. MLP model suggested in this research could be applied to support, analysis, and improve predicting air pollution and air quality management. This study shows the importance of modeling and application of ANN in presenting management strategies to reduce urban pollution.



中文翻译:

用于预测SO 2浓度的人工神经网络(ANN)和多元线性回归(MLR)模型的评估

如今,空气质量已成为影响人类健康,环境和生态系统的城市地区的主要问题。因此,政府部门,环境和卫生机构通常需要预测每日的空气污染物。该预测通常基于各种条件与空气污染之间的统计关系。这项研究旨在比较多层线性回归(MLR)和多层感知器(MLP)在预测德黑兰空气中SO 2浓度方面的性能。选择不同的参数,如气象参数,城市交通数据,城市绿地信息和时间参数,以预测SO 2的日浓度。考虑结果,相关系数(R 2),MLR模型的均方根误差(RMSE)分别为0.708和6.025,而MLP的这些值等于0.9和0.42。根据敏感性分析的结果,一天延迟时间,停车指标,季节/年和总停车面积的值是影响SO 2浓度的主要因素。这项研究中提出的MLP模型可以用于支持,分析和改进对空气污染和空气质量管理的预测。这项研究表明,在提出减少城市污染的管理策略时,人工神经网络的建模和应用非常重要。

更新日期:2021-04-13
down
wechat
bug