深厚软黏土地基固结沉降预测方法仿真研究  

Simulation Study on Prediction Method of Consolidation Settlement of Deep Soft Clay Foundation

作　　者：董海波[1] 包永堂 DONG Hai-bo;BAO Yong-tang(China University of Petroleum,Qingdao Shandong 266555,China;Qingdao Shandong University of Science and Technology,Qingdao Shandong 266590,China)

机构地区：[1]中国石油大学(华东),山东青岛266555 [2]山东科技大学计算机科学与工程学院,山东青岛266590

出　　处：《计算机仿真》2024年第12期317-320,530,共5页Computer Simulation

摘　　要：深厚软黏土地含水量较高且强度较低,传统地基沉降算法无法精准的其沉降量,对基建事业具有较大影响。为提高软土地基估计沉降预测的精确性、真实性与稳定性,基于遗传优化的小波神经网络算法构建了GAWN地基固结沉降预测模型。模型首先采用z-score算法对土地沉降中的环境采集数据进行去量纲处理,并使用Pearson算法分析数据相关性,以提高建模的时效性;接着将优化处理后的数据进行训练集与测试集划分,构造GAWN沉降预测模型的输入;然后利用GA优化模块对输入数据进行编码,并利用F值计算迭代出最优解至WNN网络中;最后利用遗传效应不断优化迭代权值与尺度值,直至完成GAWN模型对沉降量的预测输出。多组基线预测模型的仿真结果表明,在EFS自采沉降数据集上,较四类传统沉降预测算法模型相比,GAWN模型的R2指标提升了1.10%,RM及MA指标降低了41.84%和39.36%,即所构建的GAWN沉降预测模型的预测精确性有一定提升,且稳定性与真实性有较大提高。综上所述,所构建的GAWN地基固结沉降预测模型通过数据优化与神经元系统结合,有效的提高了地基沉降预测的精确性、真实性与稳定性,在沉降数据仿真领域中,具有较大的研究意义。Deep soft clay soil has high water content and low strength,and the traditional foundation settlement algorithm can not accurately calculate the settlement,which has a greater impact on infrastructure.In order to improve the accuracy,authenticity and stability of the estimated settlement prediction of soft soil foundation,the GAWN foundation consolidation settlement prediction model is established based on the wavelet neural network algorithm optimized by genetic algorithm in this paper.Firstly,the z-score algorithm is used to remove the dimension of the environmental data collected in the land subsidence,and the Pearson algorithm is used to analyze the data correlation to improve the timeliness of the model,and then the optimized data is divided into training set and test set to construct the input of the GAWN subsidence prediction model;Then the GA optimization module is used to encode the input data,and the F value is used to calculate and iterate the optimal solution to the WNN network;finally,the genetic effect is used to continuously optimize the iterative weight and scale value until the prediction output of the GAWN model on the settlement is completed.The simulation results of multiple baseline prediction models show that on the EFS self-mining subsidence data set,compared with the four traditional subsidence prediction algorithm models,the R2 index of the GAWN model increases by 1.10%,and the RM and MA indexes decrease by 41.84%and 39.36%,that is to say,the prediction accuracy of the GAWN subsidence prediction model constructed in this paper has a certain improvement.And that stability and the authenticity are greatly improve.To sum up,the GAWN foundation consolidation settlement prediction model constructed in this paper effectively improves the accuracy,authenticity and stability of foundation settlement prediction through the combination of data optimization and neural system,and has great research significance in the field of settlement data simulation.

关 键 词：沉降预测 遗传算法 小波神经网络 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]