多维关联因素筛选条件下的堆积层滑坡体积预测研究

The Prediction of Landslide Volume of Accumulation Layer under the Condition of Multi-dimensional Correlation Factor Screening

为实现三峡库区堆积层滑坡体积的高精度预测,在滑坡特征参数统计的基础上,先开展体积影响因素的分布特征分析及相关性研究,以筛选出滑坡体积的重要影响因素;再利用多种优化算法和混沌理论实现极限学习机的参数优化和误差弱化,构建出滑坡体积在多维关联影响因素条件下的混沌优化极限学习机模型,以实现其预测研究.实例分析表明:三峡库区滑坡体积影响因素较多,受区域地质条件限制,各类影响因素的分布范围均较广,且波动性较强;同时,本研究中各类优化算法均能不同程度提高预测精度及运算速度,验证了各优化方法的有效性,且混沌理论能有效弱化预测误差,进一步提高了预测精度;另外,预测结果的平均相对误差为1.71%,可靠性验证结果的平均相对误差也仅为1.49%,两者相当.该预测模型不仅具有较高的预测精度,还具有较强的可靠性,适用于滑坡体积预测,为三峡库区滑坡规模研究提供了一种新的思路.

In order to realize the high precision prediction of the landslide volume in the accumulation layer of the Three Gorges Reservoir area,based on the statistics of the landslide characteristic parameters,this paper first carried out the distribution characteristic analysis and correlation research of the volume influencing factors,in order to screen out the important influencing factors of the landslide volume;then used a variety of optimization algorithms and chaos theory to realize the parameter optimization and error weakening of the limit learning machine,and constructed the in the limit learning machine model of chaos optimization for landslide volume under the condition of multi-dimensional correlation influencing factors,to realize its prediction research.The case study shows that there are many factors affecting the landslide volume in the Three Gorges Reservoir area,which are limited by the regional geological conditions.The distribution range of various factors is wide and the fluctuation is strong.At the same time,all kinds of optimization algorithms in this paper can improve the prediction accuracy and operation speed in varying degrees,which proves the effectiveness of the optimization method,and the chaos theory can effectively weaken the prediction error.In addition,the average relative error of the prediction results is 1.71%,and the average relative error of the reliability verification results is only 1.49%.Both of them are equivalent.The prediction model in this paper which not only has high prediction accuracy,but also has strong reliability,is suitable for landslide volume prediction,and provides a new idea for the study of landslide scale in the Three Gorges Reservoir area.