摘要
沥青路面抗滑性能预测是制定养护计划及确定养护方案的重要依据。以重庆某高速公路路面抗滑性能预测为例,建立基于遗传神经网络的路面抗滑性能预测模型,并与单一的神经网络、遗传算法及回归模型进行对比,验证了遗传神经网络用于沥青路面抗滑性能预测的适用性和可靠度。研究结果表明:回归模型对高度非线性问题拟合度较差,预测精度明显不如遗传算法或神经网络算法;相较单一的神经网络模型或遗传算法,组合预测模型能极大提升预测精度,但运算效率明显不如单一的神经网络模型。因此,遗传神经网络组合预测模型在路面抗滑性能预测的精准上更具适用性。
The prediction of anti sliding performance of asphalt pavement is an important basis for formulating maintenance plan and determining maintenance scheme.Taking the anti sliding performance prediction of one expressway in Chongqing as an example,this paper establishes a pavement anti sliding performance prediction model based on genetic neural network,and compares it with a single neural network,genetic algorithm and regression model to verify the applicability and reliability of genetic neural network in the prediction of asphalt pavement anti sliding performance.The results show that the regression model has poor fitting for highly nonlinear problems,and the prediction accuracy is obviously lower than that of genetic algorithm or neural network algorithm;Compared with the single neural network model or genetic algorithm,the combined prediction model can greatly improve the prediction accuracy,but the operation efficiency is obviously lower than the single neural network model.Therefore,the combined prediction model of genetic neural network is more suitable for the accuracy of pavement anti-skid performance prediction.
作者
张含伟
苗超杰
ZHANG Hanwei;MIAO Chaojie(Sichuan Communications Investment Design Consulting and Research Institute Co.,Ltd.,Chengdu,610041;Chongqing Yuxiang Double-line Expressway Co.,Ltd.,Chongqing 401346)
出处
《公路交通技术》
2022年第4期12-18,共7页
Technology of Highway and Transport
基金
云南省交通厅科技项目(云交科教(2016)140(B))
贵州省交通厅科技项目(2020-123-002)。
关键词
沥青路面
抗滑性能
遗传算法
神经网络
组合预测模型
asphalt pavement
anti-slip performance
genetic algorithms
neural network
combination prediction model
