Nuclear Systems Safety Analysis by Artificial Intelligence, Meta-Modeling and Adaptive Simulation

Di Maio, Francesco; Pedroni, Nicola; Zio, Enrico

Permalink

https://hdl.handle.net/2142/121825 Copy

Description

Title

Nuclear Systems Safety Analysis by Artificial Intelligence, Meta-Modeling and Adaptive Simulation

Author(s)

• Di Maio, Francesco
• Pedroni, Nicola
• Zio, Enrico

Issue Date

2023

Keyword(s)

• Nuclear safety
• Simulation
• Artificial intelligence (AI)
• Adaptive simulation
• Uncertainty quantification

Abstract

For the safety analysis of nuclear systems, the response under different perturbed conditions must be studied. This can be done by means of mathematical models, implemented in corresponding Best- Estimate (BE) computer codes for numerical simulation. Given the uncertainties in the operational conditions and in the models and parameters, the analysis entails repeated system response simulations under different conditions and parameters settings to identify unsafe states of operation. The feasibility of the analysis is challenged by the fact BE codes are: i) computationally demanding (i.e., require a long time to simulate a single scenario); ii) high-dimensional (i.e., involve large number of inputs and/or outputs); iii) somewhat black-box (i.e., the mathematical function underlying the input-output relation is eventually not fully explicit and nonlinear); iv) dynamic (i.e., evolve in time); and v) affected by uncertainties (due to the scarcity of data available for parameters calibration)1. Advancements in the fields of Artificial Intelligence, Meta-Modeling and Adaptive Simulation can provide an efficient way for computationally affordable and accurate nuclear safety analysis. In this work, different computational methods are presented for efficiently tackling the computational issues related to nuclear systems safety analysis.

Type of Resource

text

Language

eng

Handle URL

https://hdl.handle.net/2142/121825

Owning Collections