基于GA-BP算法的刀具磨损预测模型  

Tool Wear Prediction Model Based on GA-BP Algorithm

作　　者：毕长波 王宇浩 马廉洁 蔡重延 李孛 张东升 吕鑫 孙智超 邓航 BI Chang-bo;WANG Yu-hao;MA Lian-jie;CAI Chong-yan;LI Bei;ZHANG Dong-sheng;LV Xin;SUN Zhi-chao;DENG Hang(School of Control Engineering,Northeastern University at Qinhuangdao,Qinhuangdao 066004,China;Suizhong Minghui Industrial Technology Inc,Huludao Liaoning 125205,China)

机构地区：[1]东北大学秦皇岛分校控制工程学院,河北秦皇岛066004 [2]绥中明晖工业技术有限公司,辽宁葫芦岛125205 [3]秦皇岛鸿泰科技股份有限公司,河北秦皇岛066000

出　　处：《组合机床与自动化加工技术》2018年第10期145-146,150,共3页Modular Machine Tool & Automatic Manufacturing Technique

基　　金：国家自然科学基金资助项目(51275083)

摘　　要：通过氟金云母车削实验,测试了刀具体积磨损量,分析了刀具磨损量与切削参数的关系。利用GA-BP算法预测,通过最小二乘法进行数值拟合。建立刀具磨损量关于切削速度,进给速度,切削深度的一元模型,以可决系数检验模型精度,结果表明一元模型具有较高可靠性。在此基础上提出多元模型,利用遗传算法对多元模型进行求解并进行了实验验证,验证结果表明多元模型较为精确。Through experiments for turning fluorophlogopite,the volume wear was tested,and the relationship between tool wear and cutting parameters was analyzed.GA-BP algorithm was used to predict,and the numerical fitting was done by the least square method.A series of one-dimensional models were built on cutting speed,feed speed and cutting depth.The accuracy of the model was verified by the coefficient of correlation.The results show that the model has high reliability.A multivariate models was proposed on the basis of the one-dimensional models.The genetic algorithm was used to solve the multivariate model and the experimental verification was carried out.The verification results show that the multivariate model is accurate.

关 键 词：遗传算法 BP神经网络 刀具磨损 车削 可加工陶瓷 

分 类 号：TH122[机械工程—机械设计及理论] TG711[金属学及工艺—刀具与模具]