No abstract

Original scientific paperIn this study, an attempt is made to present the application and comparative performance analysis of optimal control approach for automatic generation control (AGC) of electric power generating systems. Optimal controller is designed utilizing performance index minimization criterion. To conduct the study, various single and multi-area models with/without system nonlinearities from the literature are simulated under sudden load perturbation. In this comparative study, to corroborate the worth of optimal controller, the performance of optimal AGC controller is compared with that of I/PI controller optimized adopting recently published the best established techniques such as teacher learning based optimization (TLBO), differential evolution (DE), genetic algorithm (GA), particle swarm optimization (PSO), hybrid bacteria foraging optimization algorithm-PSO (hBFOA-PSO), craziness based PSO (CBPSO), firefly algorithm (FA), krill herd algorithm (KHA), moth-flame optimization (MFO), glow swarm optimization (GSO), simulated annealing (SA), bat algorithm (BA), stochastic fractal search (SFS) and hybrid SFSlocal unimodal sampling (hSFS-LUS) technique. The simulated results are compared in terms of settling time (ST), peak undershoot (PU)/overshoot (PO), various performance indices (PIs), minimum damping ratio (ξmin) and system eigenvalues. A sensitivity study is conducted to certify the robustness of optimal controller.Komparativna analiza primjene optimalnog upravljanja za automatsko upravljanje sustavima za proizvodnju električne energije. U ovom radu se razmatra primjena i komparativna analiza sustava za automatsko planiranje proizvodnje proizvoača električne energije. Sinteza optimalnog regulatora proporcionalno-integralne strukture je provedena korištenjem integralnih kriterija. Različiti modeli s jednim područjem i više područja te s i bez nelinearnosti korišteni su u simulaciji nagle promjena opterećenja. Kako bi se pokazala važnost optimalnog regulatora, u komparativnoj analizi su performanse dizajniranog optimalnog regulatora usporeene s peformansama postignutim korištenjem I i PI regulatora sintetiziranih primjenom postojećih uobičajeno korištenih metoda kao što su teacher learning optimization, diferencijska evolucija, genetski algoritam, optimizacija rojemčestica, hybrid bacteria foraging optimizacijski algoritam, craziness based optimizacija rojemčestica, firefly algoritam, krill herd algoritam, moth-flame optimizacija, glow swarm optimizacija, metoda simuliranog kaljenja, bat algoritam, stohastično fraktalno traženje (eng. stochastic fractal search,SFS) i metoda hibridnog SFS lokalnog unimodalnog uzorkovanja. Performanse primijenjenih algoritama upravljanja vrednovani su usporedbom ostvarenih vremena ustaljivanja, iznosa podbačaja i prebačaja te drugih pokazatelja performansi, minimalnih relativnih koeficijentima prigušenja i svojstvenih vrijednosti sustava upravljanja. Provedena analiza osjetljivosti potvruje robusnost parametara optimalnog regulatora za širok raspon radnih točaka i ...

In this paper, we have proposed a quantitative feedback theory based robust digital voltage mode controller for a fourth-order boost dc-dc converter operating in continuous conduction mode. Discrete-time models of the converter are established for triangular trailing-edge modulation, which is then used in the compensator design. It is observed that an up-down glitch is present in the frequency response of control-to-output transfer functions of higher order dc-dc converters, and because of this, gain margin (GM) and phase margin (PM) alone are unable to give correct information about closed-loop stability. Therefore, in this paper, we have introduced a safety margin as an additional design specification, which along with GM and PM information reflects the actual stability of the higher order dc-dc converters. A simplification in loop-shaping approach is also proposed, which considerably reduces the time and effort involved in loop-shaping as compared to present trial-and-error-based approach and also results in a robust compensator of feasible order. Robustness of the designed controllers is analyzed in frequency and time domains through computer simulations and is then verified experimentally on a 28-V 30-W laboratory prototype converter. The experimental results are in line with the analytical design studies and demonstrate that the controller, which is directly designed through a discrete-time model, is more robust than an average model-based digitally redesigned controller.

No abstract

No abstract