IGA-BP网络模型在高炉铁水硅含量预测中的应用

Application of the model of IGA-BP networks to predicting the silicon content of blast furnace hot metal

文章针对BP网络收敛速度慢和易于陷入极小值的问题,采用免疫遗传算法全局寻优和BP网络局部寻优相结合的方法,提高了BP网络的计算精度和收敛速度;应用IGA-BP网络模型对高炉铁水硅含量进行了预测,数值结果对比发现,该模型提高了预测精度的同时,迭代次数比一般BP网络模型也大大减少;仿真结果证明了方法的有效性。

By combining global optimization of the Immune Genetic Algorithm（IGA） with local optimization of the BP neural network, the calculation accuracy and convergence rate of the BP neural network are improved, and the model of IGA-BP networks is applied to predicting the silicon content of blast furnace hot metal. The numerical results of the model show that the prediction precision is improved and the iteration number is much less than that by the normal BP network model. The test and simulation results are satisfactory.