This article discusses the main problems that arise when constructing digital twins (DT) as models of the subway stations and their elements based on data processing using machine learning methods. The subject area is...
详细信息
This article discusses the main problems that arise when constructing digital twins (DT) as models of the subway stations and their elements based on data processing using machine learning methods. The subject area is considered and shortcomings are identified in the applied methods for constructing digital twins when information is provided by several sources of various types: video cameras, microphones, sensors. The problems of rebuilding digital twins when new information arrives are discussed. To ensure the possibility of building and rebuilding DTs, it is proposed to use the development environment Xcode and operation system macOS, as their usage allows reduce the time for building and rebuilding models, and to use built-in libraries from Apple, which make it possible to significantly simplify data processing based on machine learning tools. In the proposed solution, the construction and reconstruction of digital twins is carried out on Apple chips M2 and M1, data processing is performed by Apple libraries on the RISC-M2 chip. The increase in performance is 45.9%, the accuracy of event identification increases by 15.3%. The proposed solution can be applied not only in the discussed subject area, but also in building highly loaded systems, such as natural gas buses, trolleybuses, trams, commuter trains and subways were high speed of operation of monitoring systems of large objects with complex structure and behavior are required.
暂无评论