基于数据挖掘的固定型交通检测器配置优化

结合固定型交通检测器空间配置的4条原则和配置密度优化步骤,提出基于数据挖掘技术的固定型交通检测器配置优化方法.设计6种高速公路出口匝道的固定型交通检测器配置密度方案作为实例研究对象,运用数据挖掘技术的时间序列指数平滑方法、ARIMA方法和神经网络方法分别建立高速公路出口匝道小时交通量Winters预测模型、ARIMA预测模型及神经网络预测模型.采用网格搜索技术确定Winters模型参数,设计一种比传统ARIMA模型参数估计方法更精确的算法程序,来估计ARIMA模型参数,采用3项误差指标评价模型预测效果.根据预测结果及高速公路事件管理交通参数精度要求确定可行方案及最佳方案.实例研究表明,在保证满足ITS 对交通参数精度要求的同时,通过数据挖掘技术降低了交通流信息采集固定型检测器的配置密度及成本.

Application of Data Mining Technology to the Optimization of Fixed Traffic Detector Deployment

According to four rules of fixed traffic detector deployment and optimization of deployment density, this paper presents a method for the optimization of fixed traffic detector deployment based on data mining technology. Six projects for freeway off-ramp of fixed traffic detector deployment are designed. Winters, ARIMA and feed-forward neural network hourly traffic volumes forecasting model for freeway off-ramp are established by applying data mining time series technology, namely exponential smoothing, ARIMA and feed-forward neural network method. An algorithm of finding ARIMA parameters is established by applying grid searching techniques to finding exponential smoothing parameters, which is more accurate than traditional method. Three error indexes is applied to evaluating the accuracy of the forecast model. The results indicate that data mining technology is a simple and feasible method in the application of optimization of fixed traffic detector deployment, which decreases the detector density and cost.