基于Matlab神经网络工具箱,采用改进的径向基函数(RBF)网络优化计算4200中厚板轧机的轧制温度。通过径向基层散布常数的人工调整以及神经元的自适应调整,提高了收敛速度,确定了最佳的网络结构形式。网络预测结果与在4200中厚板轧机上实测的轧件温度进行了对比,预测精度很高,表明了该网络模型的优越性。该模型可为以温度参数为主要控制对象实行自动化生产的4200轧机提供可靠的参数,也可为人工神经网络在其它自动控制方面的应用提供参考。 Based on the Matlab neural network toolbox, the optimized radial basis function (RBF) network is used to optimize the rolling temperature of 4200 plate mill. Through the artificial adjustment of radial base spreading constant and the adaptive adjustment of neurons, the convergence speed is improved and the best network structure is determined. The network prediction results are compared with the measured temperature of the rolling stock on the 4200 plate mill, and the prediction accuracy is very high, which shows the superiority of the network model. The model can provide reliable parameters for the 4200 rolling mill with automatic production of temperature parameters as the main control object, and can also provide reference for the application of artificial neural network in other automatic control.