基于ASO-BP神经网络的屈服强度预测技术研究  

Research on prediction technique of yield strength based on ASO⁃BP neural network

作　　者：杨小平[1,2] 武修瑞 郑许[3,4] 任月路 朱玉涛[3,4] 何克准 YANG Xiaoping;WU Xiurui;ZHENG Xu;REN Yuelu;ZHU Yutao;HE Kezhun(School of Information Science and Engineering,Guilin University of Technology,Guilin 541004,China;Guangxi Key Laboratory of Embedded Technology and Intelligent System,Guilin University of Technology,Guilin 541004,China;ALG Aluminium lnc.,Nanning 530031,China;Guangxi Key Laboratory of Materials and Processes of Aluminum Alloys,Nanning 530031,China)

机构地区：[1]桂林理工大学信息科学与工程学院,广西桂林541004 [2]桂林理工大学广西嵌入式技术与智能系统重点实验室,广西桂林541004 [3]广西南南铝加工有限公司,广西南宁530031 [4]广西铝合金材料与加工重点实验室,广西南宁530031

出　　处：《兵器材料科学与工程》2023年第6期6-10,共5页Ordnance Material Science and Engineering

基　　金：国家高新技术研发计划(863计划)项目(2013AA12210504);广西科技重大专项(桂科AA23023027)。

摘　　要：针对传统屈服强度预测模型通用性较差的问题,提出一种采用原子搜索优化算法优化BP神经网络,建立多类型合金屈服强度预测模型的方法。以Kaggle公开数据为研究对象,对89种钢合金建立ASO-BP神经网络屈服强度预测模型,同时与PSO-BP,GA-BP,BP神经网络模型对比。结果表明:ASO-BP预测模型平均绝对百分比误差(MAPE)为6.98%,相关系数达到0.98716,效果优于其他对比模型。验证了预测多种类型合金屈服强度的合理性和可靠性,为工程实际应用和合金屈服强度检测提供较好的辅助判断。Aiming at the problem of poor generality of traditional prediction model of yield strength,a method of BP neural network optimized by atom search optimization algorithm was proposed to build yield strength prediction model of multi⁃type alloy.Based on the open data of Kaggle,the yield strength prediction model of ASO⁃BP neural network was established for 89 kinds of steel alloys,and was compared with PSO⁃BP,GA⁃BP and BP neural network models.The results show that the average absolute percentage error(MAPE)of the ASO⁃BP prediction model is 6.98%,and the correlation coefficient is 0.98716,which is better than those of other comparison models.The rationality and reliability of prediction model of the yield strength of various types of alloys are verified,which provides a good auxiliary judgment for engineering application and measurement of yield strength of alloy.

关 键 词：低合金钢 屈服强度 预测模型 原子搜索优化算法 BP神经网络 

分 类 号：TG142.33[一般工业技术—材料科学与工程] TP391.9[金属学及工艺—金属材料]