基于遗传神经网络的高速公路纵坡运行速度预测方法研究

车辆运行速度预测取决于多因素、非线性函数关系的建立,预测模型建立的准确与否取决于各个影响因素之间的相互作用的特性。将遗传算法与神经网络有机结合起来,以高速公路上的实测运行速度为基础,建立遗传神经网络训练和检验样本集,利用Matlab7.04的神经网络工具箱和遗传算法工具箱的函数,完成程序的编写,建立基于遗传算法的高速公路纵坡路段运行速度(V85)的神经网络预测模型,并将预测结果与实测数据进行比较。结果表明:所用遗传神经网络模型可靠,预测精度高,对我国采用运行速度的路线设计方法和线形质量评价有较高的参考价值。

Study of Expressway Longitudinal Slope Running Speed Forecasting Method Based on Genetic Neural Network

The forecast of vehicle running speed depends on the establishment of multi-factor and nonlinear function relationship. Whether or not accurate to establish the forecasting model depends on the characteristics of the interaction among various influencing factors. The article organically combines the genetic algorithms with the neural network, and based on the measured running speed of expressway, establishes the training and inspection sample set of the genetic neural network. The program preparation is completed by the functions of Matlab7.04 neural network toolbox and genetic algorithm toolbox. The establishment of the neural network forecasting model for expressway longitudinal slope running speed (V85) is based on the genetic algorithm. The forecasting results are compared with the measured data. The result shows that the genetic neural network model is reliable and its forecasting accuracy is high, which has the higher referring value for the line design method of running speed and the evaluation of alignment quality.