基于BP神经网络的城市环境空气质量预测模型

本研究以监测所获得的数据为基础,运用BP神经网络算法原理,建立了城市环境空气质量预测模型,并对该模型的泛化能力进行了误差评价.结果表明：通过BP神经网络建立的空气质量预测模型具有较高的预测精度,预测结果的相对误差均在5％以内,能够很好地满足实际应用的需求.更重要的是,所建立的预测模型无需了解空气质量变化的内部机制,比传统的基于复杂数学模型的预测方法更为便捷,为环境保护部门可以提供更加可靠的决策依据.

Development of Predicted Model of Urban Environment Air Quality Using Artificial Neural Network

In the present investigation, by using the back propagation algorithm; the predicted model of urban environment air quality has been developed based on the obtained monitoring data. The generalization ability has been evaluated in term of relative error. It is found that the model can predict the air quality with high accuracy and reliability. More important, based on the established model, the inner rule of variety of air quality doesn＇t need to be understood. Compare with the traditional complex mathematical model, the ANN model is more convenient to provide the suitable advice for the department of environmental protection.