基于BP神经网络的DP980钢板本构关系  

Constitutive relationship of DP980 steel sheet based on BP neural network

作　　者：林启权[1] 卜根 王镇柱[1] 董文正[1] 王凯[1] LIN Qi-quan;BU Gen;WANG Zhen-zhu;DONG Wen-zheng;WANG Kai(School of Mechanical Engineering,Xiangtan University,Xiangtan 411105,China)

机构地区：[1]湘潭大学机械工程学院,湖南湘潭411105

出　　处：《塑性工程学报》2022年第8期139-144,共6页Journal of Plasticity Engineering

基　　金：湖南省自然科学基金资助项目(2020JJ4578,2022JJ30565);湖南省教育厅重点项目(21B0104)。

摘　　要：为了准确描述双相钢温热条件下的本构关系,以DP980钢板为研究对象,采用单向拉伸实验获得了不同温度(300~500℃)、材料方向(与轧制方向呈0°、45°和90°)下钢板的流动应力。以实验温度、材料方向和真实应变作为输入参数,真实应力作为输出参数,建立了DP980钢板流动应力的BP神经网络预测模型,同时通过最小二乘拟合法获得了不同温度下DP980钢板的Voce本构模型,并与BP神经网络模型的预测精度进行对比。结果表明,DP980钢板不仅具有负温度敏感性,还随温度升高呈现出明显的各向异性;BP神经网络预测模型相比于Voce模型可以更精确地预测出DP980钢板在不同温度和材料方向下的流动应力。To accurately describe the constitutive relationship of dual-phase steel under warm condition,taking the DP980 steel sheet as the research object,the uniaxial tensile experiments were used to obtain the flow stress of steel sheet at different temperatures(300-500℃)and different material orientation(0°,45°and 90°to the rolling direction).Taking the experimental temperature,material direction and true strain as the input parameters and the true stress as the output parameter,the BP neural network prediction model for flow stress of DP980 steel sheet was established.At the same time,the Voce constitutive model of DP980 steel sheet at different temperatures was obtained by the least square fitting method and the prediction accuracy of the model was compared with that of BP newral network model.The results show that the DP980 steel sheet not only has negative temperature sensitivity,but also presents obvious anisotropy with the increase of temperature;BP neural network prediction model can more accurately predict the flow stress of DP980 steel sheet at different temperatures and material directions compared with Voce model.

关 键 词：DP980钢板 BP神经网络 流动应力 Voce本构模型 温热条件 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程] TG142[自动化与计算机技术—控制科学与工程]