基于模糊C均值聚类的时序交通流量预测

Time Series Traffic Flow Prediction Based on Fuzzy C-Means Clustering

摘要: 为了提高交通流量预测精度，本文提出了一种模糊C均值聚类(FCM)与增强粒子群–灰狼混合优化算法(PSO-GWO)融合的交通流量预测模型。该模型通过FCM算法将交通流量划分为不同类别，设计PSO-GWO混合优化器动态调整轻梯度提升机(LightGBM)的参数，结合时序特征提取与选择，构建多模式预测模型，并应用于PEMS04单传感器交通数据集。实验结果表明，本文所提出的FCM-PSO-GWO-LightGBM模型平均绝对误差(MAE)为9.22，相比传统随机森林(RF)、极端梯度提升(XGBoost)、基准LightGBM模型都具有显著优势。

Abstract: To improve the accuracy of traffic flow prediction, a traffic flow forecasting model integrating Fuzzy C-Means (FCM) clustering with an enhanced Particle Swarm Optimization-Grey Wolf Optimizer (PSO-GWO) hybrid algorithm is proposed. In this model, the FCM algorithm is employed to categorize traffic flow data into distinct classes. A PSO-GWO hybrid optimizer is designed to dynamically adjust the parameters of the Light Gradient Boosting Machine (LightGBM). By incorporating temporal feature extraction and selection, a multi-mode prediction model is constructed and subsequently evaluated on the PEMS04 single-sensor traffic dataset. Experimental results demonstrate that the proposed FCM-PSO-GWO-LightGBM model achieves a Mean Absolute Error (MAE) of 9.22, exhibiting significant performance advantages over traditional models including Random Forest (RF), Extreme Gradient Boosting (XGBoost), and the baseline LightGBM model.