神经网络在高速公路网综合评价中的应用

Application of Neural Network in Expressway Network Comprehensive Evaluation

应用多层次分析法和神经网络模型,对高速公路网评价指标及综合评价方法进行了研究。建立了适用于高速公路网评价的四大类评价指标;针对定量和定性指标特性分别采取不同的无量纲化函数对其进行无量纲化处理;通过神经网络建模思路及网络模型分析,结合高速公路网评价特点及评价指标个数,建立了三层前馈神经网络综合评价模型。经在江苏省第二轮高速公路网规划评价中进行实际应用分析,结果表明选取的评价指标科学合理,神经网络综合评价方法克服了其它综合评价方法的随意性缺陷。

This thesis applied the multilevel analysis and neutral network method, studied the evaluation indexes of expressway network and comprehensive evaluation model and set up four classes of evaluation index suitable for expressway network evaluation. In line with the quantitative and qualitative index, different dimensionless functions were adopted. Through neutral network modeling and network model analysis, taking into account of network evaluation characteristics and the number of evaluation index, a three level front feedback neutral network comprehensive evaluation model was developed. The model was used for the 2nd-round expressway network planning evaluation in Jiangsu province. The results show that the selected evaluation indexes are rational and that the neutral network comprehensive evaluation mode has overcome the defect of other method.