IF钢屈服强度预测仿真研究  

Simulation Research on Forecasting Yield Strength of IF Steel

作　　者：曲强[1] 樊林林 陈雪波[1] 汪瑾[1] 

机构地区：[1]辽宁科技大学电子与信息工程学院,辽宁鞍山114051

出　　处：《计算机仿真》2015年第5期382-385,共4页Computer Simulation

基　　金：自然科学基金(71371092)

摘　　要：研究IF钢的屈服强度与材料微观组织结构间的关系,为衡量IF钢的强度,提高预测模型的精度。屈服强度是衡量IF钢力学性能的重要指标,是IF钢的使用极限,当应力超过屈服极限后产生颈缩,应变增大,使IF钢破坏。由于影响IF钢屈服强度的因素众多,且具有高度的非线性,很难用精确的数学模型描述。为解决上述问题,首先采用互信息的特征选择方法选择预测模型的输入,然后用减法聚类法实现自适应神经模糊推理系统的结构识别,并采用混合学习算法训练该自适应神经模糊推理系统,实现对IF钢屈服强度的预测。最后,对所建模型进行仿真验证,并将仿真验证结果与传统的BP神经网络进行比较,仿真结果表明,采用自适应神经模糊推理系统的预报模型,在收敛速度及建模精度方面均优于传统的BP神经网络。The relationship between yield strength and microstructure of IF steel is studied and the accuracy of prediction model is improved. A feature selection method based on mutual information is used to choose the input feature of prediction model. The structure of adaptive neural-fuzzy inference system （ANFIS） is identified by subtractire clustering algorithm and the network parameters are trained by hybrid learning algorithm. And the model is simulated and the result of the simulation is compared with the traditional BP neural network, the simulation proves that the prediction model based on adaptive fuzzy inference system in terms of speed of convergence and accuracy of modeling is superior to the traditional BP neural network.

关 键 词：自适应神经模糊推理系统 减法聚类 互信息 屈服强度 反向传播网络 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]