融合核极限学习机与PSR的混沌交通流预测
Chaotic traffic flow prediction combining kernel extreme learning machine with phase space reconstruction
夏晶晶 1陈振2
作者信息
- 1. 河南牧业经济学院信息工程学院,河南郑州 450044
- 2. 河南农业大学信息与管理科学学院,河南郑州 450002
- 折叠
摘要
传统短时交通流预测精度低、稳定性差,提出一种结合改进蝴蝶算法优化核极限学习机与相空间重构的短时交通流预测模型.结合量子自适应种群初始化、邻域扰动和惯性权重对蝴蝶算法改进,利用改进蝴蝶算法对核极限学习机超参寻优.利用混沌理论确定样本时序最佳延迟时间和嵌入维数,利用PSR对样本重构,利用优化核极限学习机建立短时混沌交通流预测模型.采用郑州市某主干路口车流实测数据进行实证分析,其结果表明,改进模型能够有效降低预测误差,实现混沌交通流实时准确预测.
Abstract
Aiming at the low accuracy and poor stability of traditional short-term traffic flow prediction,a short-term traffic flow prediction model was proposed,in which the phase space reconstruction and the improved butterfly algorithm were combined to optimize kernel extreme learning machine.Combined with quantum adaptive initialization,neighborhood disturbance and inertia weight,the butterfly algorithm was improved,and the improved algorithm was used to optimize the hyperparameters of kernel extreme learning machine.The chaos theory was used to calculate the optimal delay time and embedding dimension of the sample time series,and the phase space reconstruction method was used to reconstruct samples and then build a new short-term traffic flow prediction model.The empirical analysis was carried out using the measured data of a Zhengzhou intersection in the city.The results show that,the improved mode can reduce the prediction error and realize real-time and accurate prediction of chaotic traffic flow.
关键词
相空间重构/核极限学习机/交通流预测/蝴蝶优化算法/量子自适应/邻域扰动/惯性权重Key words
phases space reconstruction/kernel extreme learning machine/traffic flow prediction/butterfly optimization algo-rithm/quantum adaptive/neighborhood vibration/inertia weight引用本文复制引用
基金项目
河南省林业科技发展基金(KJZXA2019042)
出版年
2024
NETL
NSTL
万方数据