基于最小二乘支持向量机的在线工件厚度辨识  

Online Estimation on Workpiece Height Based on Least Squares Support Vector Machine

作　　者：豆尚成[1] 奚学程[1] 赵万生[1] 

机构地区：[1]上海交通大学机械与动力工程学院/机械系统与振动国家重点实验室,上海200240

出　　处：《电加工与模具》2013年第4期23-27,共5页Electromachining & Mould

基　　金：国家高技术研究发展计划(863计划)资助项目(2009AA044202);国家重大科技专项(2009ZX04003-011);国家自然科学基金资助项目(51175337);中央高校基本科研业务费专项资金资助项目(HIT.KLOF.2010011);机械系统与振动国家重点实验室课题资助项目(MSVMS201111)

摘　　要：电火花线切割加工变厚度工件时的断丝和效率低下现象是线切割加工需解决的主要问题。根据工件厚度选择加工参数可避免加工过程中的断丝现象,并提高加工效率。基于工件三维模型提取的工件厚度不但能直接用于指导加工过程中的参数选取,还能用于在线修正工件厚度辨识模型。基于最小二乘支持向量机的在线算法动态更新工件厚度辨识模型,可使模型包含的信息更丰富,辨识精度更高。在加工没有三维模型的工件时,利用不断修正的辨识模型也可估计出准确的工件厚度用于指导加工过程。实验数据验证了该在线算法的有效性,辨识精度小于2 mm。When machining workpieces with a variable height, wire breakage and low machining efficiency are the two major issues. For variable height WEDM, the workpiece height must be provided so that appropriate machining parameters can then be determined according to it. The workpiece height extracted from the 3D model can be directly applied to the CNC system. Not only can this data be used to control the machining process in progress, but also can be used to update the workpiece height estimation model online. The update of the workpiece height estimation model was achieved by using the online least square support vector machine. By learning the new dynamics which have not been excited before, the height estimation model can be refined, thus improving its prediction accuracy. Experimental results show that, by online learning, the estimation model becomes more accurate and a much higher estimation precision is thus obtained. The value of estimation errors at most parts of the workpiece is less than 2mm.

关 键 词：电火花线切割加工 最小二乘支持向量机 工件厚度辨识 

分 类 号：TG661[金属学及工艺—金属切削加工及机床]