结构及其重要构件的长期沉降是影响地下空间综合交通枢纽施工安全的关键因素之一，对其开展监测与预测具有重要的工程意义。以深圳市黄木岗大型地下空间综合交通枢纽为研究背景，采用自动化监测系统，首先监测了12轴巨型斜柱施工后的119期长期沉降，分析其长期沉降规律，并评判其预警状态。监测结果表明，斜柱施工后，长期的沉降规律为先在安全状态内小幅波动，再增大至黄色预警状态，最后在安全状态与预警状态之间小幅波动。然后，利用109期监测数据分别构建斜柱沉降的反向传播（back propagation, BP）神经网络模型和遗传算法-BP（genetic algorithm-BP, GA-BP）神经网络模型，并预测之后10期的斜柱累计沉降量以对比验证两种模型效果。结果表明，相比BP神经网络模型，GA-BP神经网络模型的预测值与实测值更吻合。

The long-term structural settlement is one of the key factors on the construction safety of the large-scale integrated transportation hub underground, so it is necessary to monitor and forecast the structural settlement of the whole structure or its important components. Firstly, based on Huangmugang large-scale transportation hub underground in Shenzhen City, the long-term settlement (up to 119 periods after construction) of the 12-axis giant-inclined-column was monitored by the automatic monitoring system. The tendency of the column settlement was analyzed, and the early warning status of the column was evaluated. The monitoring results indicate that after the construction of the inclined column, the long-term settlement fluctuates within a narrow range in the safe state, then changes to the yellow alerting state, and finally has a slight fluctuation between the two states. Then, the back propagation (BP) neural network model and genetic algorithm-BP (GA-BP) neural network model for inclined column settlement were constructed using 109 periods monitoring data, and the cumulative settlement of inclined columns in the following 10 periods was predicted to compare and verify the effectiveness of the two models. The results indicate that compared to the BP neural network model, the predicted values of the GA-BP neural network model are more consistent with the measured values.