Student Research Committee, Department of Environmental Health Engineering, School of Health, Shiraz University of Medical Sciences, Shiraz, Iran.
Department of Environmental Health Engineering, School of Health, Shiraz University of Medical Sciences, Shiraz, Iran.
Environ Sci Pollut Res Int. 2023 Feb;30(8):21345-21359. doi: 10.1007/s11356-022-23602-8. Epub 2022 Oct 21.
This study aimed to model the removal of formaldehyde as an indoor air pollutant by Nephrolepis obliterata (R.Br.) J.Sm. plant using response surface methodology (RSM) and artificial neural network (ANN) models, and optimization of the models by particle swarm optimization algorithm (PSO). The data obtained in pilot-scale experiments under a controlled environment were used in this study. The effects of parameters on the removal efficiency such as formaldehyde concentration, relative humidity, light intensity, and leaf surface area were empirically investigated and considered as model parameters. The results of the RSM model, with power transformation, were in meaningful compromise with the experiments. A multilayer perceptron (MLP) neural network was also designed, and the mean of squared error (MSE), mean absolute error (MAE), and R were used to evaluate the network. Several training algorithms were assessed and the best one, the Levenberg Marquardt (LM), was selected. The PSO algorithm proved that the highest removal efficiency of formaldehyde was obtained in the presence of light, maximum leaf surface area and relative humidity, and at the lowest inlet concentration. The empirical system breakthrough occurred at 15 mg/m of formaldehyde, and the maximum elimination capacity was about 0.96 mg per m of leaves. The findings indicated that the ANN model predicted the removal efficiency more accurately compared to the RSM model.
本研究旨在采用响应面法（RSM）和人工神经网络（ANN）模型，对肾蕨（Nephrolepis obliterata（R.Br.）J.Sm.）植物去除室内空气污染物甲醛的情况进行建模，并通过粒子群优化算法（PSO）对模型进行优化。本研究使用在受控环境下进行的中试规模实验获得的数据。实验中，经验性地考察了参数（如甲醛浓度、相对湿度、光照强度和叶片表面积）对去除效率的影响，并将这些参数作为模型参数。经功率变换后的 RSM 模型结果与实验结果存在合理的折衷。还设计了一个多层感知器（MLP）神经网络，并使用均方误差（MSE）、平均绝对误差（MAE）和 R 来评估网络。评估了几种训练算法，最终选择了最佳的算法——Levenberg-Marquardt（LM）算法。PSO 算法证明，在光照、最大叶片表面积和相对湿度存在的情况下，以及在最低入口浓度下，甲醛的去除效率最高。经验系统在甲醛达到 15mg/m 时出现突破，最大消除能力约为每平方米叶片 0.96mg。研究结果表明，与 RSM 模型相比，ANN 模型能更准确地预测去除效率。
