闪速熔炼过程中存在大量多元非线性因素,难以从统计学和机理上确立操作参数。为优化闪速炉的操作参数,建立了动态T-S递归模糊神经网络(DTRFNN)的软测量模型,推导了DTRFNN的权值学习算法,将其应用到某厂的铜闪速熔炼过程中的参数软测量上,平均精确率达到97%,能为生产操作提供有益的指导。
In flash smelting process, there exist a large number of multiple nonlinear factors. From the view point of statistics and reaction mechanism, it is difficult to establish the operating parameters. A soft sensor model based on dynamic T-S recurrent fuzzy neural networks (DTRFNN) is put forward to optimize the operating parameters. The weighted leaming algorithm of DTRFNN has been deduced. This model is applied in the parameter soft sensor of copper flash smehing process in a factory. Application result shows that the average precision reaches to 97%. The proposed modeling can provide useful instruction for production operation.