A Hybrid Knowledge-Data Model to Driving the Self-Evolution of Building Digital Twins

The building physical entity (BPE) could produce a series of unstable dynamic changes under the influence of highly complex internal and external loads. However, the building information model (BIM) used in the current construction industry is difficult to simulate this change. In our previous work, we proposed a foundation model for building digital twins (BDTFM), which constructing the building digital twins based on it could replace BIM to solve the above difficulties. However, the BDTFM only gives the functional description of BDTs at the conceptual level and does not give specific technical methods to simulate the dynamic changes of BPE. Therefore, this paper proposes a knowledge-data-based hybrid model to drive BDTs to implementation self-evolution and dynamic optimization, realize intelligent reasoning and autonomous decision-making, to achieve the purpose of simulating the dynamic changes of BPE in the whole life cycle. At last, giving a case of foundation pit settlement prediction of an underdrain to illustrate how the theory proposed in this paper could be applied to practical engineering.