水利水电工程设计

针对堤防结构沉降预测中存在的非线性建模困难与模型参数优化复杂等挑战，提出一种融合金豺优化算法(Golden Jackal Optimization,GJO)的CNN-GRU混合神经网络模型。该模型结合了卷积神经网络(Convolutional Neural Network,CNN)在提取空间局部特征方面的优势与门控循环单元(Gated Recurrent Unit,GRU)在处理时间序列数据中的高效性能，并通过GJO算法对模型关键超参数进行全局优化，以提升训练收敛速度与模型的泛化能力。基于实际堤防沉降监测数据开展的试验结果表明，与传统的BP神经网络和LSTM模型相比，所提出的GJO-CNN-GRU模型在平均绝对误差(MAE)、均方根误差(RMSE)及决定系数(R²)等评估指标上均表现出更优的拟合与预测性能。该研究不仅为堤防结构健康监测提供了一种智能高效的技术方案，也为复杂工程结构变形的智能预测研究提供了新的方法借鉴与思路参考。

To address the challenges of nonlinear modeling and complex parameter optimization in embankment settlement prediction, this study proposes a novel hybrid neural network model based on the Golden Jackal Optimization(GJO)algorithm,integrating Convolutional Neural Networks(CNN)and Gated Recurrent Units(GRU).The model leverages CNN's strength in extracting spatial local features and GRU's efficiency in handling temporal sequences. The GJO algorithm is employed to perform global optimization of key hyperparameters,thereby enhancing training convergence speed and model generalization. Experiments conducted on real-world embankment settlement monitoring data demonstrate that the proposed GJO-CNN-GRU model outperforms traditional models such as the Backpropagation(BP)neural network and Long ShortTerm Memory(LSTM)in terms of Mean Absolute Error(MAE),Root Mean Square Error(RMSE),and the coefficient of determination(R²). This research not only offers an intelligent and efficient solution for embankment structural health monitoring but also provides a promising methodological reference for deformation prediction in other complex engineering structures.