基于梯度提升回归算法的刀具磨损评估模型  

Tool Wear Evaluation Model Based on Gradient Boosting Regression

作　　者：项海婧 宫爱红[1] 胡明茂[1] Xiang Haijing;Gong Aihong;Hu Mingmao(School of Mechanical Engineering,Hubei University of Automotive Technology,Shiyan 442002,China)

机构地区：[1]湖北汽车工业学院机械工程学院,湖北十堰442002

出　　处：《湖北汽车工业学院学报》2021年第4期65-69,共5页Journal of Hubei University Of Automotive Technology

基　　金：湖北省重点研发计划项目(020BAA005);工信部工业互联网创新发展工程项目(TG200802C,TC200A00W)。

摘　　要：为了提高加工过程中刀具磨损在线评估的准确性,构建了基于梯度提升回归算法的刀具磨损评估模型。以刀具的铣削力、振动信号时频域指标和声发射信号滤波后的最大幅值处频率作为特征值,以对应工况下的刀具磨损量为目标值来构造数据样本。在迭代训练过程中,刀具磨损评估模型将数据集中2组刀具的数据样本合并作为训练集与验证集,另1组刀具的数据样本独立作为测试集,并采用交叉验证和参数网络搜索的优化方法来提高模型的泛化能力。仿真结果表明,梯度提升回归模型比同等条件下线性回归模型、贝叶斯岭回归模型、弹性网络回归模型和支持向量机的R^(2)预测值提高了42%、36%、10%和13%。In order to improve the accuracy of in-process tool wear evaluation in machining engineering,a tool wear evaluation model was built based on gradient boosting regression.The frequency of maximum amplitude of the milling force,the vibration signal time frequency domain index,and the sound emission signal filtering were used as the characteristic value,then the tool wear under the corresponding working condition was used as the target value to construct the data sample.In the iterative training process,the tool wear assessment model combines two sets of tools data samples in data sets as the training set and verification set,and the other set was used as the test set,the optimization method of crossverification and parameter network search is used to improve the generalization of the model.The simulation results show that the gradient boosting regression model compared to the Bayes-ridge regression model,the elastic network regression model,and the support vector machine in the equivalent condition that R^(2) predicted value can be increased by 42%,36%,10%and 13%.

关 键 词：刀具磨损评估 梯度提升回归 时域 频域 

分 类 号：TG71[金属学及工艺—刀具与模具]