Development of a Greybox Model for the Prediction of the Performance of PVD Coated Carbide Tools

根据目前的技术水平，PVD涂层工具的真实、复杂的应用行为既不能令人满意地测量，也不能在模型中充分描述。为了能够对这些刀具进行基于知识的鉴定，有必要获得有关失效开始、磨损进展和剩余刀具寿命的知识。特别是，由于缺乏对基本关系的了解，不能模拟不连续的涂层失效。此外，由于磨损，涂层的性能和施加的热机械载荷都会随着运行时间的增加而变化。提出的研究项目一方面解决了C45N和42CrMo4+QT合金的可变涂层性能和可变工艺相关载荷之间的动态关系，另一方面解决了外部纵向车削的动态关系。从现有技术中已知，涂层性能对工具的应用行为有显著影响。众所周知，由于刀具磨损，涂层性能和热机械载荷应力都会发生显著变化。考虑到不同的特性和载荷，可以更准确地预测刀具的性能，并有助于建立对以特定载荷方式设计和选择刀具的基本理解。为此目的，在研究项目开始时，就涂层和涂层与硬质合金衬底之间的界面特性对工具进行了广泛的表征。随后为随后的建模制定了详细的数据基础。借助于一种基于对滑行试验台的调查并结合显微运动成像的新方法，首次有可能绘制作用于刀具磨损过程中的刀具刃口上的载荷应力。与同样随时间变化的热负荷一起，这些负荷应力作为有限元模型的输入变量，用于映射导致涂层失效的切割楔体中的临界应力。通过使用人工智能方法，扩展了该白盒模型的预测能力和精度。由于很难根据物理关系映射涂层性能的变化并将其用于这样的预测，因此采用了基于涂层属性和残余应力的高分辨率分析的数据驱动方法。通过观察应力和涂层性能之间的关系，可以预测工具的进一步行为，并获得对潜在机制的新见解。

According to the current state of the art, the real, complex application behaviour of PVD-coated tools can neither be measured satisfactorily nor described sufficiently in models. In order to be able to conduct a knowledge-based qualification of these tools, it is necessary to gain knowledge about the start of failure, the progress of wear and the remaining tool life. In particular, discontinuous coating failure cannot be modelled due to a lack of knowledge of the fundamental relationships. In addition, both the coating properties and the thermomechanical loads applied change with increasing operating time as a result of wear. The proposed research project addresses the dynamic relationship between variable coating properties on the one hand and the variable process-related loads on the other hand for external longitudinal turning of the alloys C45N and 42CrMo4+QT. It is known from the state of the art that the coating properties have a significant influence on the application behaviour of the tools. It is also known that both the coating properties and the thermomechanical load stresses change significantly due to tool wear. Taking into account the varying properties and loads allows for a more accurate prediction of the performance of the cutting tools and helps to build up fundamental understanding to design and select tools in a load-specific way. For this purpose, the tools are extensively characterised at the beginning of the research project with regard to coating and interface properties between the coating and the carbide substrate. This is followed by the development of a detailed data basis for the subsequent modelling. With the help of a new method based on investigations on a planing test rig in combination with microkinematography, it is possible for the first time to map the load stresses acting on the cutting edge as tool wear progresses. Together with the thermal load, which also changes over time, these load stresses serve as input variables for an FEM model for mapping critical stresses in the cutting wedge that promote coating failure. The predictive capability and accuracy of this whitebox model is extended by the use of artificial intelligence methods. Since it is difficult to map the changed coating properties on the basis of physical relationships and use them for such a prediction, data-driven approaches based on high-resolution analyses of the coating properties and residual stresses are employed. By observing the relationship between stresses and coating properties, it is possible to predict the further behaviour of the tools and gain new insights into the underlying mechanisms.