OBRABOTKAMETALLOV MATERIAL SCIENCE Том 23 № 3 2021 EQUIPMEN . INSTRUM TS Vol. 7 No. 1 2025 In contemporary view, vibration monitoring systems enable the prediction of surface quality during cutting based on digital vibration measurements [2‑3]. However, such prediction relies on the development of complex mathematical models that capture the evolutionary dynamics of the cutting process [4]. The complexity of these models and the requirement for their parametric identification present a significant challenge. Addressing this challenge will substantially enhance the capabilities of modern metalworking systems. One promising solution involves leveraging a novel digital paradigm in metalworking control systems known as the “digital twin” [5‑7]. Specifically, the application of intelligent models that describe the complex dynamics of technological processes during metal cutting holds significant potential in this emerging field of scientific knowledge [8‑9]. For example, the work by Y. Altintas and his research groups, prominent experts in the field of digital twin-based metalworking control, proposes using digital twins to generate new CNC programs that allow part machining without preliminary setups and experiments [10]. This suggests that the selection of optimal processing modes, both in addressing current challenges and in reconfiguring the control system on a metalcutting machine (i.e., achieving flexibility), can be achieved using virtual models within a digital twin. From a modern perspective, the technology of constructing digital twins, particularly the synthesis of virtual models, relies on two primary paradigms: one based on deterministic mathematical models [10] and the other on the widespread use of neural networks [8‑9]. Acritical direction in the development of digital twin technology is the diagnosis of various malfunctions. For instance, in [11], the focus is on generating labeled training datasets for various bearing malfunctions to supplement limited measured data. The authors propose a novel digital twin approach to address the scarcity of measured data in bearing malfunction diagnosis. Experimental results demonstrate an increase in the accuracy of malfunction diagnosis [12]. A similar perspective, albeit with slight variations, is presented in [13]. The authors highlight the limitations of traditional malfunction diagnosis methods that rely on experimental data, noting that in mission-critical industrial scenarios, such data is not always available. Digital twin technology, by creating a virtual representation of a physical object that reflects its operating conditions, enables the diagnosis of technical system or technological process malfunctions even when insufficient data on those malfunctions exists. The authors propose a malfunction diagnosis system based on digital twins that leverages labeled simulated data and unlabeled measured data [13]. The construction of a digital twin system that integrates sensor data from malfunctioning bearings into the subspaces of virtual models in real time is presented [14]. The authors refine the parameters of the virtual models by comparing the results of digital modeling in the time domain with the measured and captured signals [14]. At the same time, accurately modeling the complex influence of cutting tool wear on the dynamics of the cutting process remains an extremely challenging task, which cannot be solved without considering the thermodynamic aspects of the processes occurring during metal cutting [15‑16]. Based on the analysis, digital twin technology has become widely adopted in the diagnosis of malfunctions, including bearing malfunctions. Therefore, an obvious development of digital twin technology is to leverage it for predicting the impact of cutting tool wear on the quality of the parts produced during machining. Based on this, we formulate the purpose of the study as improving the accuracy of the digital twin system’s prediction of surface quality during machining under conditions of progressive cutting tool wear through parametric identification of the digital twin’s virtual models using data obtained from a vibration monitoring system of the cutting process. Research methodology Vibration monitoring system for assessing cutting tool wear and machined surface quality The experimental component of this research utilizes a vibration diagnostic subsystem, positioned on the cutting tool itself, or rather on its holder, as depicted in Fig. 1. This system is based on an industrial IEPE (ICP) general-purpose accelerometer with an integrated A603C01T charge converter amplifier. The accelerometer has a frequency range of +/− 3 dB: 0.4‑15,000 Hz; a sensitivity of +/− 10 %: 10.2 mV/(m/s2). The system also includes a single-channel ICP (IEPE) converter with a frequency range of 0.1-50,000 Hz.
RkJQdWJsaXNoZXIy MTk0ODM1