基于改进GWO-BP神经网络模型的箱涵沉降预测
2021-11-08杨阳赵青戚蓝黎启贤王毓杰邹爽
杨阳 赵青 戚蓝 黎启贤 王毓杰 邹爽



摘 要:箱涵受外部荷载等多方面影响,会出现一定程度的不均匀沉降,可能会对箱涵结构造成破坏,因此箱涵地基沉降预测十分重要。引入灰狼算法(GWO)对BP神经网络的权值和阈值进行寻优,建立了基于改进的GWO-BP预测模型,对箱涵的沉降值进行预测。将该预测模型应用于南水北调工程天津某标段的箱涵沉降预测,并将预测值与实测值进行对比,相对误差在5%以下。通过与未改进的灰狼算法优化BP神经网络模型、BP模型进行对比,结果表明改进的灰狼算法优化BP神经网络预测模型具有更好的寻优能力与寻优精度,能够有效地对箱涵沉降值进行预测。
关键词:箱涵;沉降预测;灰狼算法;BP神经网络;权值和阈值
中图分类号:U449.82 文献标志码:A
doi:10.3969/j.issn.1000-1379.2021.10.029
引用格式:杨阳,赵青,戚蓝,等.基于改进GWO-BP神经网络模型的箱涵沉降预测[J].人民黄河,2021,43(10):150-153.
Abstract: Due to the influence of external load and other aspects, the box culvert will have a certain degree of uneven settlement, which may cause damage to the box culvert structure. Therefore, the prediction of box culvert foundation settlement is very important. The gray wolf algorithm (GWO) was introduced to optimize the weight and threshold of BP neural network, and the improved GWO-BP prediction model was established to predict the settlement value of box culvert. The prediction model was applied to the settlement prediction of a box culvert in Tianjin section of South-to-North Water Transfer Project, and the relative error was less than 5%. The results show that the Improved Grey Wolf algorithm has better optimization ability and accuracy, and can effectively predict the settlement value of box culvert.
Key words: box culvert; settlement prediction; grey wolf algorithm; BP neural network; weight and threshold
在南水北調工程这类大型输水工程中,钢筋混凝土箱涵被广泛应用,箱涵结构的安全是保证整个输水工程正常运行的关键。箱涵在地下水开采、上部建筑荷载过大等因素的共同作用下,会出现一定程度的地基沉降。地基沉降过大会对箱涵结构造成破坏,影响输水效率,增加人工维修成本,因此箱涵地基沉降值的预测非常重要。
近年来,沉降预测已有许多研究成果。王鹏等[1]利用回归分析法预测基础不均匀沉降,但该方法较难反映复杂的非线性关系,故预测精度较低。陈继光[2]采用SVM模型对建筑物沉降进行预测,但该模型对样本数量要求过高,故其实际应用有局限性。……
