【摘要】 针对大面积吹填软土地基长期沉降难以预测的问题，使用静态预测模型和动态预测模型，根据现场沉降观测资料，预测某大面积超载预压工程的沉降量，按照预测步长和输入参数个数的不同设置4种预测任务，对比分析原始样本和插值处理后样本在4种预测任务中的模型可靠度。结果表明，动态模型预测可靠度优于静态预测模型，且动态模型单步预测时BP神经网络模型预测性能优于长短期记忆(long short-term memory, LSTM)网络模型，多步预测时LSTM模型预测性能优于BP神经网络模型。更多还原

【Abstract】 As to the issue that the long-term settlement of large-area hydraulic reclamation soft soil foundation is difficult to predict, the settlement of a large area overload preloading project is predicted based on the field settlement observation data using the static prediction model and dynamic prediction model. Four prediction tasks are set according to the different prediction steps and the number of input parameters. The reliability of original samples and interpolated samples in four prediction tasks model is compared and analyzed. The results show that the prediction reliability of dynamic model is better than that of static model. In the single-step prediction of dynamic model, the prediction performance of BP neural network model is better than that of long short-term memory(LSTM) network model. In multi-step prediction, the prediction performance of LSTM model is better than that of BP neural network model.更多还原