基于Transformer系列模型的高压线铁塔区域沉降预测方法

Transformer Series Models-based Prediction Settlement of High-voltage Line Tower Areas

现有高压线铁塔区域沉降值预测方法存在无法准确进行长期预测的问题。提出了一种基于Transformer系列模型的沉降值预测方法。通过建立时序InSAR气象与环境因素数据集,训练并测试了四种Transformer系列模型,以六种评价指标对模型进行了对比与分析。结果表明,Transformer系列模型具有较高的预测精度,其中Autoformer模型在高压线铁塔区域沉降值预测任务上表现最优异,具有较好的应用潜力。所提出的方法有助于及时发现并预防高压线铁塔区域因地表沉降出现的危险情况。

Currently prevailing predictive methods for settlement value of high-voltage line tower area cannot accurately predict long-term settlement,so this study proposed a Transformer series model-based settlement value predictive method.Four Transformer series models were trained and tested by establishing time-series InSAR meteorological and environmental factors datasets.And the models were compared and analyzed with six evaluation indexes.The results showed that the Transformer series models have high prediction accuracy,and the Autoformer model has the best performance in predicting settlement value of high-voltage line tower area.The proposed method can help in timely detection and prevention of the potential dangers in high-voltage line tower areas.