面向道路交通事故鉴定的水泥路面性能评价方法  

作　　者：李璇 马兆有[1,2] 寿任祯 鲍瀚涛 赵冬 魏然[1,2] 

机构地区：[1]公安部交通管理科学研究所 [2]道路交通安全公安部重点实验室

出　　处：《道路交通科学技术》2022年第5期18-22,共5页Road Traffic Science & Technology

摘　　要：针对当前道路交通事故鉴定中水泥路面性能评价的需求,本文采用反映路面使用性能和路面结构状况的7项优化指标和分级标准综合评价水泥路面的性能。由于离散Hopfield神经网络具有构建程序简单、训练样本少和客观性强等特点,所以采用MATLAB构建的离散Hopfield神经网络评价试验路段水泥路面的性能。将理想的水泥路面性能等级评价指标矩阵和试验路段6处待分类的水泥路面性能评价指标矩阵输入构建的神经网络中进行仿真学习,得到试验路段水泥路面性能的评价等级,并将评价结果与模糊复合物元法、非线性模糊法作对比,证明了离散Hopfield神经网络评价方法具有可靠性。In view of current demand for performance evaluation of cement pavement in the identification of road traffic accidents,the paper adopts seven optimizing indexes and grading standards reflecting functional performance and structure of the pavement to comprehensively evaluate the performance of the cement pavement.Because the discrete Hopfield neural network(DHNN)has the characteristics of simple construction procedure,less training samples and strong objectivity,the DHNN constructed by MATLAB is used to evaluate the performance of cement pavement in the test section.When inputting the ideal cement pavement performance grading evaluation index matrix and the to-be-classified pavement performance evaluation index matrix of six places in the test section into the constructed neural network to do simulation learning,the evaluation grades of cement pavement performance in the test section are thus obtained.Finally,by comparing the evaluation results with the fuzzy complex matter element method and the nonlinear fuzzy method,the reliability of the evaluation method of the discrete Hopfield neural network is proved.

关 键 词：水泥路面性能 离散HOPFIELD神经网络 MATLAB 评价指标矩阵 仿真学习 

分 类 号：U41[交通运输工程—道路与铁道工程]