基于遗传-小波神经网络和时空特性的交通数据融合

Traffic Data Fusion Based on Genetic Wavelet Neural Network and Spatio-temporal Characteristics

由于时间序列的交通数据重点关注断面交通流量变化,而空间序列的交通数据主要关注路网交通流分布。在综合考虑二者特性的前提下利用遗传算法的群体搜索技术及小波神经网络的较强学习能力,提出了基于遗传-小波神经网络(GAWNN)的交通数据融合模型。使用MAE、MRE和MSE三个指标对交通数据序列进行优劣对比分析。经实例验证,考虑时空特性的交通数据其数据质量优于单一的时间序列和空间序列。

Due to the traffic data of tiie time series, the traffic data of tiie spatial sequence are mainly concernedwith the distribution of traffic flow. Under the premise of the comprehensive consideration of the two characteristicsof the use of genetic algorithm for group search technology and wavelet neural netwbased on genetic wavelet neural network ＂ GA-WNN） is proposed. Using MAE,MRE and MSE to compared and analysis of traffic data sequence. The results show that the quality of traffic data with time and space is better thanthat of single time series and space series.