Kollarov O. Control of separate power-supply system with hydrogen-oxygen cells on basis of artificial neural networks

Українська версія

Thesis for the degree of Candidate of Sciences (CSc)

State registration number

0413U006770

Applicant for

Specialization

  • 05.09.03 - Електротехнічні комплекси та системи

07-11-2013

Specialized Academic Board

Д 11.052.02

Essay

Object of research - the energy processes in the separate power-supply system with a hydrogen- oxygen fuel cell as a primary source of electricity. Research objective - increase of transfer efficiency of the maximum power (energy) to a loading from the hydrogen-oxygen cell by separate power-supply system. In solving the problems were such methods used: formalization and systematic approach, the theory of artificial intelligence, automatic control theory, mathematical analysis, approximation theory and the theory of series, hill-climbing method. Laboratory study have been performed at laboratory facility of autonomous power supply system with a hydrogen- oxygen fuel cell power of 300 W. The practical value of dissertation is the calculate mode of artificial neural networks weights based on the use of Fourier-series convergence theorem, which guarantees the convergence of the artificial neural network to the target function and the topology determining mode of artificial neural networks based on the kind of mathematical models of identifiable objects which possible to identify the physical state variables of a hydrogen- oxygen fuel cell. Scientific novelty: further developed intelligent control methods of separate power-supply system through the development of the extremal control systems structure, including the model of the hydrogen-oxygen fuel cell based on artificial neural networks, which allowed to support matching mode of separate power-supply system with a load; improved mathematical model of separate power-supply system through the finding of dependence between the diffusion voltage drop, ionic conductivity of the membrane and the pressure drop across the electrodes of hydrogen-oxygen fuel cell, which allowed it to take into account the geometrical parameters of hydrogen-oxygen fuel cell in the general mathematical model of the system; were further developed polynomial identification methods of hydrogen-oxygen fuel cells through the development of piecewise periodic Gauss polynomials and orthogonal function system on their basis, which allowed the identification of a hydrogen-oxygen fuel cell in the separate power-supply system. The results of the dissertation introduced in the LC "Dontehprom" in Donetsk, and are used in scientific research and in the educational process at the SHEE "Donetsk national technical university" Board of Education. The scope: electrotechnical.

Files

Similar theses