基于动态不等时距的边坡位移灰色预测模型分析

Grey Prediction Model Analysis of Slope Displacement Based on Dynamic Unequal Time Interval

对边坡进行安全监测,建立灰色GM(1,1)模型可以预测预警灾害的发生。边坡监测难以实现等时距观测,传统不等时距GM(1,1)模型预测精度低,无法满足长期预测要求。针对其缺点,重新分配在建模过程中系统数据累加和累减的权值分配,确立最佳权值分配,并在MATLAB实现程序化建模。针对仓上露天坑尾矿库边坡建立了传统、动态和修正权值的动态不等时距GM(1,1)模型,对比分析3个模型发现:动态模型较传统模型具有明显优势,预测精度提高,残差和残差率下降;修正权值的动态模型拟合度更好,预测结果更接近真实值,更能反映系统发展趋势。

The safety monitoring of slope and the establishment of GM（1,1） model can predict the occurrence of early warning disaster. Slope monitoring is difficult to achieve equal time interval observation, and the traditional unequal interval G M（1,1） model has low prediction accuracy, and can not meet the long-term prediction requirements. Based on the shortcomings, distribution in the process of modeling data accumulation and processing system of weights, establish the optimum weight distribution, and in the process of modeling of MATLAB. The bunker on the open pit slope of tailing is established, and the traditional dynamic dynamic correction weights of unequal time interval GM（1,1）model, comparative analysis of three models found that the dynamic model has obvious advantages compared with the traditional model, improve the prediction accuracy, residual and residual rate decreased; the right to amend the value of the dynamic model fit better, forecast the result is more close to the true value, can reflect the system development trend.