Development of a Greybox model for wear prognosis of PVD coated carbide tools during high performance turning of steels

在切削过程中，涂层和切削材料的持续退化导致物理气相沉积（PVD）涂层刀具的热机械应力聚集。然而，目前开发的刀具磨损预测分析白盒模型是基于未涂层刀具的固定积分热机械载荷，因此没有考虑从线性到渐进磨损增加的转变。数据驱动的黑盒模型不能表示物理相互作用，当涉及到可变边界条件时，其鲁棒性与不确定性有关。因此，不可能准确预测刀具的使用寿命，也不可能对要求苛刻的加工过程中的涂层刀具进行基于知识的鉴定。该研究项目的主要目的是开发一个Greybox模型，用于准确预测高性能车削钢时PVD涂层硬质合金刀具的刀具寿命和剩余使用寿命。为此，在项目的第一阶段，提出了用于确定切削过程中安装热机械应力的分析白盒模型与数据驱动的黑盒模型之间的耦合。此外，还将深入研究温度相关涂层性能对刀具磨损过程的影响。首先，在切削刀片上沉积两种不同厚度的TiAlCrSiN和TiAlCrSiON涂层并进行表征。随后，确定了温度相关的弹塑性涂层性能和系统行为。涂层刀具将用于在干燥条件下粗车削热处理C45和42CrMo4钢。为了确定开发白盒模型所需的不同工艺参数下的热机械刀具载荷，计划进行类比切削试验。就地获取过程状态变量的切削试验，重点关注刀具线性到渐进磨损过渡区，将为刀具损伤定量分析提供实验基础。最后，将开发并验证一个Greybox模型，以在切削过程中实现在线剩余刀具寿命预测。为此，Whitebox模型和实验确定的涂层性能和刀具磨损过程数据将通过基于机器学习的Blackbox模型进行耦合。在项目的第二阶段，将进行精度、灵敏度和误差分析，以进一步提高所开发的灰盒模型的预测精度和可移植性。在此过程中，可以探索新兴的数字化加工技术以前未被发现的研究领域。

The continuous coating and cutting material degeneration during the cutting process contribute to an instationary thermomechanical stress collective of the physical vapor deposition (PVD) coated tool. However, the currently developed analytical Whitebox models for tool wear prediction are based on a stationary integral thermomechanical loading of uncoated tools, whereby the transition from linear to progressive wear increase is not considered. Data-driven Blackbox models cannot represent the physical interactions and their robustness is associated with uncertainties when it comes to variable boundary conditions. Consequently, an accurate prediction of tool service life and knowledge-based qualification of coated tools for demanding machining processes is not possible.The main aim of the research project is the development of a Greybox model for an accurate prediction of tool life and remaining service life of PVD coated carbide tools during high-performance turning of steels. For this purpose, a coupling between analytical Whitebox models for determination of the instationary thermomechanical stresses in the cutting process with data-driven Blackbox models is proposed in the first phase of the project. Moreover, the influence of temperature dependent coating properties on tool wear progress will be thoroughly investigated. Firstly, TiAlCrSiN and TiAlCrSiON coatings, each with two different coating thicknesses, will be deposited on the cutting inserts and characterized. Subsequently, the temperature dependent elastic-plastic coating properties and the system behavior are to be determined. The coated tools will be used for rough turning of heat-treated C45 and 42CrMo4 steels under dry condition. In order to determine the thermomechanical tool load at different process parameters, required for development of the Whitebox model, analogy cutting tests are planned. Cutting tests with in situ acquisition of the process state variables and focus on linear to progressive tool wear transition zone will form the experimental basis for quantitative tool damage analysis. Finally, a Greybox model will be developed and validated to enable online remaining tool life prediction during the cutting process. For this purpose, the Whitebox model and experimentally determined data on coating properties and tool wear progress will be coupled by means of machine learning based Blackbox models. In the second phase of the project, accuracy, sensitivity and error analyses will be carried out to further increase the prediction accuracy as well as the transferability of the developed Greybox model. In the process, previously undiscovered research areas for the emerging digitalized machining technology can be explored.