Grey Wolf Optimizer in Design Process of Stable Neural Controller – Theoretical Background and Experiment
- 385 Downloads
This article deals with an adaptive neural controller applied for a nonlinear plant with time-varying parameters. The structure of the controller is based on Radial Basis Function Neural Network. The output part of the controller (weights) is modified in several iterations of the control structure. In this application, the coefficients of the Gaussian functions are constant (it means the centers and width). The relevance of proper selection of those values is presented in tests performed for a real plant (an electrical drive). Moreover, for optimization of this part of the controller the metaheuristic – Grey Wolf Optimizer – algorithm was applied. The centers were selected in a clustering process. The synthesis of the controller includes stability analysis (using the Lyapunov method). The content of this article can be divided into two basic parts, the first shows theoretical considerations and the second is related to the experimental tests of the analyzed neural controller (executed in a laboratory, for the drive with 0.5 kW nominal power, using dSPACE card).
KeywordsGrey Wolf Optimizer Radial Basis Function Neural Network Clustering Adaptive control Electric drive
- 11.Pajchrowski, T.: Porównanie struktur regulacyjnych dla napędu bezpośredniego z silnikiem PMSM ze zmiennym momentem bezwładności i obciążenia. Przegląd Elektrotechniczny 94(5), 133–138 (2018)Google Scholar