摘要
鉴于传统GM(1,1)模型不能考虑反映外部环境变化对系统变化趋势的影响以及GM(1,N)多变量模型在模型结构上存在不足,引入OGM(1,N)模型并针对其背景值取值存在的误差采用自适应变异粒子群算法对背景值进行优化。对排土场3个监测点(G7,G8,G9)进行灰色关联分析,将改进后的模型应用于界牌岭矿山排土场边坡变形预测工作,并与GM(1,1)模型、Verhulst模型和非线性回归模型结果作对比。排土场3个监测点的预测结果:改进的OGM(1,N)模型平均相对百分误差分别为0.16%,2.26%,10.562%;GM(1,1)模型平均相对百分误差分别为16.57%,18.07%,19.095%;Verhulst模型平均相对百分误差分别为4.52%,2.34%,29.809%;非线性回归模型平均相对百分误差分别为11.44%,8.45%,11.621%,改进的OGM(1,N)模型相较其他3种模型有较高的精度。对优化前后的OGM(1,N)模型结果进行了对比,平均相对模拟百分误差约减小到优化前的1/3。综合3个监测点的预测数据效果,改进的OGM(1,N)模型在边坡变形预测中具有较好的适用性与有效性。
Considering the traditional GM(1,1)model in slope deformation prediction can not reflect the impact of external environmental changes on the trend of system change and inherent structural shortcomings in GM(1,N)model,OGM(1,N)model was introduced and its background value is optimized by adaptive particle mutation swarm optimization according to the error of background value in gray system.Three monitoring points(G7,G8,G9)on the waste dump were analyzed by grey relational analysis theory.Then the improved OGM(1,N)model was used to predict the waste dump deformation in Jiepailin mine.Compared the results with GM(1,1)model,Verhulst model and Nonlinear regression,the results of three monitoring points showed that the improved OGM(1,N)model is more precise than the other three models.In detail,the average relative percentage error of improved OGM(1,N)model is 0.16%,2.26%,10.562%,respectively.The average relative percentage error of GM(1,1)model is 16.57%,18.07%,19.095%,respectively;the average relative percentage error of Verhulst model is 4.52%,2.34%,29.809%,respectively,the average relative percentage error of nonlinear model is 11.44%,8.45%,11.621%,respectively.Comparing the result of the OGM(1,N)model before and after optimization,the average relative simulation percentage error after optimization was reduced to about more than 1/3 of that before optimization.According to the prediction results of three monitoring points,the improved OGM(1,N)model is adaptive and effective in slope deformation prediction.
作者
唐超
陈妍颖
李庶林
刘胤池
胡静云
彭府华
TANG Chao;CHEN Yanyin;LI Shulin;LIU Yinchi;HU Jingyu;PENG Fuhua(School of Architecture and Civil Engineering,Xiamen University,Xiamen,Fujian 361005,China;Changsha Institute of Mining Research Co.,Ltd.,Changsha,Hunan 410012,China)
出处
《岩石力学与工程学报》
EI
CAS
CSCD
北大核心
2020年第S01期3197-3205,共9页
Chinese Journal of Rock Mechanics and Engineering
基金
国家自然科学基金资助项目(51674218)。
关键词
边坡工程
边坡变形预测
灰色预测
OGM(1
N)
自适应变异粒子群算法
多变量预测模型
slope engineering
slope deformation prediction
gray prediction
OGM(1,N)
adaptive particle mutation swarm optimization
multivariate prediction model
