论文部分内容阅读
使用神经网络模型预测难选冶金精矿在臭氧和三价铁氧化条件下的铁浸出率。神经网络的输入结点是6个操作参数:臭氧浓度,三价铁离子浓度,液固比,氧气量,氧化时间,反应温度;神经网络的输出结点是难选冶金精矿中铁的氧化率。基于误差反向传播算法的多层前向神经网络使用33组实验值,采用6-11-1的网络结构经过反复训练得到一个良好模型,其相关系数R2为0.966。对神经网络与常规的多元线性回归2种模型进行对比。神经网络的计算结果表明:在所有操作参数中,温度是最重要的影响因素,臭氧为第二重要的影响因素。神经网络模型能够准确地预测黄金冶炼厂的难选冶金矿的预处理步骤中铁的氧化率,并可用来优化工艺参数。
Prediction of Iron Leaching Rate of Refractory Metallurgical Concentrate under Ozone and Ferric Oxidation Conditions Using Neural Network Model. The input nodes of the neural network are six operating parameters: ozone concentration, ferric ion concentration, liquid-solid ratio, oxygen amount, oxidation time, reaction temperature; the output node of neural network is iron oxidation rate of refractory metallurgical concentrate . Based on the error backpropagation algorithm, the multi-layer feedforward neural network uses 33 sets of experimental values and uses 6-11-1 network structure to obtain a good model after repeated training, with a correlation coefficient R2 of 0.966. The neural network is compared with the conventional multiple linear regression model. The results of the neural network show that temperature is the most important factor among all the operating parameters. Ozone is the second most important factor. The neural network model can accurately predict the oxidation rate of iron in the pretreatment step of the gold smelter refractory metallurgical ore and can be used to optimize the process parameters.