【摘要】 针对转炉控制中对吹炼终点温度的控制问题,提出了基于混合递阶遗传RBF神经网络(HGA-RBF)的转炉炼钢终点温度预报模型。研究了RBF网络的特点,用递阶遗传算法克服了网络的结构和参数选择的随机性问题;并结合最小二乘法,提高了收敛速度。仿真结果表明,此算法在一定程度上提高了RBF网络的优化收敛速度和训练测试精度。某钢铁公司提供的实际冶炼数据试验,也证明了该模型预报精度较高,对提高生产的质量有重要意义。更多还原

【Abstract】 To the temperature control problem in the BOF period of steel-making,the predictive models of hybrid hierarchy genetic algorithm (HGA)and radial basis function neural network(RBF)in basic oxygen furnace process(BOF)are presented.Based on the research on the hidden layer structure and the parameter character of RBF network,the two parameters are coded and optimized by genetic evolution individual- ly.The predictive model of endpoint temperature based on Matlab is put forward.The simulation results show the method has fast convergence speed and remarkable accuracy.The achieving application in some steel-making company proves that the model is of high predictive accuracy, and important to increase the production quality.更多还原