Efficient Creation of Behavior Models with Variable Modeling Depths Used in Digital Twins

TL;DR

This paper introduces an automated method for creating behavior models with variable depths for Digital Twins, enabling resource-efficient modeling that maintains accuracy across different use cases.

Contribution

The paper presents a novel approach for automatically generating behavior models with varying depths, addressing a gap in existing methods.

Findings

Lower depth models nearly match higher depth models in accuracy

Significant reductions in computing time and memory usage achieved

Applicable to multiple industrial use cases

Abstract

Behavior models form an integral component of Digital Twins. The specific characteristics of these models may vary depending on the use case. One of these key characteristics is the modeling depth. Behavior models with a lower modeling depth depict the behavior of the asset in an abstract way, while those with a higher modeling depth depict the behavior in detail. Even if very detailed behavior models are flexible and realistic, they also require a lot of resources such as computing power, simulation time and memory requirements. In some applications, however, only limited resources are available. The automated creation of Digital Twins is of crucial importance for their widespread use. Although there are methods for the automated creation of behavior models for Digital Twins with a specific modeling depth, there is currently no method for the automated creation of behavior models with varying modeling depths. This article presents such an approach and demonstrates its advantages using two industrial use cases. It is demonstrated that the automatically created behavior models of lower modeling depth yield results that are almost identical to those of models with a higher modeling depth, but with significantly reduced computing time and required memory. This enables the efficient use of behavior models in a variety of use cases, regardless of the availability of resources.