NARX神经网络模型在昆明市环境空气质量预测中的应用

Application of NARX Neural Network Model in Environmental Air Quality Prediction in Kunming

神经网络模型作为一种重要的手段被广泛应用于数学计算、物理建模、水文模拟、环境预测、人工智能等研究领域。为验证神经网络模型在高原山地城市环境空气质量预测中的作用,以昆明市环境空气自动监测站气象因子和污染物浓度数据为基础,构建NARX神经网络模型,对污染物浓度进行预测。结果表明,基于NARX神经网络建立的预测模型具有很强的非线性动态描述能力,能够对环境空气6参数做出较为准确的预测,其预测浓度相对误差显著低于CMAQ、NAQPMS空气质量数值模式以及LSTM统计模型预测结果。优化后的NARX神经网络对污染物浓度变化趋势的预测较其他几个模式更为敏感。

As an important means,neural network modeling is widely used in mathematical calculation,physical modeling,hydrological simulation,environmental prediction,artificial intelligence and other research fields.In order to verify the role of the neural network modeling in the prediction of urban environmental air quality in plateau mountains,NARX neural network model was constructed to predict pollutant concentration,based on the data of meteorological factors and pollutant concentration in Kunming environmental air automatic monitoring station.The results showed that the prediction models based on NARX neural network has strong nonlinear dynamic description ability and can make relatively accurate prediction of ambient air 6 parameters.The relative errors of the prediction concentration were significantly lower than that of the numerical model of air quality,CMAQ,NAQPMS and LSTM statistical model.The optimized NARX neural network models were more sensitive to the prediction of pollutant concentration change trend than other models.