Abstract:
[Objective] Digital twin technology can improve the quality of operation management of pumping station groups in water diversion and transfer projects. However, the current application of digital twin technology in the operation management of pumping station groups remains exploratory. Moreover, certain problems persist, such as difficulties ensuring the operational performance of digital twin scenes and combining the integration of data, knowledge, and models with business, all of which hamper the application of digital twin technology. To bridge this gap, this study constructs a six-dimensional theoretical model and basic framework of a digital twin engine and proposes a high-performance digital twin engine construction method in combination with the business requirements of a pumping station group's operation management. [Methods] In terms of theoretical modeling, this paper divides the critical elements of a digital twin engine into six aspects: Physical scenes, twin scenes, services, twin data, connections, and feedback and decision-making. These elements are interrelated, thus mapping the physical scenes into twin scenes and realizing the digital twin engine's dynamic operation and virtual-real integration through services, twin data, and connections. On the basis of the theoretical model, the digital twin engine framework combines the business elements of the pumping station group's operation management. These include data collection, data storage, support, function, and display layers, which provide a data foundation, operation environment, functionservices, andinteraction windowfortheengine. Furthermore, thedigitaltwinengineconstruction methodincludesthe followingfive aspects: Twin scene generation engine, data management engine, dynamic virtual-physical mapping engine,operationsimulation and analysis engine, and realityinteraction and feedback control engine. In particular, the twin scenes generation engine generates high-performance and smooth twin scenes through lightweight BIM processing and data-knowledge-modelfusion. The data management engine builds a multiprecision, full-factor twin data resource pool. The dynamic virtual-physical mapping engine realizes real-time and dynamic iterative updating of physical scenes in twin scenes.Moreover, the operationsimulation and analysis engine supports the core business capability of the pumping station group's operation management by providing program recommendations, operation process preview, performance analysis, and other capabilitiesfor the pumping station group. The reality interaction and feedback control engine also provides the pumping station group with operational decision-making capability and a safe control environment. [Results] In practical applications through engineering cases, the digital twin engine was mainly constructed in the browser/server mode, with the desktop application as a supplement. Theresultsrevealedthatthe digitaltwinengineeffectivelysupportedthe pumpingstation group's operational management business requirements. It was found that the operation optimization capability saved 4.14% in operationcosts and 1 .59 % in energy consumption while maintaining high operation efficiency. Simultaneously, the operation scheme generated by the engine enabled a simulation preview of the entire operation process, accompanied by high-performance dynamic virtual-physical mapping. The dynamic virtual-physical mapping engine and timing database significantlyreduced the response time of data mapping, maintaining the response time for 10 000 data mappings within 300 ms. Furthermore, after undergoinglightweight processing, thetwinscene maintaineda highrunningframerate, whether deployed as a desktop application or a web-based application. Interms ofcomputerresourceconsumption, when performing a simultaneous simulation preview and utilizingthe weather system, a substantial amount of data and particle effects needed to be processed, necessitating a high-performance computer graphics processor. In contrast, digital twin engines typically operated underlow-performance stress conditions and did not demand high computer performance. [Conclusions] This work providestheoretical and methodological support, as well as serves as a practical reference, to help construct a digital twin enginethattargetsthe operation management of pumping station groupsin water diversion projects. Nonetheless, this paper describes adigitaltwinengineconstruction methodthat marksa preliminarystepinintegrating data, mechanisms, algorithms, and knowledge specific to pumping station group systems in water diversion projects. Future research should focus on enhancingthe engine's performance, determiningcomprehensiveenginefunctionsthatarecustomizedto meet various business needs, and exploringthe applications of deep mining multimodel coupling. Β© 2024 Tsinghua University. All rights reserved.
