Greybox models for in-situ quantification of probabilistic tool life of CVD coated cemented carbide gear hobs

点击翻译按钮获取中文摘要

The overall objective of the present project proposal is the detailed description of the thermomechanical tool loading during gear hobbing and the resulting tool wear of CVD-coated cemented carbide tools by using deterministic models coupled with in-situ acquired process data and machine learning models to enable service life optimized regulation of the process. Gear hobbing is an established process for gear manufacturing. The load spectrum in gear hobbing leads to abrasive tool wear, which, when increased, negatively affects the dimensional accuracy of the workpiece as well as the stability of the tool cutting edge. Local inhomogeneities in the microstructure of the hard material coating and substrate as well as varying parameters in the coating microstructure result in probabilistically distributed service lives, which change in-situ depending on the real, complex operating conditions. By predicting the remaining service life of a tool depending on the material strength and the stress during machining, the total service life of coated tools can be significantly increased by closed-loop control. To date, the onset of failure, wear progress and remaining service life cannot be identified or predicted with sufficient certainty either by modeling or experimentally. In order to gain a deeper understanding of the process-microstructure-service life relationship, the tribological system is therefore being comprehensively investigated in the present project proposal, taking into account the material system and the manufacturing-related residual stresses. The focus is on gear hobbing tools made of cemented carbide, which are coated with a TiCN base layer and an α-Al2O3 top layer by CVD. These are first characterized in an analogy test. Based on this, a micromechanical model of the coating-substrate microstructure is built up, which enables the determination of the manufacturing-related residual stresses on the one hand and the simulation of the abrasive tool wear on the other hand. From deterministic white-box models for the prediction of stress and strength of the hob as well as by in-situ process data and black-box models, grey-box models are developed, which allow a material-sensitive prediction of the (remaining) tool life within the process time.