金氰化浸出过程混合建模及仿真算法分析研究

Hybrid Modeling and Simulation Algorithm of Gold Cyanide Leaching Process

针对某金氰化浸出过程,首先建立了多级动态机制模型,并在此基础上采用基于免疫原理的RBF神经网络数据模型的学习算法,计算实际生产过程中难以测量的动力学反应速度,从而估算动态机制模型中的未知参数,再与物料守恒方程串联,建立了预测金浸出率的串联型混合模型。其次与基于免疫原理的RBF神经网络建立的纯数据模型对比仿真来验证模型的有效性;最后根据实际生产过程的实际值与预测值之间的偏差进行模型更新,并通过仿真分析验证了更新机制的准确性。结果表明:串联混合模型大大提高了浸出过程浸出率的预测精度,模型的更新机制提高了模型的精准性与泛化能力。

Firstly,a multi-stage dynamic mechanism model was established for a gold cyanide leaching process.On this basis,the learning algorithm of RBF neural network data model based on immune principle was used to calculate the dynamic reaction rate which was difficult to measure in the actual production process,so as to estimate the unknown parameters in the dynamic mechanism model,and then connected with the material conservation equation in series.Therefore,a series mixed model was established to predict the leaching rate.Secondly,the model was compared with the pure data model based on the immune principle of RBF neural network to verify the effectiveness of the model.Finally,the model was updated according to the deviation between the actual value and the predicted value in the actual production process,and the accuracy of the updating mechanism was verified by simulation analysis.The results show that the series mixing model greatly improves the prediction accuracy of the leaching rate in the leaching process,and the updating mechanism of the model improves the accuracy and generalization ability of the model.