摘要
针对目前短时交通流预测存在的问题 ,提出一种基于非参数回归的短时交通流量预测与事件检测综合算法框架并对框架中的每个步骤进行详细说明。为了进一步提高上述算法的精度与速度 ,对传统的非参数回归算法做了两方面改进 :基于密集度的变 K搜索算法与基于动态聚类和散列函数的历史数据组织方式。通过这些改进 ,使得上述基于非参数回归的算法成为一种“无参数”、可移植、高预测精度的实时预测算法 ,并能有效地用于短时交通流的预测问题中。现场实验充分表明该算法完全满足实时交通流预测的需要。
Although short-term traffic flow forecasting algorithm is widely used today, some questions still exist. To address these issues, an KNN-NPRCK Nearest Neighbors-Non-Parametric Regression) based integrated algorithm which can make traffic flow forecasting and traffic event detection at the same time is proposed. Then in order to improve the accuracy and computing speed of the proposed algorithm, authors put forward two improvements which are improved in variable K search method based on 'dense degree' and advanced data structures based on dynamic cluster method and hash-function transformation. Field test fully proves two improvements, which the proposed algorithm can adequately meet real-time system requirements and accuracy requirements.
出处
《中国公路学报》
EI
CAS
CSCD
北大核心
2003年第1期82-86,共5页
China Journal of Highway and Transport
