基于神经网络冷再生层最大剪应力预测

The Maximum Shear Stress Prediction Research of Cold Regeneration Based on Neural Network

目的针对乳化沥青冷再生路面内部剪应力过大易导致路面产生车辙等路面破坏问题,对其内部剪应力进行预测,减少此类病害,更好地选择路面结构参数,提高冷再生层内部抗剪强度.方法以乳化沥青冷再生层的厚度、模量,水泥稳定碎石的厚度、模量以及土基模量为输入参数,以冷再生层最大剪应力为输出参数,运用遗传算法对初始参数进行优化,运用灰色神经网络理论构建冷再生层最大剪应力预估模型;构建多元线性回归模型预测最大剪应力,对二者的预测能力进行分析.结果笔者建立的神经网络模型预测值与实测值拟合效果良好,最大误差仅为4.119 2%,能够进行准确预测.多元线性回归和灰色神经网络预测模型,都可用于冷再生层最大剪应力的预测,但灰色神经网络模型对冷再生层最大剪应力数据的预测结果较优.结论把灰色神经网络预测模型与沥青路面结构的设计联系起来,可以更好地控制乳化沥青冷再生路面的剪切破坏.

The internal shear stress is so much in the emulsified asphalt cold regeneration that it tends to cause problems, such as road cut. The internal shear stress was predicted to reduce such diseases, and the pavement structure parameters were better selected to improve the internal shear strength. The thickness and modulus of emulsified asphalt cold regeneration layer and cement stabilized macadam and modulus of soil base were used as the input parameters. The maximum shear stress of the cold regeneration layer was used as the output parameter. The genetic algorithm was used to optimize the initial parameters. The maximum shear stress prediction model of cold reclaimed layer was constructed by using gray neural network theory. The multiple linear regression model was constructed to predict the maximum shear stress and the prediction ability of the two model was analyzed. The neural network model has a good fitting effect between the predicted and measured values,and the maximum error is only 4. 1192%, so that it can do accurate prediction. Multivariate linear regression and gray neural network prediction model can be used to predict the maximum shear stress of cold regeneration layer, but the gray neural network model has better prediction results. The gray neural network prediction model is connected with the design of asphalt pavement structure, which can control the shear failure of emulsified asphalt cold regeneration road.