-
摘要: 针对道路表面的坡面水流受降雨和坡面粗糙程度的影响, 是一个高度非线性空间分布的过程, 一般模型很难精确描述。建立了基于人工神经网络的道路表面水膜厚度预测模型, 以降雨强度、坡度、坡长和坡面的粗糙程度为输入层, 水膜厚度为输出层, 隐含层为6个神经元, 通过试验数据的训练, 确定了网络的权重和阈值。应用结果表明该模型预测的水膜厚度与测量值的相关系数为0 98, 误差平方和为3 08, 这说明该模型用于道路表面水膜厚度预估是可行的。Abstract: Rainfall and slope rough degree have great effect on rain water depth on road surface, and their relation is highly nonlinear, it can not be described by simple models. This paper used artificial neural network to establish the prediction model of water film thickness. Input layers of the model were rainfall intensity, gradient, slope length and contexture depth of road, output layer was rain water depth, there were 6 neural cells in hidden layer. The weights and thresholds of the model were attained through training neural network with testing data.Applied results show that the relative coefficient of predictive values and measure values is 0.98, error is 3.08. The result indicates that the predictive model is reasonable.
-
表 1 神经网络模型的训练结果
Table 1. Trained result of neural network model
序号 模型 误差平方和 相关系数 1 ANN (4, 4, 1) 0.064 5 0.982 0 2 ANN (4, 5, 1) 0.048 7 0.986 4 3 ANN (4, 6, 1) 0.026 6 0.992 6 4 ANN (4, 7, 1) 0.026 3 0.992 7 
下载: 导出CSV
表 2 检验样本
Table 2. Inspection samples
序号 坡面构造深度/mm 坡度 降雨强度/ (mm·min-1) 坡长/m 水膜厚度/mm 1 1.5 0.020 3.6 5 4.0 2 1.5 0.020 5.4 7 8.0 3 1.5 0.020 6.3 4 7.0 4 1.5 0.050 5.3 3 4.0 5 1.5 0.050 6.3 1 2.0 6 1.5 0.080 3.7 1 1.0 7 1.5 0.080 5.4 7 5.5 8 1.5 0.080 6.5 7 6.0 9 0.7 0.025 5.3 4 3.0 10 0.7 0.025 6.3 6 4.0 11 0.7 0.025 7.4 6 7.0 12 0.7 0.055 5.5 5 2.0 13 0.7 0.055 7.3 4 3.5 14 0.7 0.075 4.6 7 2.0 15 0.7 0.075 6.2 3 2.0 16 0.7 0.075 7.4 5 3.5 17 0.9 0.020 2.4 1 1.0 18 0.9 0.050 1.2 4 1.0 19 0.9 0.050 3.0 3 1.5 20 0.9 0.080 3.3 1 0.5
下载: 导出CSV
-
[1] 单成昕. 汽车理论基础[M]. 北京: 人民交通出版社, 1990. [2] 庄继德. 现代汽车轮胎技术[M]. 北京: 北京理工大学出版社, 2001. [3] 芮孝芳. 产汇流理论[M]. 北京: 水利电力出版社, 1995. [4] 赵振宇, 徐用懋. 模糊理论和神经网络的基础和应用[M]. 北京: 清华大学出版社, 1996. [5] 闻新, 周露, 王力丹, 等. Matlab神经网络应用设计[M]. 北京: 科学出版社, 2000. [6] 文康. 地表径流过程的数学模拟[M]. 北京: 水利电力出版社, 1995. [7] 芮孝芳. 径流形成原理[M]. 南京: 河海大学出版社, 1991. [8] 杨荣富, 丁晶, 刘国东. 具有水文基础的人工神经网络初探[J]. 水利学报, 1998, 33 (8): 23—27. https://www.cnki.com.cn/Article/CJFDTOTAL-SLXB808.004.htmYANG Rong-fu, DING Jing, LIU Guo-dong. Preliminary study on the artificial neural network based on hydrological property[J]. Journal of Hydraulic Engineering, 1998, 33(8): 23—27. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-SLXB808.004.htm -
点击查看大图
图(1) / 表(2)
计量
- 文章访问数: 250
- HTML全文浏览量: 87
- PDF下载量: 352
- 被引次数: 0
