Enhancing Distribution System Performance via Distributed Generation Placement and Reconfiguration Based on Improved Symbiotic Organisms Search

Minimal power loss is highly desired for an efcient and economical operation in distributed systems. Tis paper presents an improved symbiotic organisms search (ISOS) for system reconfguration (SR) and distributed generation placement (SR-DGP) simultaneously. Te proposed ISOS combined the simple quadratic interpolation (SQI) strategy into SOS to improve the search process. Te ISOS was adopted to defne the optimal system topology, location, and capacity of distributed generators (DGs) to minimize power losses. Te proposed ISOS was evaluated on the 33-node and 69-node systems. Te proposed ISOS successfully reduced the power losses by 73.1206% and 84.2861% for the 33-bus and 69-bus. Moreover, ISOS was also compared with other approaches, where ISOS obtained better results than other approaches for all test systems. Hence, ISOS showed its efectiveness in dealing with the SR-DGP problem.


Introduction
Tere is a momentous amount of power loss associated with the distribution system. An inefcient power loss results in the system performing inefciently as well as a poor voltage profle. To improve efciency and advance the voltage profles, the system needs to be reconfgured properly. Furthermore, distribution generation (DG) can contribute to lowering power losses and improving the voltage profles and system capacity by integrating local power sources such as wind turbines, solar photovoltaic (PVC), and diesel generators. It is nevertheless imperative that system reconfguration (SR) and DG placement (DGP) should be carried out in a systematic manner since it can result in inefective and undesirable network performance. After reconfguring the distributed system, it is imperative to maintain the radial topology of the feeders. It is also important to determine the DGs placement so as to minimize power losses while maintaining constraints. Terefore, it is very critical to solve the SR-DGP problem properly to improve the system's performance [1].
Numerous studies have examined the SR of the distribution system to minimize power loss. It has been reported that metaheuristic algorithms can be used to reconfgure the distribution system in recent studies such as bacterial foraging optimization (BFO) [2], genetic algorithm (GA) [3][4][5], cuckoo search algorithm (CSA), freworks algorithm (FWA) [6], harmony search algorithm (HSA) [7], heuristic rulesbased fuzzy multiple objectives [8], and particle swarm optimization (PSO) [9,10]. To improve the efectiveness of the distribution system, several studies have combined both SR and DGP in recent years. By deploying a fuzzy-bees technique in [11], power losses were reduced, the voltage profle was enhanced, and feeder load balancing in the distribution system was balanced through reconfguration and distribution of multiple DGs. In [12], the authors employed the gravitational search algorithm (GSA) to defne DG uncertainties and reconfguration. In [13], an adaption shufed frog leaping algorithm (ASFLA) has been adopted for dealing with the coordination of DG allocation and network reconfguration in several cases of 33-node and 69-node systems. To optimize the voltage stability index (VSI) and power losses, an enhanced CSA [14] was implemented based on graph theory. A combination of gray wolf optimizer and PSO was used in [15] to address reconfguration problems while taking DGs into consideration. Te heuristic algorithm used in [16] provided some promising results for network reconfguration involving the allocation of DGs. According to Badran et al. [17], the frefy (FF) could be used to optimize reconfguration and DG outputs. For reconfguration with DGs distribution, Raut and Mishra [18] applied a combination of levy fights and a sine-cosine algorithm (SCA). In [19], DGP and system reconfguration were achieved using HSA in 33-node and 69-node networks, in which voltage profles and power loss reduction optimization were objectives. VSI and loss sensitivity factor (LSF) were employed to calculate the DG positions. Harmony search algorithm (HSA) was introduced by Rao et al. [20] to reconfgure and determine the optimal DGs placement. LSF was also adopted to fnd the DG positions. For optimizing the integration of reconfguration with DG allocation, an improved equilibrium optimization algorithm (IEOA) [21] was proposed. Tis technique was verifed on 33-node and 69-node systems with various load scenarios. For dealing with DG placement and reconfguration, a search group algorithm (SGA) was applied to four distribution networks [22]. A similar problem was studied by the marine predators algorithm (MPA) [23]. A literature review indicates that there have been only a few studies conducted for defning the optimal confguration simultaneously with the DGs placement. By using this approach, maximum loss reduction can be achieved. It may be difcult to solve such a combined optimization problem because it includes several types of variables and constraints and as a result, there is no perfect optimization method to solve this problem.
In this study, we have developed an improved symbiotic organisms search (ISOS) for dealing with the SR-DGP problem to obtain the minimum power losses. To advance the quality of the original SOS, the ISOS was implemented by combining simple quadratic interpolation (SQI). We applied ISOS to determine the SR-DGP in the 33-node and 69-node systems. Based on numerical results, it has been demonstrated that a combination of SR-DGP substantially enhanced the system's power losses and voltage profle. ISOS was also more efcient in terms of obtaining optimal solution quality than other compared methods according to comparison results.
In summary, this paper makes the following contributions: (i) A new ISOS was proposed to handle the SR-DGP problem for minimizing system power losses. (ii) Te ISOS was tested on the 33-node and 69-node systems. After the SR-DGP application, the system performance was improved regarding power losses and voltage profle. (iii) As shown by the outcome evaluations, ISOS gained high solution quality compared to other techniques.

Problem Formulation
Te SR-DGP objective aims to minimize power losses by optimizing the network topology, locations, and capacities of DGs, which may be given in the following equation: where R l is the resistance of the l th branch, I l is the current passing through that branch, and N Br is the number of branches in a distributed system. Te SR-DGP problem needs to satisfy the following constraints: (i) Power balance: where N DG is the number of DG units, N Bus is the number of bus, Q S and P S are the slack bus's reactive and active powers, respectively, Q DG,m and P DG,m are the reactive and active power outputs of the m th DG unit, respectively, Q L,k and P L,k are the reactive and active power losses in the l th branch, respectively, and Q DM,n and P DM,n are the reactive and active power at the n th load bus, respectively. (ii) Bus voltage constraints: the voltage at buses (V k ) may be limited as follows: (iii) Power fow constraints: the power fowing in the l th branches must not exceed the limit I max,l : (iv) DG constraints: the output of the m th DG unit should be limited as follows: (v) DG penetration constraints: the penetration of all DGs follows the given condition [19]: (vi) Radial topology constraint: after reconfguration, all loads must be supplied and the radial structure must be secured [24,25]: 2 Journal of Control Science and Engineering in which Y denotes a matrix resembling the connection of buses and branches in the system:

Original SOS.
Te SOS has been developed based on symbiotic interactions between two creatures in a natural ecosystem [26]. Tis method begins the exploration phase by constructing a population called an ecosystem, where each organism in the ecosystem represents a solution. Te ecosystem is modifed for each iteration via three symbiotic stages: mutualism, commensalism, and parasitism.
In the mutualism stage, the n th organism is chosen at random from the ecosystem to associate with the m th organism via mutualistic connections. New organisms are created as follows [26]: where O best is the best organism in the ecosystem, vectors O m and O n are the m th and n th organisms in the ecosystem, respectively, MV denotes the average of the m th and n th organisms, representing a mutualistic relationship; bf 1 and bf 2 (beneft factors) are chosen at random as either 1 or 2.
Te ftness values are estimated for organisms. New organisms are updated in the following equations: In the commensalism stage, the m th organism is likewise chosen randomly from the population to connect with the n th organism, equivalent to the mutualism step. Trough commensalism interaction, the m th organism gets advantages, whereas the n th organism is neither damaged nor profted. A new organism can be defned as follows [26]: Te new organism can be updated according to equation (10).
In the parasitism communication of two diverse organisms in the parasitism stage, one organism (i.e., parasite) gets benefts while the other (i.e., host) is negatively afected. During this period, the m th organism takes over the parasite's role. By replicating the vector O m , a Parasite_Vector (PV) is generated. Several PV vector elements are randomly altered to develop a new candidate solution (O PV ) [26]. Te n th organism is picked at random from the existing population to act as a host. Both PV and O n vectors have their ftness values calculated. Te O PV vector is updated or discarded as follows:

SQI Strategy.
Te efciency of an optimization method is based on its exploitation and exploration. Having global exploration capability indicates that the optimization algorithm is efectively utilizing the entire search space. Local exploitation ability refers to the ability of the optimization algorithm to search for the best solution near a new solution that has already been discovered. However, a metaheuristic algorithm may become stuck in a local optimum and fail to converge because of its stochastic nature. To enhance the search ability of SOS, the SQI method is integrated with SOS. By using SQI, a new set of solution vectors is generated that lie on a point of minima quadratic curve that passes through three randomly selected solution vectors. Hence, SQI is able to accelerate the algorithm's convergence [27]. After the parasitism phase at each iteration, the d th dimension of a new candidate organism can be defned based on a threepoint SQI as follows [28]: where O n and O k are the two organisms randomly selected from the population, f m , f n , and f k denote ftness function values for m th , n th , and k th organisms, respectively, and D is the dimension of the problem, and d � 1, 2, . . ., D.
Te ftness values are computed for the organism O new i . Te updating of the new organism can be defned as follows: O i � SW 1 , · · · , SW N SW , L DG,1 , · · · , L DG,N DG , P DG,1 , · · · , P DG,N DG , (16) where N SW is the number of opened switches. Each organism is randomly generated in its boundaries, in which the opened switch's positions and DG unit's locations are natural numbers. Tus, the control variables for opened switches positions (SW i ), locations (L DG,i ) and sizes (P DG,i ) of DG units are constructed using the given formula:

Application of ISOS to SR-DGP
where N SW indicates the total number of opened switches.
Every ftness function value for each organism of ISOS is calculated as follows: in which K P represents penalty constants for inequality constraint violations. If the dependent variables (bus voltages, feeder capacity, and DGs penetration) violate the constraints, a method is applied to adjust the variables towards their bound: in which x symbolizes the V i , I k , and PE DG values; x lim denotes the limitations of V i , I k , and PE DG . Te steps for applying the ISOS approach to the SR-DGP problem are given as follows: Step 1: the data of the SR-DGP problem is declared Step 2: ISOS parameters (NP and maxIter) are defned Step 3: the population O is initialized Step 4: the ftness values are calculated using equation (18) for each organism in the population O. Set the iteration to 1 (Iter � 1) Step 5: the best organism is defned Step 6: identify the best solution Step 7: perform the mutualism phase Step 8: perform the commensalism phase Step 9: perform the parasitism phase Step 10: perform the SQI method Step 11: if Iter < maxIter, increase the iteration (Iter � Iter + 1) and proceed to Step 5. Otherwise, the algorithm is stopped Finally, the ISOS application to the SR-DGP problem is given in Figure 1.

Simulation Results
Te ISOS algorithm was carried out on Matlab 2019b. For each test case, the ISOS method was executed in thirty trials independently. Te ISOS control parameters were set as follows: NP � 50 and maxIter � 200. Te ISOS performance was compared with other techniques in previous papers.

33-Node
System. Initially, the ISOS technique was applied to the 33-node system with 37 branches and 33 nodes, as shown in Figure 2. Te line and load data for the 33-node system can be found in Supplementary Table S1. For the original case, the system contains 202.677 MW of power loss [29]. Tree DGs were located in the system with a maximum capacity of 300 kW. Table 1 shows the results acquired from the ISOS method for this system. From Table 1, the network confguration was obtained with the opened switches: 7-14-9-30-28. Concurrently, the integrations of DG units were defned at buses 33, 12, and 25 and their capacities were 739 kW, 469 kW, and 1020 kW, respectively. Te ISOS acquired minimal power losses of 54.4785 kW; i.e., a 73.1206% reduction compared with the original case. Furthermore, the voltages at load buses were also enhanced, which can be seen in Figure 3. Te minimum voltage was improved from 0.9131 pu to 0.9678 pu. Tis showed that the SR-DGP using ISOS signifcantly impacted the voltage improvement and power loss decrease. Table 1 also gives the outcomes obtained by ISOS and other techniques in the literature for this system.   Journal of Control Science and Engineering Compared with other methods, the proposed ISOS acquired the lowest power losses. It may be shown that the power losses of 54.4785 KW produced by the ISOS was better than SOS, FF [17], ISCA [18], FWA [19], HSA [20], GA [20], RGA [20], EOA [21], and IEOA [21]. Tus, the ISOS approach can provide a good-quality solution for this system. Te ISOS and SOS convergence curves are shown in Figure 4. ISOS converges to a better value than SOS, which indicated its good performance regarding convergence ability.

69-Bus
System. ISOS method was employed in the 69node system with 73 branches and 69 nodes, as shown in Figure 5. Te line and load data for the 69-node system can be found in Supplementary Table S2. For the original case, minimum voltage amplitude and power losses were 0.9092 pu and 225.03 kW [29], respectively. Te results attained by ISOS for the 69-bus system are given in Table 2. Te opened switches: 69-70-14-55-61 were defned by the ISOS to generate the optimum topology of the 69-bus system. Moreover, DG units with capacities of 475 kW, 28 29 30  31  32  33  34   28  29  30  31  32  33  34   35   47  48  49   47  48  49   50   72  53  54  55  56  57  58   53  54  55  56  57  58   59   59   60  61  62  63  64   60  61  62  63   Journal of Control Science and Engineering 7 respectively. Compared to the base case, the minimal power losses was 35.3549 kW, which is corresponding to a reduction of 84.2861%. Te voltages at load buses were improved as shown in Figure 6. Te minimum voltage in the system was enhanced from 0.9092 pu to 0.9806 pu. Accordingly, SR-DGP using ISOS signifcantly enhanced the voltage profle and reduced the power loss of this system. Table 2 also indicated a comparison between ISOS and other techniques for the 69-node system. As seen in Table 2, the minimal power loss of 35.3549 kW attained by ISOS is better than those attained by SOS, FF [17], ISCA [18], FWA [19], HSA [20], GA [20], RGA [20], EOA [21], and IEOA [21]. Te convergence characteristics of ISOS and SOS are depicted in Figure 7. Tis fgure showed that ISOS converged to better results than SOS, as shown in Figure 7. Terefore, ISOS is very efective for dealing with SR-DGP in the 69-bus system.

Conclusion
Tis study has successfully implemented the ISOS method to solve the SR-DGP problem for minimizing power loss. Te ISOS efectiveness has been validated on the 33-node and 69node systems. It was found that SR-DGP using ISOS led to a reduction of power loss and an improvement of voltage profle compared to the original case. Te power loss reductions for 33-node and 69-node were 73.1206% and 84.2861%, respectively. Also, the fndings indicated that ISOS delivered higher solution quality with regard to power loss reduction than other techniques, as seen from the outcome evaluations. ISOS also showed better convergence characteristics than SOS for both test systems. Tus, ISOS provided a viable solution for DG placement issues and managing network reconfguration in distribution systems.

Data Availability
Te system data used to support the fndings of this study are included within the supplementary information fle.

Conflicts of Interest
Te author declares that there are no conficts of interest.    Journal of Control Science and Engineering

Supplementary Materials
Te line and load data for 33-node and 69-bus systems can be found in Supplementary Tables S1 and S2 in Supplementary document. (Supplementary Materials)