摘要
为准确预测尾砂胶结充填体强度,建立了以孔隙率、粒级不均匀系数、曲率系数、充填料浆浓度、1/灰砂比、养护龄期6个因素作为输入因子,尾砂胶结充填体强度为输出因子的GEP充填体强度预测模型。以多个矿山室内充填物料配比试验数据为例,验证GEP预测模型的可行性。结果表明,GEP预测结果与实际值相对误差仅为5.13%。与BP神经网络模型的预测值进行对比,结果表明,GEP预测模型比BP神经网络模型预测结果(平均相对误差14.09%)更加精确,与实测值的拟合度更好。
Gene Expression Programming (GEP) model is established, in order to accurately predict the strength of cemented backfill. We take porosity, particle level uneven coefficient, coefficient of curvature, filling slurry concentration, 1/sand ratio, curing age as the input factors, and the cemented railings backfill strength as output factor. Taking the test data of several mines as examples, the relative error of GEP forecast and the actual value is only 5.13~. The results show that the predicted value of GEP forecasting model is better than that of the BP neural network model(average relative error 14. 090%) and the fitting degree of the GEP model is better than that of the BP neural network model too.
出处
《有色金属(矿山部分)》
2015年第B08期13-18,共6页
NONFERROUS METALS(Mining Section)
关键词
基因表达式编程
充填体强度
充填试验
预测模型
BP神经网络
Gene Expression Programming (GEP)
backfill strength
backfill test
prediction model
BP neural network