This paper conducts in-depth research and analysis on the commonly used models in the simulation process of air pollutant diffusion. Combining with the actual needs of air pollution, this paper builds an air pollution system model based on neural network based on neural network algorithm, and proposes an image classification method based on deep learning and Gaussian aggregation coding. Moreover, this paper proposes a Gaussian aggregation coding layer to encode image features extracted by deep convolutional neural networks. Learn a fixed-size dictionary to represent the features of the image for final classification. In addition, this paper constructs an air pollution monitoring system based on the actual needs of the air system. Finally, this article designs a controlled experiment to verify the model proposed in this article, uses mathematical statistics to process data, and scientifically analyze the statistical results. The research results show that the model constructed in this paper has a certain effect.

中文翻译：

---

基于神经网络的空气污染系统研究

本文对空气污染物扩散模拟过程中常用的模型进行了深入的研究和分析。结合空气污染的实际需求，建立了基于神经网络算法的基于神经网络的空气污染系统模型，并提出了一种基于深度学习和高斯聚集编码的图像分类方法。此外，本文提出了一种高斯聚集编码层来对由深度卷积神经网络提取的图像特征进行编码。学习固定大小的词典，以表示图像的特征以进行最终分类。此外，本文还根据空气系统的实际需求构建了空气污染监测系统。最后，本文设计了一个受控实验以验证本文提出的模型，使用数学统计来处理数据，并科学地分析统计结果。研究结果表明，本文构建的模型具有一定的效果。