Spatial-temporal traceability for cyber-physical industry 4.0 systems

Abstract

<p>The COVID-19 outbreak has posed significant challenges to end-to-end global supply chain visibility and transparency, with city lockdowns, factory shutdowns, flight cancellations, cross-border closures, and other uncertainties, disruptions, and disturbances. To address these challenges, reliable and accurate spatial-temporal information of physical objects and processes is essential to understand the industrial context and predict potential risks or bottlenecks for further decision-making. Product traverse both indoor (e.g., shopfloors and warehouses) and outdoor (during transportation) contexts. Despite significant advances in spatial-temporal traceability for outdoor environments using Global Positioning System (GPS) and Geographic Information Systems (GIS), satisfactory performance has not yet been achieved in indoor context, which accounts for the majority of operations. This limitation results in disjointed visibility and inaccessible transparency across the holistic supply chain. This research introduces universal and interoperable spatial-temporal elements for cyber-physical industrial 4.0 systems (CPIS) and develops a multi-modal bionic learning (MMBL) method for accurate and enduring indoor positioning. Proximity, mobility, and contextual reasoning mechanisms are designed to capture the interplay, evolution, and synchronization among objects at the operations level. To validate and evaluate the effectiveness of the proposed solution, we first conduct laboratory experiment and then apply the method in a real-life case company. Comparative analysis is conducted. MMBL clearly outperforms the other methods with 95% of the errors are within 3.41 m and maintains effectiveness after a year of use, which represents a significant step forward in achieving spatial-temporal traceability in CPIS.</p>