基于神经网络的带钢热连轧机轧制力预报

Based on Neural Network Strip Steel Strip Machine Rolling Force Prediction

采用贝叶斯统计学原理改进传统神经网络算法,通过在神经网络的目标函数中引入了表示网络结构复杂性的约束项,避免了网络的过拟合以提高网络的泛化能力。将改进的神经网络应用于某钢铁公司1700 mm热连轧机带钢轧制力中,其预报精度、训练时间和网络稳定性均优于传统神经网络预测,其改进传统的轧制力预报方式,从而进一步提高轧制力预报精度和辊缝设定精度,以期进一步带钢厚度质量。

Based on the bayesian statistical principle,improve the paper improves the traditional neural network algorithm,through introducing constraint which express network structure complexity in neural network objective function,avoiding network over fitting to increase the network generalization ability.The improved neural network is applied to a steel company 1700 mm strip in strip rolling force machine,its forecast accuracy,training time and network stability are all better than traditional neural network forecast.,Iit improved the traditional rolling force prediction mode,so as to further improve the rolling force prediction precision and roll gap setting precision,so asas well as to further the strip thickness quality.