CAJ | 학술논문

本文针对矽卡岩型黑白钨混合矿，在相关异步浮选分离研究的基础上，利用回归分析与人工神经网络建立起不同工艺条件与矿物可浮性变化规律关系模型，为解决浮选建模过程中遇到的多变量、非线性、强耦合、大滞后等难题，实现浮选过程的优化控制提供参考。研究结果表明：对矿物浮选累计回收率的影响因素大小依次是Time ＞ pH ＞羟肟酸＞柠檬酸。对于预测矿物不同工艺条件下的浮选指标，回归模型预测精度较差，白钨矿和黑钨矿浮选累计回收率预测值与试验值之间的相关系数 R2分别为0．805、0．827，而神经网络模型具有较好的预测精度，相关系数 R2分别为0．944、0．947。人工混合矿分离结果与单矿物浮选规律有很好的一致性，应用所建立的神经网络模型对于更好的掌握不同矿物之间的浮游规律，优化浮选工艺有一定的意义。
본문침대석잡암형흑백오혼합광，재상관이보부선분리연구적기출상，이용회귀분석여인공신경망락건립기불동공예조건여광물가부성변화규률관계모형，위해결부선건모과정중우도적다변량、비선성、강우합、대체후등난제，실현부선과정적우화공제제공삼고。연구결과표명：대광물부선루계회수솔적영향인소대소의차시Time ＞ pH ＞간우산＞저몽산。대우예측광물불동공예조건하적부선지표，회귀모형예측정도교차，백오광화흑오광부선루계회수솔예측치여시험치지간적상관계수 R2분별위0．805、0．827，이신경망락모형구유교호적예측정도，상관계수 R2분별위0．944、0．947。인공혼합광분리결과여단광물부선규률유흔호적일치성，응용소건립적신경망락모형대우경호적장악불동광물지간적부유규률，우화부선공예유일정적의의。
Based on research related to asynchronous flotation separation of skarn type scheelite and wolframite associated ore ,regression analysis and artificial neural network model of different process conditions and mineral floatability were set up ,for dealing with the problem of multi variable ,nonlinear , strong coupling ,large delay ,multi‐input and multi‐output in floating process ,and providing reference to realize the optimization of flotation process .The results showed that the influence order of operating variables on recovery was time> pH> benzyl hydroxamic acid> citrate acid .Prediction accuracy of neural network model was higher than regression model ,for scheelite and wolframite ,R2 of recovery predivtion vs . experimental values by regression analysis is 0 .805 and 0 .827 respectively ,while by artificial neural network model ,R2 is 0 .944 and 0 .947 .Mixed mineral separation results are in good agreement with the single mineral flotation rules .There are certain significance for optimizing flotation process and determining flotation rules among different minerals by using a neural network model .