Multi-gene genetic programming of critical boron concentration and power peak factor for nuclear reactor fuel reload calculations

Created by W.Langdon from gp-bibliography.bib Revision:1.8414

@Article{Filho:2025:pnucene,

• author = "Marcos A. G. S. Filho and Alan M. M. Lima and Victor H. C. Pinheiro",
• title = "Multi-gene genetic programming of critical boron concentration and power peak factor for nuclear reactor fuel reload calculations",
• journal = "Progress in Nuclear Energy",
• year = "2025",
• volume = "180",
• pages = "105596",
• keywords = "genetic algorithms, genetic programming, Artificial Intelligence, Power peak factor, Nuclear reload problem, ANN",
• ISSN = "0149-1970",
• URL = "https://www.sciencedirect.com/science/article/pii/S0149197024005468",
• DOI = "doi:10.1016/j.pnucene.2024.105596",
• abstract = "Nuclear fuel reload aims to search for a core configuration of partially burned and fresh nuclear fuel, optimising the operational cycle length while assuring safety limits. For each configuration, operational cycle length and safety limits are evaluated in terms of boron concentration and power peak factor, respectively. Other safety parameters are not currently predicted with MGGP. In practice, a licensed numerical model is provided by the reactor manufacturer to estimate these physical parameters, and each configuration is simulated in approximately 300 s. Considering all possible core combinations, this approach becomes computationally unfeasible. This work introduces the Multi-Gene Genetic Programming (MGGP) to generate an explicit closed form mathematical function to estimate the nuclear reload physical parameters more efficiently. Results for a study of a single cycle of the Angra-I reactor show that the algorithmically generated mathematical function calculates boron concentration and power peak factor with a coefficient of determination of 0.997 and 0.95, respectively. For each configuration, the time of assessment is approximately 8E-4 s, which is several orders of magnitude faster than common licensed numerical tools, potentially enabling expensive optimisation studies. Also, MGGP performance is compared with an implemented Artificial Neural Network (ANN), and model results are compared",

}

Genetic Programming entries for Marcos A G S Filho Alan M M Lima Victor H C Pinheiro

Citations