A Study on Service-Oriented Digital Twin Modeling Methods for Weaving Workshops

Abstract

With the rapid development of intelligent manufacturing, Digital Twin technology, as an advanced tool for the intelligentization of weaving workshops, has endowed weaving services with real-time simulation and dynamic optimization capabilities while also placing higher demands on the digital capabilities of workshops. The diverse and multi-manufacturer equipment in weaving workshops exacerbates the complexity of multi-source heterogeneous data. Moreover, traditional data collection methods, which are mostly based on fixed frequencies, increase the network load during real-time high-frequency data reception, making stable, long-term operation difficult. Conversely, low-frequency collection might miss important state changes, thus affecting the quality of weaving big data. To address these issues, this paper proposes a service-oriented Digital Twin modeling method for weaving workshops. This method combines OPC Unified Architecture (OPC UA) with a state change-based data collection approach, utilizing a sliding time window (STW) to identify anomalous data and employing median interpolation to correct these anomalies. The goal is to enhance the representation capability of the Digital Twin in the weaving workshop by improving the data quality. For a specific service of predicting the warp-out time of 288 air-jet looms in a workshop, the average error of the predicted warp-out time using the dynamic data set proposed in this study was reduced from 0.85 h to 0.78 h compared to the static data set based on fixed frequency, an improvement of 8.2%, thereby validating the effectiveness of the proposed method.