基于融合深度框架的齒輪加工質量在線評估方法
首發時間:2021-09-30
摘要:齒輪加工精度會影響機構的運行平穩性和耐久性,因而有必要保證每個齒輪的加工質量。然而若對所有齒輪進行人工質檢將耗費大量時間與人力成本。通過分析齒輪加工振動數據來估計齒輪加工誤差的方法可提高生產效率、保證齒輪精度,并指導工藝參數優化。為此,本文提出一種深度融合在線評估方法DFOEM用于齒輪加工誤差的監測與估計。其中,設計了一種多層特征融合殘差網絡(MFFRN),并提出一系列正則化方法以提高估計的魯棒性與泛化性。通過對多工況下的不同型號齒輪加工振動數據進行分析驗證,表明設計的MFFRN網絡具有較高的特征提取能力,且提出的DFOEM融合方法估計性能優于所有單模型方法。
關鍵詞: 機械運行與維修; 齒輪加工; 質量評估; 深度回歸; 信號處理; 融合方法
For information in English, please click here
Gear Machining Quality Estimation Method Based on Fused Deep Method
Abstract:Gear machining precision will influence the stabilityand durability of the mechanism, so it is necessary to ensure the machining quality of each gear. However, manual quality inspection for all gears will consume a lot of time and labor costs. The method of estimating gear machining quality by processing gear machining vibration signals can increase production efficiency, maintaingear quality, and assist the optimization of manufacturing parameters. Therefore, this paper proposes a deep fusion online evaluation method DFOEM for the monitoring and estimation of gear machining errors. In the method, a multi-layer feature fusion residual network (MFFRN) is designed, and a regularization methods is proposed to improve the robustness and generalization of the estimation. Through the analysis and examinationof the vibration data of different types of gears under multiple operating conditions, it is shown that the designed MFFRNnetwork has good feature extraction capabilities, and the proposed DFOEM fusion method has better estimation performance than orther methods.
Keywords: Mechanical Operation and Maintenance, Gear Machining Quality Prediction Deep Regression Signal Processing Ensemble Method
引用
No.****
動態公開評議
共計0人參與
勘誤表
基于融合深度框架的齒輪加工質量在線評估方法
評論
全部評論