| 基于神经网络和多标度特征分析的古滑坡变形预测及趋势评价 |
| |
| 引用本文： | 田 倩,吴 健,赵 东.基于神经网络和多标度特征分析的古滑坡变形预测及趋势评价[J].大地测量与地球动力学,2022,42(10):1056-1062. |
| |
| 作者姓名： | 田 倩 吴 健 赵 东 |
| |
| 摘 要： | 提出一种新的古滑坡变形预测方法。首先结合集合经验模态分解(EEMD)和奇异值分解(SVD)对古滑坡变形数据进行分解，然后利用分项组合神经网络预测古滑坡复活区的变形，最后利用多重分形消除趋势波动分析(MF-DFA)进行古滑坡多标度趋势评价。以王家坡滑坡为例分析本文方法的有效性。结果表明，组合分解模型EEMD-SVD较单项分解模型具有更强的数据分解能力，可有效实现滑坡变形数据的信息分解；基于神经网络的分项组合预测模型适用于滑坡变形预测，所得预测结果的相对误差基本在2%左右，预测精度较高，且外推预测显示滑坡变形仍会进一步增加，增加速率为1.23~1.36 mm/周期；MF-DFA模型的多标度特征分析结果显示，滑坡变形具有多重分形特征，变形有进一步增加的趋势，这与预测结果较为一致，可佐证前述预测结果的准确性。
|
| 关 键 词： | 古滑坡 数据分解 神经网络 MF-DFA模型 趋势预测 |
Deformation Prediction and Trend Evaluation of Paleo-Landslide Based on Neural Network and Multi-Scale Feature Analysis |
| |
| TIAN Qian,WU Jian,ZHAO Dong.Deformation Prediction and Trend Evaluation of Paleo-Landslide Based on Neural Network and Multi-Scale Feature Analysis[J].Journal of Geodesy and Geodynamics,2022,42(10):1056-1062. |
| |
| Authors: | TIAN Qian WU Jian ZHAO Dong |
| |
| Abstract: | A new method for paleo-landslide deformation prediction is proposed. Firstly, the paleo-landslide deformation data are decomposed by ensemble empirical mode decomposition(EEMD) and singular value decomposition(SVD), and then the deformation of the resurrected area of the paleo-landslide is predicted by the component combined neural network. Finally, the multi-fractal detrended fluctuation analysis(MF-DFA) is used to evaluate the multi-scale trend of the ancient landslide. Taking Wangjiapo landslide as an example, the effectiveness of the proposed method is analyzed.The results show that the combined decomposition model EEMD-SVD has stronger data decomposition ability than the single decomposition model which can effectively realize the information decomposition of landslide deformation data. The sub item combination prediction model based on neural network is suitable for landslide deformation prediction. The relative error of the prediction results is mostly about 2%, with high prediction accuracy. The extrapolation prediction shows that the landslide deformation will increase further, with an increase rate of 1.23-1.36 mm/cycle.Through the multi-scale feature analysis of MF-DFA model, we concluded that the landslide deformation has multifractal characteristics, and the deformation tends to increase, which is consistent with the prediction results, and proves the accuracy of the above prediction results. |
| |
| Keywords: | paleo-landslide data decomposition neural network MF-DFA model trend prediction |
|
| 点击此处可从《大地测量与地球动力学》浏览原始摘要信息 |
| 点击此处可从《大地测量与地球动力学》下载免费的PDF全文 |
