International Journal of Control, Automation, and Systems (2012) 10(4):1-6 DOI ISSN:1598-6446 eISSN:2005-4092 http://www.springer.com/12555 A Multi-objective Fuzzy Adaptive PSO Algorithm for Location of Automatic Voltage Regulators in Radial Distribution Networks Taher Niknam, Mohammad Rasoul Narimani, and Rasoul Azizipanah-Abarghooee Abstract: This paper proposes a multi-objective optimal location of Automatic Voltage Regulators (AVRs) in distribution systems at the presence of Distributed Generators (DGs) by a Fuzzy Adaptive Particle Swarm Optimization (FAPSO) algorithm. The proposed algorithm utilizes an external reposi- tory to save founded Pareto optimal solutions during the search process. The proposed technique al- lows the decision maker to select one of the Pareto optimal solutions (by trade-off) for different appli- cations. The performance of the suggested algorithm on a 70-bus distribution network in comparison with other evolutionary methods such as Genetic algorithm and PSO is extraordinary. Keywords: Automatic voltage regulator, fuzzy adaptive particle swarm optimization, multi-objective, pareto method. 1. INTRODUCTION In distribution systems, voltage control is a difficult task since voltages are eventually influenced by dynamic load fluctuations. In this regard, utilities reinforce their power systems in order to have a better control over voltage variations [1]. Utilization of analytical tools such as Optimal Power Flow (OPF) and voltage stability analysis can minimize bus voltage deviations. Meanwhile, voltage profile can be improved by usage of some devices such as transformers with on-load tap changers (OLTCs), fixed and controlled capacitors banks, DGs and AVRs [2]. It is worthwhile to note that in distribution networks the application of some equipment such as AVRs depends on the investment cost of these devices and that is why the optimal location of AVRs becomes an important issue. Also, the intense attention to DGs impact in power systems, especially on the distribution networks is crucial [3]. During the years many single-objective optimization methods such as Simulated Annealing (SA) and artificial neural networks [4] have been applied to solve the problem of optimal placement and sizing of capacitor banks. Furthermore the aforementioned techniques; nowadays, fuzzy logic [1] and evolutionary algorithms [5,6] are also applied to solve the problem of optimal placement of capacitor banks in distribution networks. In order to optimize tap position and the ON/OFF state of the capacitor banks, analytical tools such as optimal power flow have been applied [7]. In [6] and [9] the optimal location of OLTCs and capacitor banks have been illustrated by an evolutionary multi-objective approach. In [10-12], the optimal location of AVRs is surveyed separately from the placement and sizing of the capacitor banks problem. In the same way, in [13], authors determined the location of AVRs by using a sequential algorithm. In [14], for choosing the optimal location of AVRs in radial distribution systems a simple Genetic Algorithm (GA) has been applied. In this approach the multiobjective version of the AVR placement has been studied. The considered objective functions of the proposed approach are not the same and it is difficult to solve this type of problem by usage of the classical approaches that are applied to optimize single objective problems. Therefore, the proposed algorithm first models the multiple-objectives using fuzzy sets to release their imprecise nature. In this paper the multi-objective solution method in the suggested FAPSO algorithm has been applied in order to extract Pareto optimal solutions of the problem which works according to dominate and non-dominate concept. This method can find all non-dominated solutions and allows the decision maker to select one of them (by trade-off) for different applications. In other words, the decision-maker can choose any of the Pareto optimal solutions based on his/her own preferences. For more validating the obtained results of the proposed approach are compared with those which are obtained genetic algorithm (GA) and Particle Swarm Optimization (PSO). 2. MULTIOBJECTIVE OF LOCATION OF AUTOMOTIC VOLTAGE REGULATORS Objective functions and constraints: The objective functions are the total electricity cost, electrical losses and voltage deviation and described in (1) to (3) as: 1 1 min ( ) ( ) , DG N sub sub DGi DGi i f X P price P price = = ´ + ´ å (1) © ICROS, KIEE and Springer 2012 __________ Manuscript received July 6, 2011; revised March 13, 2012; accepted May 24, 2012. Recommended by Editorial Board mem- ber Euntai Kim under the direction of Editor Young-Hoon Joo. Taher Niknam, Mohammad Rasoul Narimani, and Rasoul Azi- zipanah-Abarghooee are with the Department of Electrical and Electronics Engineering, Shiraz University of Technology, Shiraz, Iran (e-mails: niknam@sutech.ac.ir, m.narimani@sutech.ac.ir, r.azizipanah@sutech.ac.ir).