自适应神经模糊推理系统在交通污染物浓度预测中的应用

Adaptive Neural Fuzzy Inference System for Prediction of Traffic-related Pollution

城市交通带来的废气排放已经成为城市大气污染的主要来源之一。交通污染问题的成因和机理较为复杂,变化规律具有较强非线性和周期性特征。将自适应神经模糊推理系统(adaptive neuro fuzzy inference system,ANFIS)应用于交通污染物浓度时序数据预测时呈现出良好的泛化能力。本文以长沙市CO小时浓度数据为研究目标,通过分析CO浓度时序数据的自相关性、偏自相关性,以及交通流对CO浓度的时滞性影响,确定ANFIS预测模型的输入变量。结果表明,相较于传统的时间序列预测模型以及机器学习模型,ANFIS模型预测结果具有更高的精度,能够对交通环境污染进行预测及预警,为防止城市灾害性大气污染事件发生奠定理论研究基础并提供有效决策支持。

The exhaust emissions from urban traffic have become one of the main sources of urban air pollution.The cause and mechanism of traffic-related pollution are complicated, and the change law of traffic-related pollution has strong nonlinear and periodic characteristics. When adaptive neural fuzzy inference system(ANFIS) is applied for predicting time series data of traffic-related pollution,the good generalization ability is demonstrated. This paper is to study the CO hourly concentration data in Changsha.After the autocorrelation and partial autocorrelation of CO hourly concentration data and the time delay effect of traffic flow on CO hourly concentration have been analyzed,the input variables of the ANFIS are determined.The results indicate that the ANFIS has higher accuracy than traditional time series prediction model and machine learning model.ANFIS can provide prediction and early warning of traffic-related pollution,establishing the foundation of theoretical research and providing effective decision support for prevention of urban catastrophic air pollution events.