摘要
充填体强度在膏体充填技术的安全与高效方面起到显著作用,采用贝叶斯算法优化SVR模型构建了混合骨料充填体强度预测模型。探究了膏体料浆质量分数、水灰比、灰砂比及废石占比各单因素及交互作用对充填体强度的影响规律。结果表明:相比传统机器学习方法贝叶斯优化SVR模型在测试集上决定系数R^(2)表现为0.979且误差指标更小,可对充填体强度进行精确预测。各特征因素对于充填体强度影响大小依次为水灰比>质量分数>灰砂比>废石占比。在交互作用中水灰比与灰砂比对于水化反应及充填体强度起决定性作用;质量分数与废石占比对于搭建骨架结构和反映密实程度起重要作用;灰砂比与质量分数对充填体强度提升呈现梯度放大效应。本研究可为充填体强度预测、优化物料配比及改善经济成本提供理论依据,对实际矿山充填工程具有重要的指导价值。
The strength of the filling material plays a significant role in ensuring the safety and efficiency of paste filling technology,A strength prediction model for mixed aggregate backfills was constructed by optimizing the SVR model using Bayesian algorithms,The influence of individual factors and their interactions on the strength of the filling body was investigated,including the mass fraction of paste slurry,water-cement ratio,cement-sand ratio,and the proportion of waste rock in the mixed aggregate.The results show that:compared to traditional machine learning methods,the Bayesian optimized SVR model exhibits a determination coefficient R^(2)of 0.979 on the test set and a smaller error metric,enabling accurate prediction of the strength of the filling body.The order of the influence of each characteristic factor on the strength of the filling body is as follows:water-cement ratio>mass fraction>cement-sand ratio>waste rock proportion.In the interaction,the water-cement ratio and cement-sand ratio play a decisive role in the hydration reaction and the strength of the filling body;Quality score and waste rock proportion play a significant role in establishing the framework structure and reflecting the degree of compaction;The ratio of cement to sand and mass fraction exhibit a gradient amplification effect on the strength enhancement of the filling body.This study provides a theoretical basis for predicting the strength of backfill,optimizing material ratios,and improving economic costs,and has important guiding value for practical mine backfill projects.
作者
李聪
何维
肖亚辉
罗松
米小伟
王康平
梁买东
晏承园
王俊
LI Cong;HE Wei;XIAO Yahui;LUO Song;MI Xiaowei;WANG Kangping;LIANG Maidong;YAN Chengyuan;WANG Jun(Zijin Mining Group Co.,Ltd.,Shanghang 364200,Fujian,China;Longnan Zijin Mining Co.,Ltd.,Longnan 742100,Gansu,China;School of Land Resources Engineering,Kunming University of Science and Technology,Kunming 650093,China)
出处
《有色金属(矿山部分)》
2026年第2期88-95,共8页
NONFERROUS METALS(Mining Section)
基金
云南省基础研究专项-青年项目(202101AU070022)
昆明理工大学人培基金(KKZ3202021040,KKSY201921017)。
关键词
贝叶斯优化
SVR模型
充填体强度预测
特征值重要性
影响规律
bayesian optimization
SVR model
strength prediction of filling body
eigenvalue importance
influence law
