The concept of a multi-level system of digital twins (on the example of geomagnetic data)

The problem of spatial interpolation of data with the required accuracy is especially relevant for processes and phenomena of natural origin, which are characterized by pronounced spatial anisotropy of information. An example of this is geomagnetic data as a result of observing the parameters of the Earth's magnetic field and its variations. Technical means of recording such data, as a rule, are limited by the possibilities of physical placement due to the high requirements for the hardware and software necessary for information and measurement monitoring, as well as the associated economic and time costs. The impossibility of placing monitoring technical means in certain geographic locations increases the information entropy of the processes that take place there and are relevant for study and observation. The solution to this problem is partly possible by using modern geoinformation technologies, the algorithms of which allow interpolating spatial data by known values. However, the accuracy of spatial interpolation is not always satisfactory: in particular, the restoration of geomagnetic data in this way provides a root mean square error of 7.3 to 11.2 nT for various spatial regions, which significantly exceeds the error of 1 nT permissible by the standards. In this regard, in order to solve the indicated problem, the concept of a multi-level system of digital twins of technical objects is proposed. The concept is based on the idea of multiple repetition of pairs "digital twin - physical prototype" to form a hierarchical multilevel structure of fractal type. Duplex inter-level information interaction is carried out through the use of specialized data stores that aggregate information coming from the components of the level (while the simultaneous presence of both elements of the pair is optional). The physical prototypes are both directly technical objects and pairs of the form "digital twin - physical prototype", as a result of which the structure formed in this case functions according to the principles of information backup. The sequential creation of the levels of the system of digital twins allows to significantly increase the coverage area of the earth and near-earth surfaces by physical and virtual technical means, the functioning of which is modeled using machine learning methods that form interpolated values based on retrospective analysis of known data. It is expected that for geomagnetic data, the use of a multilevel system of digital twins will allow collecting information on the state of the geomagnetic field and its variations in hard-to-reach geographic locations, for example, in the Arctic zone, where the placement of magnetometric equipment is extremely difficult. Studies have shown that the application of the proposed concept for solving the problem of spatial interpolation of geomagnetic data provides a root-mean-square error from 0.53 to 1.02 nT, which falls within the range regulated by the standards.

Keywords

Authors

References