Identifying Parameters of Reactor Operation Digital Twin by Using Swarm and Evolutionary Algorithms

Abstract Digital twin (DT) technology is currently gaining popularity in a variety of industrial applications. Recently, the China Institute Nuclear Power proposed data driven by nuclear reactor digital twin. The solves forward problem for physical field simulation and inverse parameter identification. can be viewed as black-box optimization where input parameters are identified Latin hypercube sampling (LHS) method based on given partial observation. While LHS relatively naive, this paper, we to apply swarm evolutionary algorithms solve problem. Specifically, present an experimental comparative study four popular SEAs, Particle Swarm Optimizer (PSO), Differential Evolution (DE), Evolutionary Strategy (ES), evolution strategy with covariance matrix adaptation (CMA-ES). results show that, problem, SEAs more accurate efficient than LHS. We also investigate algorithm’s searching capability under conditions noisy observations no initial guesses. that DE CMA-ES have strong robustness high ability.