بهینه سازی کلونی مورچه ها اقتباس شده برای پیکر بندی دوباره کارآمد از سیستم های توزیع متعادل و نامتعادل برای به حداقل رساندن از دست دادن

This paper presents an efficient method for the reconfiguration of radial distribution systems for minimization of real power loss using adapted ant colony optimization. The conventional ant colony optimization is adapted by the graph theory to always create feasible radial topologies during the whole evolutionary process. This avoids tedious mesh check and hence reduces the computational burden. The initial population is created randomly and a heuristic spark is introduced to enhance the pace of the search process. The effectiveness of the proposed method is demonstrated on balanced and unbalanced test distribution systems. The simulation results show that the proposed method is efficient and promising for reconfiguration problem of radial distribution systems.

Distribution networks are generally structured in a mesh but operated in the radial configuration for effective co-ordination of their protective schemes and to reduce the fault level. The reconfiguration of a distribution system is a process that alters the feeder topological structure by managing the open/close status of sectionalizing and tie-switches in the system under contingencies or under normal operating conditions. Reconfiguration of the radial distribution system is a very effective and efficient means to reduce distribution network losses, improve voltage profile, manage load congestion and enhance system reliability. The aim of distribution network reconfiguration is to find a radial operating configuration that optimizes certain objectives while satisfying all the operational constraints without islanding of any node(s). A lot of research work has been carried out to solve distribution network reconfiguration problems. These research efforts can be broadly classified into traditional approaches and artificial intelligence (AI) based approaches. The traditional approaches include heuristic optimization techniques and classical optimization techniques. Merlin and Back [1] were first to report a method for distribution network reconfiguration to minimize feeder loss. They formulated the problem as a mixed integer nonlinear optimization problem and solved it through a discrete branch-and-bound technique. Later, [2], [3], [4], [5], [6] and [7] also suggested different branch exchange heuristic algorithms. The complexity of reconfiguration problem increases with the exponential growth in the size of modern distribution networks and the heuristic techniques fail to provide a quality solution. Therefore, the researchers diverted toward various stochastic-based search techniques. Nara et al. [8] introduced genetic algorithm (GA) for reconfiguration of distribution networks for loss minimization. Later, several GA based methods [9], [10], [11], [12], [13] and [14] have been used for the reconfiguration of distribution networks. Mendoza et al. [13] proposed a new methodology for minimal loss reconfiguration using GA with the help of fundamental loops. They restricted the search space of GA by modifying the genetic operators. Enacheanu et al. [14] presented a method based on GA for loss minimization in the distribution networks using matroid theory and graph theory. Some other population-based meta-heuristic techniques, e.g., immune algorithm [15], evolutionary algorithm [16], simulated annealing [17] and [18], tabu search [19], [20] and [21], particle swarm optimization [22] also attempted the reconfiguration problem. The ant colony optimization (ACO) is a population-based meta-heuristic technique, has emerged as a powerful tool for solving combinatorial optimization problems, initially proposed by Marco Dorigo in 1992 in his Ph.D. thesis [23]. The search technique is inspired by the behavior of ants in finding paths from the nest to food and back to the nest. It was implemented to solve the traveling salesman problem (TSP) by Dorigo and Gambardella [24]. Later, Stutzle and Hoos [25] developed the max–min ant system (MMAS) to solve TSP and quadratic assignment problem. Then the basic ACO was further improved and a model-based search (MBS) algorithm was introduced by Blum and Dorigo [26]. Das et al. [27] attempted ant colony approach to compute minimum Steiner tree. The original idea of ACO has been diversified to solve a wider class of problems. Recently, the ant algorithm also has been applied to various optimization problems of the power systems, such as short-term generation scheduling problem [28], unit commitment [29], hydroelectric generation scheduling [30], distribution system planning [31] and [32], joint optimization for capacitor placement and reconfiguration of the distribution systems [33]. Su et al. [34] proposed state transition rule, local and global updating rules to make the ACO computationally efficient to minimize real power loss in the distribution networks. Ahuja et al. [35] introduced the inherent feature of hyper-mutation of Artificial Immune System (AIS) into the ACO algorithm for multi-objective optimization reconfiguration problem to avoid local minima. Carpaneto and Chicco [36] employed restricted branch exchange to improve the computation efficiency of the conventional ACO. The reconfiguration of the distribution system for loss minimization is a complex, combinatorial optimization problem. The application of these population-based search techniques to solve the reconfiguration problem of the distribution networks faces an additional difficulty of maintaining the radiality constraint throughout the evolutionary process. The methods available in the literature provide different ways of maintaining radiality constraint, but, they are incomplete in the sense that they may generate infeasible individuals during initialization as well as during the evolutionary process. These infeasible individuals are either rejected or corrected using some mechanism and the process is repeated till feasible individuals are obtained, which may be time consuming. The main contribution of this paper is to propose a new codification to generate only feasible radial topologies of distribution system while solving the reconfiguration problem using the ACO. For this purpose, some rules are framed with the help of the graph theory and the conventional ACO is adapted by these rules, hence named Adaptive Ant Colony Optimization (AACO). The other contributions of this work are: initial feasible population is created randomly to maintain the diversity, a heuristic spark (HS) is introduced to make AACO computationally efficient, and the desirability is defined on the basis of node voltages to guide the ant search. In this paper, the reconfiguration problem of the balanced and unbalanced distribution networks to minimize real power loss is solved using AACO. The organization of the paper is as follows. The formulation of the loss minimization problem is discussed in Section 2. The conventional ACO is explained in Section 3. The modifications proposed in the conventional ACO are discussed in Section 4, in Section 5 the proposed codification for AACO is illustrated with the help of an example. In Section 6, the application results of the proposed method on balanced and unbalanced distribution systems are presented and finally concluded in Section 7.

The reconfiguration of distribution networks is assuming significant importance in the context of modern distribution systems. In this paper, a new codification for the population-based meta-heuristic techniques to solve reconfiguration problem of distribution networks is presented. In the proposed method, the topological concepts of the loop vectors, common branch vectors and prohibited group vectors have been introduced with the help of graph theory and some rules are framed to avoid the generation of infeasible individuals during each stage of the proposed AACO. The importance of each of these rules while randomly creating the initial population of 100 individuals can be observed from Table 7. It can be depicted from the table that the number of infeasible individuals as well as the computation time for the initialization is reduced when these rules are incorporated subsequently. Therefore, these rules not only provide flexibility to initialize the search process by random selection but also generate only feasible individuals. This preserves the diversity for the search process. It can be concluded from the table that the importance of these rules increases with the increase in the size of the distribution systems.These rules also correct the infeasible individuals, whenever appear during the evolution process. The average computation time required to correct infeasible individuals is shown in Table 8. In fact, the proposed rules guarantee to generate only feasible radial topologies throughout the evolutionary process, irrespective of the size of the distribution system.The proposed method incorporates the advantages of heuristics to increase the pace of the search technique. The introduction of the HS in the initial population reduces the computation time significantly. Various parameters of the conventional ACO are redefined in the context of the reconfiguration of the distribution networks. The objective function for the loss minimization is optimized using the proposed AACO for balanced and unbalanced distribution systems. The simulation results show that the method provides a promising tool for the reconfiguration problem of distribution network and can be extended to incorporate multi-objective problem without any significant computational burden. The proposed codification can be investigated in other advanced versions of ACOs and other population-based search techniques as a future work.