基于皮尔-遗传神经网络的高路堤沉降预测研究

为确保高速公路的建设质量，许多高速公路进行了路基沉降观测工作，以便对路堤填筑完成以后的沉降发展做出准确的预测，高路堤沉降预测模型较多，但均较难达到理想的精度。文章则依据高路堤填土施工期路基沉降实测资料，考虑到沉降单调增长的特殊性，根据位移分解原理，采用皮尔曲线提取沉降趋势，用基于免疫进化的新型遗传神经网络模型外推偏差，从而提出了一种高路堤沉降预测的新型智能方法。实际工程证明了所提智能预测方法的有效性和可行性。

Research of the Settlement Prediction for High Embankment Based on the Pearl Curve and Genetic Neural Network

To insure the construction quality of the highway, settlement observation of the road bed was carried out in many highway in order to accurately predict the development of settlement after the building of the embankment. There are many models to predict high embankment settlement, but the prediction using these models can not obtain the ideal accuracy.Based on the monitoring data of settlement during construction, considering the monotonously increasing character of the settlement, a new intelligent prediction method combining pearl curve and genetic neural network is proposed.In this method, based on the principles of displacement decomposition, the trend of settlement time series is extracted by pearl curve and the deviation of pearl is approximated by the genetic neural network whose arohitecture and algorithm parameters evolves simultaneously through combining modified BP algorithm and immunized evolutionary programming proposed by the author. The results of practice engineering show the validity and feasibility of the new method.