摘要
针对某钢厂生产的磨球钢探伤合格率较低的问题,利用梯度提升树算法建立了加热工艺到探伤合格率的机器学习模型。通过该模型,确定了各项加热工艺对探伤合格率的影响程度。结果表明,预热的温度是对探伤合格率影响系数最大的主要因素。固定其他参数,利用模型计算不同预热温度下的探伤合格率,发现最佳预热温度为690~720℃。为进一步验证和优化这一结果,采用了热力耦合的有限元方法,计算了不同预热温度下的最大应力。结果显示,应力最小的预热温度范围为710~720℃。控制加热温度在这一范围上下后,磨球钢的实际生产的探伤合格率由原来的90.8%提高到了93.9%。
To improve the qualification rate of non-destructive testing for grinding ball steel produced by a certain steel plant,a machine learn-ing model was established.This model uses the gradient boosting tree algorithm to determine how the heating system affects the qualification rate of non-destructive testing,Through this model,it determined the influence of various heating systems on the qualification rate of non-de-structive testing.The results indicate that preheating temperature is the main factor with the greatest impact on the qualification rate.We fixed other parameters and used the model to calculate the pass rate of non-destructive testing at different preheating temperatures.It was found that the optimal preheating temperature range is between 690℃to 720℃.Additionally,we used the finite element method with thermal coupling to calculate the maximum stress at different preheating temperatures.The results show that the preheating temperature range with the minimum stress is 710℃to 720℃.After controlling the heating temperature,the pass rate of non-destructive testing for grinding ball steel increased from 90.8%to 93.9%.
作者
武甲文
秦哲
郭庆涛
赵成林
岳洪进
肖玉宝
王善宝
WU Jiawen;QIN Zhe;GUO Qingtao;ZHAO Chenglin;YUE Hongjin;XIAO Yubao;WANG Shanbao(AnSteel Beijing Research Institute Co.Ltd.,Beijing 102211,China;Lingyuan Iron and Steel Co.Ltd.,Chaoyang 122599,China;Iron&Steel Research Institute of AnSteel Group Corporation AnSteel Co.Ltd.,Anshan 114003,China)
出处
《工业加热》
2025年第10期7-10,21,共5页
Industrial Heating
基金
国家重点研发计划资助项目(冶金关键部件表面工程技术示范应用,2021YFB3702005)。
关键词
探伤合格率
加热工艺
机器学习
热力耦合有限元方法
grinding ball steel
large bar
heating system
gradient boosting tree
thermal-mechanical coupling mode
