【作者】 汪洋；

【导师】 张略；

【作者基本信息】 苏州大学 ， 机械工程， 2020， 硕士

【摘要】 微结构表面是指面型精度可达亚微米级,具有特定功能微小拓扑形状的表面,由于其具有质量轻、体积小、成本低等优势,目前被广泛使用在光学领域、航空航天、军事方面等领域中。微结构表面加工的方法也成为了一个研究热点,加工微结构表面的加工技术也在飞跃发展。其中,快刀伺服加工技术由于具有能加工光滑连续的非对称三维结构、可加工材料多、加工效率高等优势,是目前加工微结构表面的主要方法之一。在微结构表面加工过程中,由于微结构表面的面型特征,在加工过程中切削深度是一个时变量,引起切削力的动态变化,从而导致加工过程的不稳定,并可能导致再生型颤振的发生。颤振的发生会产生如工件精度无法达到标准,刀具磨损等后果,因此对加工过程中的颤振进行控制具有重要意义。在本文中,首先对面向微结构表面加工的快速伺服刀架进行了总体结构设计,对其主要组成部件进行了相关的选型和计算,保证满足设计的使用要求,并在最后对设计好的整体快速伺服刀架的实际性能进行了相关的静力学仿真和动力学仿真,并与理论计算结果进行了对比,保证设计出的刀架满足使用要求并符合安全标准。接着,对于微结构表面加工过程中发生的再生型颤振进行了数学建模,并根据其数学模型进行了加工过程的稳定性分析。根据建立的颤振数学模型选取了相关的切削参数,进行切削仿真对比,观察稳定切削与颤振发生时的刀具振动情况,通过对仿真结果的分析,选取了检测颤振发生的特征量,作为判断颤振发生的依据,并设计了检测刀具颤振发生的实验方案。由于压电材料具有响应快、输出力大的优点,在本文中设计了采用压电执行器对微结构表面加工过程中的颤振进行抑制的主动控制实验方案。并结合BP神经网络控制算法,通过颤振数据的输入,输出控制电压,实现压电制动器对刀具颤振进行前馈控制。在设计对颤振进行实时反馈控制方案时,由于BP神经网络自身存在局限性,在训练过程时可能会陷入局部最优的情况,无法得到最优控制电压,因此采用粒子群算法对传统的BP神经网络算法进行了优化。并进行了两种控制方法的实验,将未控制下的颤振情况与控制后得到的振动情况进行了对比,实验结果表明,两种控制方法均得到了良好效果。更多还原

【Abstract】 Micro-structured surface refers to a surface with sub-micron level accuracy and a micro-topology shape with specific functions.Because of its excellent advantages of light weight,small size and low cost,it is currently widely used in the optical field,aerospace,military and other fields.The method of micro-structured surface processing has also become a research hotspot,and the processing technology for micro-structured surfaces is also developing rapidly.Among them,the fast tool servo machining technology is one of main methods for processing micro-structured surfaces due to its advantages of being able to process smooth and continuous asymmetric three-dimensional structures,machinable materials and high processing efficiency.During the processing of micro-structured surface,due to the surface features of the micro-structured surface,the cutting depth is a time-varying variable during the processing process,causing dynamic changes in cutting force,which results in instability in the processing process and may cause regenerative chatter happened.The occurrence of chatter vibration will have consequences such as the workpiece precision could not reach the standard,tool wear and so on.Therefore,it is of great significance to control chatter during machining.In this paper,the structural design of a fast servo tool post for micro-structured surface machining is firstly carried out,and the relevant components are selected and calculated to ensure that the design requirements are met.The actual performance of overall fast servo tool has been subjected to relevant static and dynamics simulations,and compared with the theoretical calculation results to ensure that the designed tool post meets the requirements of use and meets safety standards.Then,the mathematical model of regenerative chatter during micro-structured surface machining was performed,and the stability analysis of the machining process was performed according to the mathematical model.According to the established chatter mathematical model,the relevant cutting parameters were selected,and the cutting simulations was compared to observe the tool vibration when stable cutting and chatter occurred.Through the analysis of simulation results,the feature amount for detecting chatter occurrence was selected as the basis for judging the occurrence of chatter,and an experimental scheme for detecting the chatter of the tool was designed.Because piezoelectric materials have the advantages of fast response and large output force,active control experimental scheme that uses piezoelectric actuators to suppress chatter vibration during the processing is designed in this paper.Combined with the BP neural network control algorithm,through the input of chatter data,the output of control voltage can be obtained,the piezoelectric actuator is used to realize feedforward control of chatter of the tool.When designing a real-time feedback control scheme for chatter,the BP neural network control algorithm has its limitations.During the training process,it may fall into a local optimal situation,and the optimal control voltage cannot be obtained.Therefore,the traditional BP neural network algorithm was optimized by using particle swarm optimization.And two control experiment schemes are designed,and the uncontrolled chatter is compared with the vibration after control.The experimental results show that the two control methods have obtained perfect effect.更多还原