摘要
针对金川镍矿充填法采矿,文章利用酒钢低活性水淬渣对早强新型充填胶凝材料进行了研究,并对酒钢水淬渣进行物化特性分析,选择生石灰、脱硫灰渣、粉煤灰和芒硝作为复合激发剂进行胶凝材料配方正交试验。在此基础上,建立BP(back propagation)神经网络模型对试验样本进行学习和训练,获得激发剂配方与胶凝材料特性之间的隐含知识。并借助隐含知识进行不同激发剂配方胶凝材料特性预测,揭示胶凝材料特性随激发剂掺量的变化规律,由此确定了新型充填胶凝材料最优配方生石灰、脱硫灰渣、粉煤灰、芒硝、亚硫酸钠、酒钢渣粉的掺量分别为3%、5%、5%、1%、2%、84%。对该最优配方进行验证试验,获得3d和7d的强度分别达到1.735 MPa和2.876 MPa,完全满足金川矿山下向胶结充填法采矿对胶凝材料的强度要求。
Considering the filling mining of Jinchuan nickel mine ,the development of early strong ce‐mentitious material by using low activity slag of Jiuquan steel was studied .Firstly ,the physicochemi‐cal characteristic of water quenching slag was analyzed ,and the quicklime ,desulfurization ash ,fly ash and mirabilite were selected as composite activator and the orthogonal test of cementitious materials was finished .On this basis ,the back propagation(BP) neural network model was established for the learning and training of the test sample ,and the implied knowledge between the activator and cemen‐titious materials properties was gotten .The characteristic of different activator formulations of ce‐mentitious materials was predicted by using the implied knowledge ,and the change rules of cementi‐tious material properties with the activator content were revealed .The optimal ratio of new filling ce‐mentitious material was as followed :the quicklime ,desulfurization ash ,fly ash ,mirabilite ,sodium sulfite and slag powder were 3% ,5% ,5% ,1% ,2% and 84% ,respectively .The results of verifica‐tion test showed that the compressive strength of 3 d and 7 d was 1.735 MPa and 2.876 MPa ,respec‐tively ,which fully meets the strength requirements of downward filling mining with cementitious ma‐terial in Jinchuan mine .
出处
《合肥工业大学学报(自然科学版)》
CAS
CSCD
北大核心
2016年第9期1189-1195,共7页
Journal of Hefei University of Technology:Natural Science
基金
国家高技术研究发展计划(863计划)资助项目(SS2012AA062405)
关键词
矿渣微粉
低活性
胶凝材料
BP神经网络
配比优化
slag powder
low activity
cementitious material
back propagation(BP) neural network
ratio optimization