| 摘要点击次数: 73 全文下载次数: 0 查看全文 查看/发表评论 下载PDF阅读器 |
| 基金项目：中冶集团武汉勘察研究院有限公司科研项目“锤击桩施工数智化关键技术研究”(2024920008) |
|
| 摘要： |
| 深基坑变形预测对于保障周边环境稳定与工程安全至关重要。将遗传算法（GA）优化的人工神经网络（GA-BP）应用于深基坑变形预测，并将其与未经优化的BP神经网络、粒子群优化的BP神经网络（PSO-BP）的预测性能进行对比。通过分析某工程深基坑支护结构变形预测结果，采用拟合优度（R2）、均方根误差（RMSE）、平均绝对误差（MAE）及平均绝对百分比误差（MAPE）等评价指标，全面评估3种模型的性能。结果显示，GA和PSO优化的BP网络所有评价指标均优于未经优化的BP网络，其中GA-BP模型在提高预测精度方面表现突出，尽管个别指标略逊于PSO-BP，但整体性能验证了GA在提升深基坑变形预测准确性方面的有效性。未来可以探索GA和PSO两种算法的结合使用，以进一步提高模型的预测准确度和泛化能力，增强模型的解释性，为工程实践和决策提供更为可靠的支持。 |
| 关键词：深基坑支护结构 变形预测 人工神经网络 遗传算法 粒子群优化 GA-BP模型 PSO-BP模型 |
|
| Abstract: |
| The accurate prediction of deformation in deep foundation pits is essential for ensuring the stability of the surrounding environment and the safety of engineering structures. This study applies a genetic algorithm (GA)-optimized backpropagation neural network (GA-BP) for deformation prediction and compares its performance with that of a standard BP neural network and a particle swarm optimization (PSO)-optimized BP network (PSO-BP). By analyzing the predicted deformations of a support structure in an actual deep foundation pit project, the performance of the three models is comprehensively evaluated using metrics including the coefficient of determination (R2), root mean square error (RMSE), mean absolute error (MAE), and mean absolute percentage error (MAPE). The results indicate that both the GA-BP and PSO-BP models outperform the unoptimized BP model across all evaluation metrics. The GA-BP model demonstrates a significant improvement in prediction accuracy. Although it is slightly inferior to the PSO-BP model on certain individual metrics, its overall performance validates the effectiveness of the GA in enhancing the predictive accuracy for deep foundation pit deformations. Future research could explore the hybrid application of GA and PSO algorithms to further improve the prediction accuracy, generalization capability, and interpretability of the model, thereby providing more reliable support for engineering practice and decision-making. |
| Keywords:deep foundation pit support structure deformation prediction artificial neural network genetic algorithm particle swarm optimization GA-BP model PSO-BP model |
| 石 荔.遗传算法优化人工神经网络的深基坑支护结构变形预测[J].地质学刊,2026,50(2):235-242 |
|
|
苏公网安备32010202012246号