04-05-2011, 12:37 PM
Abstract
We propose a new method of controller design fora boost converter in this paper. An optimization model for thecontroller design is developed, and solution is sought through anovel optimization algorithm which combines standard GeneticAlgorithm (GA) with evolution of a queen bee in a hive. Computedand measured results suggest that the new algorithm converges tothe best designs on a limited number of iterations.Index Terms—Boost converter, optimization, queen bee.
I. INTRODUCTION
BOOST converters provide an output voltage higher thanthe input voltage and are increasingly employed as frontendconverters for battery sources, photovoltaic solar systems,and fuel cells [1], [2]. When boost converter is employedin open loop mode, it exhibits poor voltage regulation andunsatisfactory dynamic response, and hence, this converter isgenerally provided with closed loop control for output voltageregulation. The mode of operation of the converter varies fromON to OFF state of the power switch and traditionally smallsignal linearization techniques have largely been employed forcontroller design. However, boost converter is a nonminimumphase system due to the presence of a right-half-plane zerowhich causes sluggish dynamic response [3]. Hence, the traditionallydesigned controllers do not guarantee fast dynamicresponse over wide bandwidth of supply and load variations.Several methods of controller design based on a small-signallinearized model and implementation for closed-loop control ofboost converters are seen in the literature [4]–[9]. These includedesign based on linear control theory such as Ziegler–Nichol’smethod [5], root locus technique [6], circle-based criterion [7],hysteresis method [8], robust control [9], etc. The concept offuzzy logic is also recently applied for the controller design[10]. Recent works in the design of boost converter control withthe use of state-energy plane [11], slidingmode controllers [12],and hybrid control algorithm [13] have shown the developmentof a robust controller working excellently well over a widerange of operating conditions.This paper proposes an optimization model for the feedbackcontroller design, and the solution is achieved through a noveloptimization algorithm. The optimization algorithm combinesthe evolution of a queen bee in a hive [14] with a standardgenetic algorithm (GA) [15]. The survival of a queen bee in a hive is perceived as “survival of fittest” and is used forperformance enhancement of a standard GA. The modifiedalgorithm—labeled as Bees GA in this paper—is then tailoredto suit feedback controller design of a boost converter. Thelarge-signal model of a boost converter is employed to evaluatethe objective function to achieve a robust controller structure.The new methodology is verified through computed andmeasured results. Furthermore, the new proposition is shownto be superior to a standard GA.
II. FORMULATION OF THE PROBLEM
A. Modeling of Boost ConverterA closed-loop boost converter using a MOSFET as a switchingelement is shown in the Fig. 1. The specifications of theconverter considered in this paper are the following: inputvoltage, Vin = 36 V; switching frequency, fs = 2 kHz; inductanceL = 33 mH; equivalent resistance of the inductor, rL =3 Ω; capacitance, C = 1000 μF; equivalent resistance of thecapacitor, rc = 0.5 Ω, and load resistance, RL = 100 Ω.In the feedback control system, the actual output voltage v0,and its reference value V ∗0 are first compared using comparatorand the error, e(t) so obtained is processed by the PIDcontroller. The output voltage of the PID controller u(t) isan analog signal and must be converted into gating pulse toMOSFET with adjustable duty cycle. This task is performedby the modulator which compares the PID controller outputvoltage with a ramp signal so that the output of modulator isgating pulse with its duty cycle varying in accordance with PIDcontroller output voltage.The differential equation describing the converter behaviorduring the ON state of the switch is
Download full report
http://ieeexplore.ieeeiel5/41/4796879/04...er=4663697
