哈尔滨工程大学学报
哈爾濱工程大學學報
합이빈공정대학학보
Journal of Harbin Engineering University
2015年
9期
1229-1233
,共5页
表面粗糙度预测%高速铣削%最小二乘支持向量机%粒子群算法%回归分析%预测精度%45钢
錶麵粗糙度預測%高速鐉削%最小二乘支持嚮量機%粒子群算法%迴歸分析%預測精度%45鋼
표면조조도예측%고속선삭%최소이승지지향량궤%입자군산법%회귀분석%예측정도%45강
surface roughness prediction%high speed milling%least square support vector machine%particle swarm optimization%regression analysis%prediction accuracy
为了提高高速铣削加工表面粗糙度预测的精确性以及模型的通用性,提出了一种基于粒子群最小二乘支持向量机( PSO-LSSVM)算法的高速铣削加工表面粗糙度预测方法. 以工件硬度以及铣削参数为影响因素,采用回归分析方法、最小二乘支持向量机( LSSVM)以及PSO-LSSVM方法,分别建立了45钢高速铣削加工表面粗糙度预测模型,并对模型的预测精度进行了试验验证和对比分析. 结果表明:相同样本条件下,回归分析方法的预测误差较大,PSO-LSSVM预测模型平均预测误差仅为LSSVM方法平均预测误差的50%. PSO-LSSVM预测模型具有较高的预测精度和泛化能力,能够准确地预测高速铣削不同硬度的工件表面粗糙度,同时为铣削参数的选择和表面质量的控制提供了依据.
為瞭提高高速鐉削加工錶麵粗糙度預測的精確性以及模型的通用性,提齣瞭一種基于粒子群最小二乘支持嚮量機( PSO-LSSVM)算法的高速鐉削加工錶麵粗糙度預測方法. 以工件硬度以及鐉削參數為影響因素,採用迴歸分析方法、最小二乘支持嚮量機( LSSVM)以及PSO-LSSVM方法,分彆建立瞭45鋼高速鐉削加工錶麵粗糙度預測模型,併對模型的預測精度進行瞭試驗驗證和對比分析. 結果錶明:相同樣本條件下,迴歸分析方法的預測誤差較大,PSO-LSSVM預測模型平均預測誤差僅為LSSVM方法平均預測誤差的50%. PSO-LSSVM預測模型具有較高的預測精度和汎化能力,能夠準確地預測高速鐉削不同硬度的工件錶麵粗糙度,同時為鐉削參數的選擇和錶麵質量的控製提供瞭依據.
위료제고고속선삭가공표면조조도예측적정학성이급모형적통용성,제출료일충기우입자군최소이승지지향량궤( PSO-LSSVM)산법적고속선삭가공표면조조도예측방법. 이공건경도이급선삭삼수위영향인소,채용회귀분석방법、최소이승지지향량궤( LSSVM)이급PSO-LSSVM방법,분별건립료45강고속선삭가공표면조조도예측모형,병대모형적예측정도진행료시험험증화대비분석. 결과표명:상동양본조건하,회귀분석방법적예측오차교대,PSO-LSSVM예측모형평균예측오차부위LSSVM방법평균예측오차적50%. PSO-LSSVM예측모형구유교고적예측정도화범화능력,능구준학지예측고속선삭불동경도적공건표면조조도,동시위선삭삼수적선택화표면질량적공제제공료의거.
In order to improve the accuracy and application scope of a surface roughness prediction model, such a model, for high speed milling, is proposed based on the particle swarm optimization-least square support vector ma-chine ( PSO-LSSVM) method. By regarding the hardness of workpieces and the milling parameters as the influence factors on the model, based on regression analysis, LSSVM and PSO-LSSVM, the prediction models of surface roughness in high speed milling of 45 steel were established, then the prediction accuracy of the models was com-pared and verified through experiments. The results show that under the same sample conditions, the mean predic-tion error of the PSO-LSSVM model is only 50% of the LSSVM model. Therefore, the prediction model established based on PSO-LSSVM has a high prediction accuracy and generalization ability. It can predict the surface roughness for workpieces with different hardnesses precisely and can provide the basis for proper selection of milling parame-ters and control of surface quality.
