基于BP神经网络的沥青老化预测系统研究

发布时间：2018-01-10 18:05

本文关键词：基于BP神经网络的沥青老化预测系统研究　出处：《重庆交通大学》2014年硕士论文　论文类型：学位论文

【摘要】：沥青材料在行车荷载和自然因素（氧气、水、阳光等）的长时间作用下非常容易发生老化，导致沥青路面性能下降。影响沥青老化的因素主要包括行车荷载作用、沥青氧化、雨水作用、阳光中紫外线辐射、时间的长短等，其过程具有非线性、混沌性、长期记忆性等特点。本文基于历史数据，运用BP神经网络模型，对沥青老化指标的变化进行研究，建立沥青老化BP网络模型，对各影响因素下沥青的老化性能进行预测。首先运用了试验分析的手段分别分析了3个主要影响沥青老化的因素-温度、水、紫外线对沥青的老化作用，，然后经过综合分析3因素对老化的影响程度得出紫外线温度水，其次收集了国内几条主要沥青路面沥青的实际老化指标，通过对这些老化指标的处理作者能够得到不同地区不同使用年限沥青25℃针入度、15℃延度、软化点等数据。第三，研究BP神经网络的原理，并推导具体理论算法，了解BP神经网络的训练过程，为BP神经网络的结构设计奠定理论基础。第四，采用多因素(最高气温、最低气温、年平均降雨量、年平均日照时间、使用年限等因素) BP神经网络对沥青三大指标进行预测，得出在各种因素作用下下沥青三大指标的变化情况，得到通过MATLAB工具箱实现的适用于指标预测的具体网络。最后，通过对重庆地区的高速路上的回收沥青的性能与网络预测的实际性能进行对比，发现具有良好的规律性，验证了BP神经网络预测的准确性。大量可重复的BP网络实验结果显示，运用BP神经网络模型可以对沥青老化情况进行较高精度的预测，这证明了本文所采用的方法和所建立的模型是可行的和有效的，所以，在实验室很快得到沥青路面不同的老化数据具有了实现的可能，为相关研究提供参考。
[Abstract]:Asphalt materials in traffic load and natural factors (oxygen, water, sun, etc.) the long time effect is very prone to aging, resulting in a decline in the performance of asphalt pavement. The influence factors including the effect of asphalt aging, the traffic load of asphalt oxidation, rain, sunlight in the ultraviolet radiation, such as the length of time, the process is nonlinear the characteristics, chaos, long memory. Based on the historical data, using the BP neural network model, to study the changes of asphalt aging index, establish the BP network model of asphalt aging, aging properties and influencing factors of asphalt prediction.
First of all by means of experimental analysis respectively to analyze the 3 main factors influencing asphalt aging temperature, water, the aging effect of ultraviolet on the asphalt, and then after a comprehensive analysis of 3 factors on the impact of aging that ultraviolet temperature of water, the actual second collects the domestic several major asphalt pavement asphalt aging index, by the author the aging index can be different in different regions of the asphalt use age 25 C 15 C penetration, ductility, softening point data. Third, the principle of BP neural network, and derive the specific theoretical algorithm, understand the training process of BP neural network, which lays the theoretical foundation for the structure design of BP neural network. Fourth. The multi factors (maximum temperature, minimum temperature, average annual rainfall, the average annual sunshine time, age and other factors) of BP neural network on the three index of asphalt for prediction, obtained in the Changes of factors under the three indices of asphalt, concrete network application is realized through MATLAB toolbox to index prediction. Finally, by comparing the actual performance and the prediction of highway asphalt recycling in Chongqing area. The discovery has good regularity, to verify the accuracy of BP neural network forecast.
The experimental results of a large number of repetitive BP network show that the application of BP neural network model can predict with high accuracy of asphalt aging, which proves that the method used in this paper and the model is feasible and effective, therefore, in the laboratory soon got different asphalt pavement aging data has the potential to achieve and provide a reference for related research.

【学位授予单位】：重庆交通大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：U414

【参考文献】