基于高阶多元回归和遗传优化的白云鄂博铌粗精矿最佳焙烧条件研究  

Study on Optimal Roasting Conditions of Bayan Obo Niobium Minerals Based on High Order Multiple Regression and Genetic Optimization

作　　者：徐策 廖振鸿 冉孟杰 陈雯[1] XU Ce;LIAO Zhenhong;RAN Mengjie;CHEN Wen(Department of Mineral Resources Development and Utilization,Changsha Research Institute of Mining and Metallurgy Co.,Ltd.,Changsha 410012,China;School of Automation,China University of Geosciences,Wuhan 430074,China)

机构地区：[1]长沙矿冶研究院有限责任公司,长沙410012 [2]中国地质大学自动化学院,武汉430074

出　　处：《有色金属（选矿部分）》2024年第11期92-99,共8页Nonferrous Metals（Mineral Processing Section）

基　　金：国家自然科学基金重点基金资助项目(92062223)。

摘　　要：针对白云鄂博铌粗精矿还原焙烧产生富铌矿相过程中,内部反应复杂,耦合因素多,调控成本高等问题。提出了一种基于高阶多元回归和遗传优化的最佳焙烧条件研究方法,通过模拟计算的手段,大大降低了时间和人力成本。对于焙烧结果的评价,构建了迁移率指标,用于探究铌粗精矿中的含铌矿物和焙烧后的富铌矿相之间的元素迁移情况。通过斯皮尔曼相关系数对各元素之间的相关性进行分析,结果表明,Ti、Sr、La、Ce、Nd等元素与铌元素具有较强正相关性,可以作为协调调控的元素。基于四阶多元回归算法建立元素迁移模型,得到了元素迁移率与FeO含量和CaO/SiO_(2)含量之间的四阶线性模型。对于模型准确性的验证,计算均方根误差RMSE为2.6382;可决系数(R^(2))为:0.87492,证明了模型具有较高的精度。针对最佳焙烧试验条件的寻优问题,利用启发式算法—遗传算法,对训练的模型寻找全局最优解;经过200代迭代后,得到最优适应度函数为-13.3882%,对应的试验条件为:CaO/SiO_(2)含量为1.001,还原Fe的含量为18.3961。在此最优操作条件下,焙烧后富铌矿相中Nb元素的迁移率提升了30%。该计算方案具有很好的可扩展性,可在更高维度的试验条件下推广,相比传统试验方法具有更好的精度和效率。In the process of producing niobium-rich ore phase by reduction roasting of Baiyun Obo niobium crude concentrate,the internal reaction is complicated,there are many coupling factors and the control cost is high.A research method of optimum roasting conditions based on high order multiple regression and genetic optimization is proposed.By means of simulation calculation,time and labor cost are greatly reduced.For the evaluation of roasting results,a mobility index was established to investigate the element migration between niobium-bearing minerals in niobium concentrate and niobium-rich ore phases after roasting.Spearman correlation coefficient was used to analyze the correlation between the elements.The results showed that Ti,Sr,La,Ce,Nd and niobium had strong positive correlation,and could be used as coordinated control elements.The element migration model was established based on the fourth-order multiple regression algorithm,and the fourth-order linear model between the element mobility and the content of FeO and CaO/SiO_(2) was obtained.To verify the accuracy of the model,the calculated root-mean-square error RMSE is 2.6382.The coefficient of determination(R^(2))is 0.87492,which proves that the model has high accuracy.Aiming at the problem of optimal roasting conditions,the heuristic algorithm-genetic algorithm is used to find the global optimal solution for the trained model.After 200 iterations,the optimal fitness function is-13.3882%,and the corresponding experimental conditions are:the content of CaO/SiO_(2) is 1.001,and the content of reduced Fe is 18.3961.Under the optimal operating conditions,the mobility of Nb element in the niobium-rich ore phase after roasting is increased by 30%.The calculation scheme has good scalability and can be extended under higher dimensional experimental conditions,and has better accuracy and efficiency than traditional experimental methods.

关 键 词：铌矿物 还原焙烧 高阶多元回归 遗传算法 

分 类 号：TD954[矿业工程—选矿] TD925