Discovery of the thermal process model from noisy data

T. A. Andreeva^ab, N. Yu. Bykov^ab, A. K. Klimova^b, A. Ya. Lukin<sup>a

a</sup> Peter the Great St. Petersburg Polytechnic University, St. Petersburg, Russia
^b ITMO University, St. Petersburg, Russia

Abstract: A modification of the algorithm of the model generative design in the form of a partial differential equation for working with noisy data is proposed. Using the algorithm, the model of the heat and mass transfer process was restored from synthetic and original experimental data on heating the medium by a flooded heat source. The thermophysical parameters of the medium are determined, the possibility of using the algorithm to indicate the convection process based on data on the space-time distribution of temperature is shown.

Keywords: generative design method, data-driven model, thermal conductivity equation, convection.

Received: 11.04.2023
Revised: 25.05.2023
Accepted: 25.05.2023

DOI: 10.21883/PJTF.2023.15.55858.19588

Bibliographic databases:

•