论文部分内容阅读
以结构钢板的大量实测数据为基础,利用 MatLab 人工神经网络工具箱,建立了结构钢板变形区的应力状态系数与锻前、锻后厚度的 RBF 神经网络预测模型。通过分析应力状态系数的影响因素,结合传统的数学模型,确立了网络的输入层参数,并对函数newrb中宽度系数 spread 的实验调整,确定了最佳的网络结构形式,提高了模型的预测精度,并与 BP 和 Elman 神经网络模型相比较,结果表明,RBF 神经网络能更好地适用于金属应力状态系数模型。
Based on a large number of measured data of structural steel plate, the prediction model of RBF neural network with stress state coefficient and pre-forging thickness of the deformation zone of the structural steel plate is established by using MatLab artificial neural network toolbox. By analyzing the influencing factors of stress state coefficient and combining with the traditional mathematical model, the input layer parameters of the network are established, and the experimental adjustment of the width coefficient spread in function newrb is established, the best network structure is determined, and the prediction accuracy of the model is improved , And compared with the BP and Elman neural network models. The results show that the RBF neural network can be better applied to the metal stress state coefficient model.