基于NNGA模型的纤维膏体充填复合材料强度预测  

Strength Prediction of Fiber Cemented Paste Backfill Composite based on NNGA Model

作　　者：张倓 黄湛煐 聂华伟[2] ZHANG Tan;HUANG Zhanying;Nie Huawei(Guizhou Equipment Manufacyuring Vocational College,Guiyang 551400,China;Guizhou Communication Vocational College,Guiyang 551400,China)

机构地区：[1]贵州装备制造职业学院,贵州贵阳551400 [2]贵州交通职业技术学院,贵州贵阳551400

出　　处：《矿业工程》2023年第4期65-68,共4页Mining Engineering

基　　金：贵州省科技厅科学技术基金(黔科合基础[2020]1Y280)。

摘　　要：为了更快更准确获取添加聚丙烯纤维情况下不同配合比对矿山膏体充填材料强度的影响,构建了一种基于遗传算法反向传播神经网络NNGA模型,利用模型对纤维增强膏体充填材料的3d、7d和28d龄期的抗压强度进行预测和分析,R2值为0.98317,7d样本中绝对误差最小达到了0.008080,两个模型的最大相对误差下降了约44.18%,对比NNGA和BPNN模型的预测结果表明,NNGA混合模型预测的泛化能力更强,波动程度更小和稳定性更好,在工程实际中更具有可行性。In order to more quickly and accurately obtain the effect of different mix ratios on the strength of the mine paste filling material under the condition of adding polypropylene fibers,a back-propagation neural network NNGA model based on genetic algorithm was created,and the model was used to predict and analyze the compressive strength of fiber-reinforced paste flling materials at 3d,7d and 28d age.The R2 value was 0.98317,the R2 value was 0.98317,and the minimum absolute error in the 7d sample reached 0.008080 and the maximum relative error decreased by about 44.18%.Comparing the prediction results of NNGA and BPNN models,it shows that NNGA hybrid model has stronger generalization ability,less fluctuation and better stability,and is more feasible in engineering practice.

关 键 词：纤维 抗压强度 膏体充填 预测 

分 类 号：X705[环境科学与工程—环境工程] TD823.7[矿业工程—煤矿开采]