论文部分内容阅读
传统不等时距GM(1,1)模型预测结果往往会出现预测值的残差较大而且残差率的变化也大,还需再建立残差模型加以修正。针对传统不等时距GM(1,1)模型的不足,重新分配在数据累加和累减过程中时距的权重,通过研究与验算确定最佳的时距权重,建立了修正时距权重不等时距边坡位移的灰色预测模型,并在数据累减还原过程中依据残差率的变化趋势动态修正时距权重,使其预测结果与监测结果更为接近。该预测模型充分考虑了预测系统的时变性和灰色性,降低了预测系统的整体预测误差,提高了预测精度。实例分析表明:该预测模型拟合精度较高,预测结果正确可靠,能够反映边坡位移的发展趋势,对边坡位移的短、中期变化有较为理想的预测效果,具有一定的理论价值和工程实践意义。
When the traditional unequal-time GM (1,1) model prediction results tend to have large residuals of prediction values and large changes in the residual rate, a residual model needs to be established to correct the problem. Aiming at the shortcomings of the traditional unequal time-varying GM (1,1) model, the weight of time-distance in the process of data accumulation and accumulation is redistributed, and the optimal time-space weight is determined through research and checking. The gray prediction model of isochronous slope displacement is used. The weight of time-interval is dynamically modified according to the trend of the residual rate during the process of data reduction and reduction, so that the prediction result is closer to the monitoring result. The prediction model fully considers the time-varying and gray-level of the prediction system, reduces the overall prediction error of the prediction system and improves the prediction accuracy. The case study shows that the prediction model has a high fitting accuracy and a correct and reliable prediction result, which can reflect the development trend of slope displacement and has an ideal prediction effect on short-term and mid-term changes of slope displacement, and has certain theoretical value and engineering Practical significance.