Microsoft Word - 11-3166_s_ETASR_V9_N6_pp4925-4932 Engineering, Technology & Applied Science Research Vol. 9, No. 6, 2019, 4925-4932 4925 www.etasr.com Duong & Nguyen: Network Reconfiguration for an Electric Distribution System with Distributed … Network Reconfiguration for an Electric Distribution System with Distributed Generators based on Symbiotic Organisms Search Thanh Long Duong Faculty of Electrical Engineering Technology Industrial University of Ho Chi Minh City Ho Chi Minh City, Vietnam duongthanhlong@iuh.edu.vn Thuan Thanh Nguyen Faculty of Electrical Engineering Technology Industrial University of Ho Chi Minh City Ho Chi Minh City, Viet Nam nguyenthanhthuan@iuh.edu.vn Abstract—This paper proposes a method of network reconfiguration based on symbiotic organisms search (SOS) algorithm for reducing power loss of the electric distribution system. The SOS is a recent developed meta-heuristic algorithm inspired from the symbiotic interaction strategies of organisms for surviving and propagating in the ecosystem. Compared to other algorithms, SOS does not need any control parameters during the searching process. The advantages of the proposed SOS method have been validated in two electric distribution systems. Three network cases have been considered for each system, consisting of performing network reconfiguration on the system without distributed generator (DG) placement, the system installed type-P DGs and the system installed type-PQ DGs. The comparison results with particle swarm optimization and other previous methods show that the proposed SOS can be a promising technique for the problem of electric network reconfiguration. Keywords-symbiotic organisms search; electric distribution system; network reconfiguration; active power loss I. INTRODUCTION Power loss in electric distribution systems (EDSs) takes a high portion in the total power loss of the electricity system. There are lots of techniques for reducing power loss such as distributed generation placement [1], capacitor placement [2-4], and network reconfiguration. Among these methods, network reconfiguration is the most efficient. It is implemented by transferring a part of the load from some branches to others through changing the status of switches located on each branch. The result of network reconfiguration method is a radial topology of EDS that satisfy constraints such as the limits of node voltage and branch current. From its introduction [5], the problem of network reconfiguration has been solved by many methods. They can be divided into two main groups: heuristic and metaheuristic. The first main group includes methods based on knowledge of operating the EDN and technical strategies. Typical methods in this group are grouped in branch and bound methods and branch exchange methods. The idea of branch and bound method is that the EDN is modeled by a purely resistive network. Initially all switches are closed. Gradually the switches on the branches with the smallest currents are opened step by step until the radial topology of the EDN is reached [6]. In the branch exchange method, initially all switches are closed and then an open switch is replaced by a closed switch for reducing power loss [7]. The advantages of this method group are that they are easy to implement and the knowledge related to these methods is very close to the area of operating EDS. However, the obtained results are usually local extremes. In addition, changing the objective function is difficult for solving the problem of network reconfiguration. The second group method includes methods based on metaheuristic algorithms. They are usually inspired from phenomena of nature or society. The main advantage of this method group is that they are very efficient for handling constraints and different objective functions. Compared to the former group, the later has the ability to provide a sufficiently good solution to the problem of network reconfiguration. Thus, more and more new metaheuristic algorithms have being applied to this problem such as genetic algorithm [8], particle swarm optimization (PSO) algorithm [9-11], gravitational search algorithm (GSA) [12], fireworks algorithm (FWA) [13], runner root algorithm (RRA) [14, 15], and cuckoo search algorithm (CSA) [16, 17]. However, for solving the problem of network reconfiguration by this method group, the issue of adjusting the control parameters of the algorithms has to be faced, e.g. for applying the GA to the problem of network reconfiguration, the rates of selection and mutation have to be tuned, for applying the PSO, two constants C1 and C2 in the equation of velocity of particles have to be selected etc. Therefore, it is worth considering the application of recent developed metaheuristic algorithms which have less control parameters, for solving the network reconfiguration problem. Symbiotic organisms search (SOS) is inspired from the symbiotic interaction strategies of organisms for surviving and propagating in the ecosystem [18]. Compared to other metaheuristic algorithms, the biggest advantages of SOS are that it does not require specific algorithmic parameters for operations. In SOS, from each organism in the ecosystem, four new organisms are generated based on mutualism, commensalism, and parasitism phases. Two new organisms are used for updating the initial organism and the other two Corresponding author: Thuan Thanh Nguyen Engineering, Technology & Applied Science Research Vol. 9, No. 6, 2019, 4925-4932 4926 www.etasr.com Duong & Nguyen: Network Reconfiguration for an Electric Distribution System with Distributed … organisms are used for updating the other organisms in the ecosystem. In [18], SOS has been used to solve 26 benchmark functions and has outperformed other algorithms such as GA, PSO, etc. Recently, SOS has been applied in many areas such as construction engineering, power or energy optimization engineering problems [19]. In the latter, SOS has been successful applied in the problems of optimal placement and sizing of distributed generators (DGs) [20, 21], reduction of power loss of a transmission system [22], capacitor placement [23], economic emission dispatch problem [24], and optimal power flow [25]. For the problem of network reconfiguration, SOS has been used in [26]. However, in this study, the solution vector of SOS (called SOS1) for the problem is expressed by the state of all the electric switches in the EDS. The normally closed electric switches are shown by the one state and vice versa, the open electric switches are represented by the zero state. Coding technique makes the solution vector become large and makes SOS ineffective for finding optimal network configuration. In this paper, the proposed method based on SOS is used to solve the problem of network reconfiguration to find the optimum EDS configuration for power loss reduction. The solution vector of the problem is only presented by the open switches in integer form. The applied method was tested on the 33-bus and 69-bus systems and the obtained results were compared with the ones of other techniques. The main contributions of this paper can be summarized as: • SOS is adapted to solve efficiently the network reconfiguration problem. • The performance of SOS is effectively applied on different electric distribution systems, namely the EDS without DGs and with different types of DGs. • SOS can find the optimal radial topology of EDS for all cases, with higher success rate than the other methods mentioned in the literature. II. THE PROBLEM OF NETWORK RECONFIGURATION The power loss of EDS is determined by the sum of the power losses of each branch of the system. Although having mesh structure, EDSs always operate in radial topology. So, there are branches that contain switches in the open state with no current flow. Therefore, the total power loss of the EDS is determined by: ∆� = � ���� �� �� � ��� ��� (1) where ��� is a number of branches of the EDS, �� is the i-th branch’s resistance, Pi and Qi are respectively the active and reactive power flow on the i-th branch. �� is a coefficient which is equal to zero for an open branch and one for a close branch, and �� is the ending voltage of the i-th branch. Equation (1) is subjected to the following constraints: • Radial network constraint: The radial topology of EDS must be maintained and no load nodes have not to be served. In this study, the empirical formula from [27] is used to check whether the candidate configuration is radial or not: det(�) = �−1 !� 1, �#$%#& 0, (!) �#$%#& (2) where det(A) is the determinant of matrix A. A is a (branch x node) matrix presenting a connection of EDS. • Node voltage constraint: The voltage magnitude at each node must lie in the permissible range [�*�+,�*,-] to keep power quality. In general, �*�+,�*,- are usually set equal to [0.95, 1] pu: �*�+ ≤ �/ ≤ �*,- 0%)ℎ 2 = 1,2, . .��56 (3) where ��56 is the total number of nodes in EDS. • Branch current constraint: The current magnitude of each branch of EDS must lie in their permissible range to avoid over load: 7� ≤ 7*,-,� 0%)ℎ % = 1,2, . .��� (4) III. SOS FOR NETWORK RECONFIGURATION SOS is based on the interactive behavior of organisms in nature. Many species living in nature have a reliant relationship with other species, called symbiosis. Commonly, there are three popular reliant relationships in nature, mutualism, commensalism, and parasitism [18-19]. The main procedures of SOS for the problem of network reconfiguration problem are described below. A. Initializating Candidate Solutions Similarly to other metaheuristic algorithms based on population, SOS also generates an initial population of organisms termed as the ecosystem. Each organism represents a candidate solution of the optimization problem. The degree of adaptation of each organism in the ecosystem is reflected by its fitness value. The process of creating the initial population of organisms is implemented as: 8� = 89:;,< +�#($ ×?8@�A@,< −89:;,/B (5) where Xi (i=1,2… N) is an organism in the ecosystem. 8@�A@,< and 89:;,< are the upper and lower limits of the control variable k (k=1, 2, …, D) of the organism, N and D are the population size and the dimension of the problem. The control variables of the network reconfiguration problem are the switches located in each branch of the EDS. Therefore, the control variables used in SOS for the network reconfiguration problem are presented under integer form. In addition, the search space of each control variable is also determined by closing the corresponding initial open switch Xinitial,k. By closing Xinitial,k, a loop occurs in the EDS called Loopk. All the switches in this loop are candidate open switches for the control variable Xi,k. It is clear that the size of Loopk is the upper limit of the control variable Xi,k and the lower limit of the control variable Xi,k is one, which is the first position in Loopk. In other words the control variables of the network reconfiguration problem are the position of switches in each loop created by closing one by one the initial open switches in the EDS. Therefore, in the initialization state, the ecosystem is generated as: 8� = roundG89:;,< +�#($×?8@�A@,< −89:;,/BH (6) Engineering, Technology & Applied Science Research Vol. 9, No. 6, 2019, 4925-4932 4927 www.etasr.com Duong & Nguyen: Network Reconfiguration for an Electric Distribution System with Distributed … where round is a function that rounds each element of Xi to the nearest integer. From the initial population of organisms, the radial topology of each organism is checked and if it is satisfied, the power flow is run to calculate its fitness function. Then, based on the fitness value, the best organism Xbest of the ecosystem is found. If the radial topology of organism Xi is not satisfied, a very high value is assigned to its fitness function value. B. Generating New Solutions Based on Mutualism Strategy In nature, mutualism helps in gaining benefits for both participating organisms in the ecosystem. An example of this strategy can be seen from the relationship between flowers and bees. From this metaphor, the organism Xi in the population will interact with the organism Xj which is selected randomly from the population to create two new solutions as in (7). Participating in the mutualism strategy, both organisms gain benefits. The radial topology of the two new organisms is checked and if it is satisfied, the power flow is run to calculate their fitness function. Then, if their fitness values are better than the one before interaction, the organisms Xi and Xj will be updated: I8�,+J; = roundK8� +rand(0,1)× (8�JMN −OPJQN:� ×RS1)T8/,+J; = roundG8/ +rand(0,1)×(8�JMN −OPJQN:� ×RS2)H (7) OPJQN:� is determined by (8). BF1 and BF2 are the benefit factors representing for level of benefit of each organism determined by: OPJQN:� = U �UVW (8) �RS1 = round (1 +rand�)RS2 = round (1 + randW) (9) C. Generating New Solutions with Commensalism Strategy In nature, many birds live on bigger herbivores. This is a typical example of commensalism. In this strategy, a species will benefit while the other derives neither benefit nor harm. The organism Xi will benefit from the commensal interaction between organism Xi and Xj. To implement this strategy, organism Xj is selected randomly from the ecosystem to interact with Xj as presented in (10) for creating a new organism. Then, organism Xi will be updated only if the fitness of the new organism is better than that of Xi before the interaction. 8�,+J; = roundG8� +rand(−1,1)×(8�JMN − 8/)H (10) D. Generating New Solutions with Parasitism Strategy Contrary to commensalism strategy, in the parasitism one species will benefit while the other species as a host will be a victim and even may die. In SOS, organism Xi is considered as a parasite. It modifies randomly some control variables to create a new organism called 8 ,�,M�NJ. Therefore, 8 ,�,M�NJ is first assigned by Xi, and then some of the control variables of 8 ,�,M�NJ are modified by (11). Note that the number of control variables of 8 ,�,M�NJ is also selected randomly. The organism Xj which is considered a host is selected randomly. If the fitness function value of the new organism created from Xi is better than that of the organism Xj, it will replace the position of Xj in the ecosystem. In other words, Xj will be killed by Xi. 8 ,�,M�NJ,< = roundG89:;,< + �#($ × ?8@�A@,< −89:;,