Full Article: PDF
Scientific Object Identifier: http://s-o-i.org/1.1/TAS-05-145-36
DOI: https://dx.doi.org/10.15863/TAS.2025.05.145.36
Language: English
Citation: Aptukov, M.I., & Kozhevnikov, V.A. (2025). Development of a program for temperature regime configuration of vegman servers using optimization algorithms. ISJ Theoretical & Applied Science, 05 (145), 285-300. Soi: https://s-o-i.org/1.1/TAS-05-145-36 Doi: https://dx.doi.org/10.15863/TAS.2025.05.145.36 |
Pages: 285-300
Published: 30.05.2025
Abstract: The article discusses a software version of setting the temperature regime of VEGMAN servers using optimization algorithms. The object of the research is the VEGMAN server, while the subject is its thermal operating mode. The aim of the work is to develop a program that implements an optimal algorithm for selecting PID-controller coefficients, ensuring a stable server temperature at the minimally required fan speed. To achieve this aim, the following tasks were addressed: – meta-heuristic optimization algorithms were studied, including particle-swarm optimization, ant-colony optimization, Bayesian optimization, and the genetic algorithm; – a Python program implementing these algorithms was developed; – the program was tested on VEGMAN servers; – a comparative analysis of the algorithms was carried out, and the one most effective by the combined criteria was selected (lowest post-tuning oscillation amplitude, shortest tuning time, fastest convergence to the target temperature, minimal fan speed); – the results of PID-coefficient tuning were visualised. The research was conducted at LLC “KNS Group”, where practical material was collected. Testing took place on LLC “KNS Group” servers using the YADRO Kiwi system. During the project, a program was created for automated tuning of the VEGMAN server cooling system based on optimizing the parameters of PID controllers that drive the system fans. As a result, PID-coefficient values were identified that ensure efficient cooling-system performance. Improved fan operation sustains an optimal thermal regime for VEGMAN servers, enhancing their overall reliability.
Key words: VEGMAN servers, meta-heuristic optimization algorithms, PID-coefficients, experiment, metrics, frameworks.
|
