【作者】 张文思；

【导师】 张家梁；

【作者基本信息】 东华大学 ， 机械制造及其自动化， 2007， 硕士

【摘要】 在切削过程中，任何故障都将严重影响机械产品质量、生产效率，扰乱正常的生产计划和增加生产成本，因此，进行切削过程监测优化，对于提高机床利用率和产品质量、降低废品率和加工成本、减轻劳动强度、降低材料消耗和提高生产安全起着重要作用。本文详细论述了切削过程监测及其优化的信号处理方法及其相关算法，其中信号处理方法包括信号预处理、时域分析与建模、频域分析与建模、小波分析；监测算法包括模糊理论、神经网络算法；优化算法包括遗传算法，为切削过程监测及其优化奠定基础；鉴于切削力信号在切削过程监测与优化中的重要性，以Labview为工具，通过硬件选取、传感器标定、软件编制(包括采集处理界面、d11驱动程序编制等)实现基于PCI和USB的两套切削力数据采集平台的开发，为后续的切削力试验模型、优化、刀具磨损监测研究作好铺垫。在切削过程优化方面，通过试验分析了切削用量对切削力的影响情况，并在此基础上，提出基于最小二乘回归切削力数学模型和基于径向基神经网络的切削预测模型，并用实验验证了其可行性与有效性；提出在满足加工质量要求的前提下效率最大的切削工艺参数优化方案并利用遗传算法(GA)完成优化，并通过试验验证了其有效性和可行性。在切削过程监测方面，以切削过程监测中的关键技术之一—刀具状态监测技术展开相关研究。通过试验和理论两个角度，研究刀具状态对稳态和动态切削力信号的影响情况；利用小波分析方法研究AE信号的特征量与刀具状态的关联性；采用时域自回归模型(AR)研究刀具状态对振动(AC)信号的影响情况，以提取能够反映刀具磨损状态的特征量。将AE信号的第六层小波分析系数均方值、振动信号的五阶自回归模型残差、切削力信号中刀具弯曲振动频率14kHz对应的幅值组成特征矢量，并借助模糊理论实现对刀具磨损的监测。经实验验证，利用融合特征向量并通过模糊判决识别方法对刀具切削状态识别有着较高的准确率和可靠性，特别适合中后期磨损刀具状态。相比传统的刀具磨损监测手段，将声发射信号、振动信号、切削力信号的特征量组成特征矢量，并利用模糊理论识别刀具状态更具有实用性和可靠性。更多还原

【Abstract】 In the cutting process, efficiency of processing and the quality ofproducts will be badly influenced by any faults of machine tools andequipments and the cost of operation will be added. Consequently, Usingthe monitoring and optimization to the processing of machining has theimportant effect on improving efficiency of the lathe and quality ofproduct, reducing the rate of waster, the cost of machining and theconsume of the materials, easing the intension of labor and improvingthe safety of the produce.This paper particularly introduce signal processing technologyincluding pretrement methods, analysis and modeling of time-domain,analysis and modeling of frequency-domain and wavelet analysis;monitoring algorithm including fuzzy theory and ANN; optimizationalgorithm including GA. These signal processing technology andcorrelative algorithm act as the foundation of for cutting processmonitoring and its optimization. Because of the importance of cuttingforce, the paper achieve the development on two systems of cuttingforce acquisition based on PCI and USB by Labview according to threeparts which are choice of hardware, demarcating of sensor and writingof software. These systems act as the foundation of development on test model of cutting force, processing optimization and tool wearmonitoring.In cutting process optimization, based on the analysis of cuttingdosage’s effect on cutting force, put forward that cutting force forecastmodel based on Least Two-multiply regression and RBF ANN andvalidate its feasibility and feasibility; besides, bring forward that cuttingparameters optimization scheme aiming to maximal efficiency satisfyingquality and achieve its by GA. Experimentation prove that it has muchmore theory visualizability and maneuverability.In cutting process monitoring, the object of development is tool wearcondition monitoring which is the key technology of cutting processmonitoring, according the testing and academic angles, research the toolwear condition’s effect on steady and dynamic cutting force; research therelevancy between the AE signal’s characteristic parameters and toolwear; research tool wear condition’s effect on AC signals by AR modelso as to acquire the characteristic parameters which can effect the toolwear. Based on the characteristic vector that reflect tool wear conditionis made of those characteristic parameters which are sixth waveletanalysis coefficient of AE signals, remnant of fifth rank AR model ACsignals, breadth value of curly libration frequency of tool (14kHz).Achieve the tool wear state identification by fuzzy theory method.Experimentation prove that it has much more theory visualizability and maneuverability than the usual prediction method which based on ANN,especially suit subsequent tool wear monitoring.更多还原