摘要
为提升交通事故时间序列预测精度,建立一个基于相关向量机(RVM)的交通事故时序序列预测模型。结合RVM的建模与求解思想,建立交通事故时间序列预测函数关系式;设计交通时序参数预测模型实现流程,并选取均方根误差(RMSE)、模型训练时间等作为评价指标;以我国交通事故数、万车死亡率、10万人口死亡率为例,验证所建模型的有效性。实例验证表明:所建模型对不同的交通事故时间序列指标预测效果良好,预测精度高于灰色预测、自回归移动平均模型、支持向量机(SVM)等经典模型。
A model was built for enhancing accuracy of traffic accident time series prediction based on the RVM model.Firstly, a prediction function of traffic accident time series was established on the basis of the thinking of RVM about modelling and solving. Secondly, steps of prediction method of traffic accident time series were designed, and root-mean-square error（RMSE） as well as model training time were taken as evaluating indicators. Finally, the effectiveness of the prediction method was confirmed by the data on traffic accidents, traffic accident fatalities of per 1 000 cars, traffic accident fatalities of per 100 000 people in China in a period of som 60 years. Results indicate that the prediction method has a good prediction effect in traffic accidenttime series, and its forecast accuracy is higher than those of grey prediction, autoregressive moving average model, support vector machine（SVM） or other classical forecasting models.
出处
《中国安全科学学报》
CAS
CSCD
北大核心
2016年第6期52-56,共5页
China Safety Science Journal
基金
国家科技支撑计划课题(2014BAG05B01)
关键词
交通事故
时间序列
相关向量机(RVM)
交通运输工程
人工智能
traffic accident
time series
relevance vector machine（RVM）
transport engineering
artificial intelligence
