基于随机森林模型的粗轧过程孔型优化设计

Hole pattern optimization design of rough rolling process based on random forest model

以粗轧轧辊孔型为优化对象,对影响折皱缺陷的因素进行了分析,提出了一种基于随机森林的粗轧过程孔型参数优化设计方法。首先,采用DEFORM-3D软件建立三维轧制热力耦合有限元模型,分析了折皱缺陷的产生原因,探究了折皱缺陷的影响参数;然后,开发了基于随机森林算法的机器学习模型,建立了变形温度、孔型圆角尺寸、槽底宽度和轧辊与轧件间的摩擦因数与宽展值之间关系的设计数据库;结合约束条件,得到了孔型参数的最优解,最后,对优化后的孔型进行了验证。研究结果表明:当孔型槽底宽度b k为165 mm、圆角尺寸R为14 mm时,取得最优解,此时宽展值最小;采用优化后孔型进行生产,消除了折皱缺陷,成品合格率明显提高。

Taking the rough rolling roller hole pattern as the optimization object,the factors affecting the wrinkle defects were analyzed,and an optimization design method of hole pattern parameters in rough rolling process based on random forest was proposed.Firstly,the three-dimensional rolling thermal mechanical coupled finite element model was established using DEFORM-3D software to analyze the causes of wrinkle defects and explore the influence parameters of wrinkle defects.Then,the machine learning model based on random forest algorithm was developed.The design database of the relationship between deformation temperature,hole pattern fillet size,groove bottom width,and friction factor between roller and[JP3]rolled parts and spread value was established.Combined with the constraints,the optimal solution of hole pattern parameters was obtained.Finally,the optimized hole pattern was verified.The research results show that when the groove bottom width b k is 165 mm and the fillet size R is 14 mm,the optimal solution is obtained,and the width spread is the smallest.After the optimized hole pattern for production,it is found that the wrinkle defects are eliminated,and the qualified rate of the finished product is obviously improved.