A Simulation Fidelity Assessment Framework

Abstract

A proposition for the correct by design of abstraction with respect to the simulation objectives based on the concepts of approximate bisimulation, Galois connections and ontology is presented. It addresses the fundamental problem of fidelity in simulation, namely, given a class of models and a class of properties that must be preserved, how to extract modeling abstractions that preserves the properties of interest which allows to conclude about the system being simulated. Fidelity and validity are explained in the framework of the experimental frame and discussed in the context of modeling abstractions. A formal method for the fidelity quantification is explained by abstraction inclusion relations for syntactic and semantic compatibility. Abstraction inclusion in dynamic systems for semantic compatibility by approximate bisimulation and the problem of finding surjection maps compatible with simulation objectives are discussed. Syntactic compatibility is explained by ontologies followed by a brief discussion on the Galois connections and building Galois surjections compatible with the simulation objectives at the end.

Graph models for engineering design: Model encoding, and fidelity evaluation based on dataset and other sources of knowledge

“…Models are developed to serve specified purposes and use. The objectives of the model, the designer, and relevant applications must be considered when developing engineering models (Ponnusamy et al, 2014).…”

“…The measure of exactness of the representation in simulation is often referred to as fidelity. To measure fidelity, adequate referencing and quantification are necessary for the modeling process (Ponnusamy et al, 2014).…”

“…Standardized measurement and specification of fidelity can support the validation of the virtual representations. Fidelity specification standardization initiatives for modeling and simulation have been investigated in literature for fidelity analysis and formalization in literature (Roza et al, 1999;Shephard et al, 2004;Ponnusamy et al, 2014;Roca, 2021). Ponnusamy et al (2014) present a simulation fidelity assessment framework to enable the verification and validation of simulation products to comply with the specifications and fitness for their intended use.…”

“…The integration and exploitation of data in the DACM framework can provide a standardized way to approach early design reasoning; however, the usability of the models may be affected by the model fidelity. The objectives of the model and the designer, and relevant applications must be considered when developing engineering models (Ponnusamy et al, 2014). As mentioned above, a valid causal ordering is a necessary condition for qualitative reasoning using the DACM framework.…”

“…Like fields such as metrology, which is constructed around standardized measurement approaches and fundamental concepts, the design of simulation models for engineering purposes, must also be facilitated by formalization and conceptualization to enable fidelity assessments. The problem has been recognized and an effort to formalize the discipline is ongoing in literature (Gross, 1999;Gross et al, 1999;Ponnusamy et al, 2014). In line with existing research, the current article develops a standardized process for machine learningbased modeling to develop simulation models and measure fidelity.…”