基于支持向量机钛合金铣削力预测分析  

Prediction Analysis of Titanium Alloy Milling Force Based on Support Vector Machine

作　　者：向国齐[1] 陆涛[2] 

机构地区：[1]攀枝花学院资源与环境工程学院,四川攀枝花617000 [2]西华大学机械工程与自动化学院,四川成都610039

出　　处：《机床与液压》2016年第3期142-146,共5页Machine Tool & Hydraulics

基　　金：攀枝花市科学技术和知识产权局(0290100061);多学科设计优化中代理模型与智能算法研究(2013JYZ009)

摘　　要：钛合金材料广泛应用于各个领域,其材料在加工过程中受铣削力影响易于产生变形而影响加工质量,为此需对铣削力进行预测分析。针对实际加工工程中铣削力函数不能显式表示的问题,提出一种基于支持向量机铣削力模型预测的方法。利用正交试验设计选取合适的设计参数样本点建立铣削力预测模型,并获得预测值与实验值的拟合曲线,试验值通过有限元建模获得,分别对预测值与试验值结果进行误差率及显著性检验分析。为验证支持向量机方法的有效性,建立BP神经网络模型对试验值预测。与BP神经网络模型预测比较,结果显示支持向量机模型预测的结果更能精确预测。Titanium alloys are widely used in various fields,the processing quality of this materials will be affected by the milling force and deformation easily caused,so that the milling force should be predicted. Pointing on the problem that the milling force function not explicitly expressed in practical machining process,a prediction algorithm is proposed for milling force based on support vector machine( SVM). The milling force prediction model was established by using the orthogonal experiment method to select suitable design parameters sample,the prediction and experimental fitting curves were worked out,experimental values were obtained by using Finite Element Modeling( FEM),and the error rate and its significant testing analysis of the predicted values and the experimental results were made respectively. In order to test the effectiveness of the proposed algorithm of SVM,back progation neural network( BPNN) was also used to establish the predict model for experimental valuse. Prediction results show that the predication results are more accurate by using SVM,comparing with the BPNN.

关 键 词：铣削力 BP神经网络 支持向量机 正交试验 

分 类 号：TG501[金属学及工艺—金属切削加工及机床] TH12[机械工程—机械设计及理论]