基于MIV-BP神经网络的公交场站污染物暴露水平测评

为了研究城市居民在公交场站污染物暴露水平下的影响程度, 分析交通、公交场站、气象条件及周边环境对污染物浓度的影响, 本文通过测量早晚高峰以及平峰期间的PM2.5浓度、CO浓度、风速、温度、车流量等10个因素, 采用平均影响值（MIV）与BP（back propagation）神经网络相结合的方法, 确定公交场站污染物暴露水平的主要影响因素, 并在此基础上建立MIV-BP神经网络测评模型。评估结果表明：公交场站PM2.5的暴露浓度与公交停靠站类型、车流量、小车流量、大车流量、风速、湿度、降雨量有关；CO的暴露浓度与瞬时停靠车辆数、公交停靠站类型、湿度、风速、温度、车流量有关；改进后的MIV-BP神经网络模型较BP神经网络具有更高的预测精度和准确度, 可有效对公交场站污染物暴露水平进行测评。

Evaluation of Pollutant Exposure Level at Bus Stop Based on MIV-BP Neural Network

In order to study the influence degree of the pollutants exposure levels on urban residents at the bus terminal and analyze the influence of traffic, the bus terminal, meteorological condition, surroundings on pollutant concentration, 10 factors including PM2.5 concentration, CO concentration, wind speed, temperature, vehicle flow etc. during peak hours in the morning and evening and peak periods were measured. Then the main influencing factors of pollutant exposure levels in bus stations were determined by using the combination of mean impact value (MIV) and BP (back propagation) neural network, and on this basis MIV-BP neural network assessment model was established. The final evaluation results indicate that the exposure concentration of PM2.5 at the bus station is related to the type of bus stop, traffic flow, car flow, traffic flow, wind speed, humidity, rainfall; CO exposure concentration is related to the number of instantaneously parked vehicles, the type of bus stop, humidity, wind speed, temperature, traffic flow; the improved MIV-BP neural network model has higher prediction accuracy and degree of accuracy than the original BP neural network, and can effectively evaluate the pollution exposure level of bus stop.