LinkExchange Member
The maintenance of the high voltage electricity transmission network in England and Wales (the National Grid) is planned so as to minimise costs taking into account: location and size of demand for electricity, generator capacities and availabilities, electricity carrying capacity of the remainder of the network, i.e. that part not undergoing maintenance. This complex optimization and scheduling problem is currently performed manually (computerised viability checks can be performed after the schedule has been produced). The paper [PS145] reports work aiming to automatically generate low cost schedules using GAs. So far: a small demonstration problem has been identified, a fitness function has been devised, to date work has concentrated upon devising a representation based upon 'greedy optimizers', which combine permutation GAs with scheduling heuristics, the best of these heuristics has been incorporated in the QGAME GA programming environment and optimal solutions have been readily found.
The paper [PS146] presents an approach to solve power system generation expansion planning (GEP) problems. Since a GEP problem has highly nonlinear characteristics, together with discrete nature, it is difficult to obtain a global optimum with conventional mathematics-based algorithms. Therefore, the authors suggest a new GA, the NGMS (nongeneration operation and main-sub population) GA. The NGMS GA adopts two new methods to overcome the premature convergence characteristic. The first is the nongeneration operation which is a kind of eliticism. The second is a main-sub population model which can search new feasible space and prevent duplication in a population. To show its effectiveness and feasibility, the results of the suggested algorithm are compared with those of conventional dynamic programming and conventional GAs. Although it is not guaranteed to obtain the true global solution, the possibility of finding a global optimum as well as sub-optima is more likely than in a conventional GA.
The optimal design of electric power distribution systems has been usually stated in scientific papers as a classical mixed-integer mathematical programming problem, where an objective function, representing the distribution system expansion costs, is minimized subject to technical constraints related to the distribution network. Then, the well known branch-and-bound algorithms have been habitually applied to solve the optimization problem, which is based on implicit enumeration and evaluation of feasible solutions, by using suitable criteria to search such solutions. However, in many cases these algorithms require large amounts of CPU time to achieve the optimal distribution network solution. The paper [PS147] presents the application of the GAs to solve the optimal design of distribution systems, reducing drastically the computer time. The computer results have been excellent, since the CPU computer hours used by the branch-and-bound algorithms have been reduced to several minutes for the GAs. Furthermore, significant savings in CPU time are obtained by GAs, with respect to the classical branch-and-bound algorithms, when the size of the designed distribution network grows. Therefore, GAs seem to be especially useful tools for application to the optimal design of large distribution systems, as the usual systems are in practice.
The paper [PS148] presents a new problem solving environment that can utilize different forms of resources to approach better solutions in the distribution network planning domain. To implement this concept, a design optimization framework, named design by expectation (DBE), is developed. DBE exploits the adaptation of GAs in searching the optimal solution, and releases the strict requirements for objective functions and constraints in conventional optimization techniques. One example, street lighting design, is used to demonstrate the applicability of the proposed approach.
Restorative power system planning problems can be treated as a combinatorial problem. In order to solve these problems practically, heuristic techniques, instead of mathematical programming, are usually used. The algorithm, however, might be specific to the given problem and there is no guarantee that an optimal solution will be obtained. In the paper [PS149], generalized methods for modelling and an algorithm for restorative planning problems are discussed. GAs and neuro-computing are applied to solve the problems. It is found that, using GAs, a complicated evaluation process can be treated. Since this method based on GAs is flexible, other methods, such as branch exchange, can be put into the GAs method. A modified GAs method, which incorporates branch exchange, is developed. It is found that the developed method is effective in planning power system restoration problems.
The paper [PS150] presents an application of new approach to distribution network planning. The technique is applied to a real problem, in Portugal. The principle behind it is the minimization of the risk, i.e. the possible regret associated to an expansion plan, given a set of uncertainties in the future. General future scenarios are organized in a tree of futures. A diversity of uncertain data are represented by fuzzy values, and fuzzy power flow or reliability models are adopted. The general solution techniques are based on GAs, and decisions are guided through a multicriteria approach.
The paper [PS151] develops a coarse-gain parallel GA for solving a service restoration problem in electric power distribution systems. A power utility performs service restoration in order to restore out-of-service areas at fault. Developing an effective service restoration procedure is a cost-effective approach to improve service reliability and to enhance customer satisfaction. The main objective in service restoration procedure is to restore as much load as possible by transferring de-energized loads via network reconfigurations to other supporting distribution feeders without violating operating and engineering constraints. Details of the parallel GA developed in this paper are described. The proposed method is implemented on transputers for parallel computation. The feasibility of the developed algorithm for service restoration is demonstrated on several distribution networks with promising results.
The paper [PS152] presents an application of GA to service restoration in distribution systems. The feasibility of the proposed method is demonstrated on a typical distribution system model. The result shows that the method can solve the problem efficiently, and this tendency becomes dominant by increasing problem dimensions.
In the paper [PS153] a GA has been adopted in the field of power system planning to select the optimal drop point in a regional network for an external source. Automation in the research process has been improved and the research period has been reduced. The algorithm has also been employed in the Transmission to East China Power System project from the San Xia Hydroplant. The result is excellent.
The paper [PS154] presents a comprehensive overview of a new approach to distribution network planning. The principle behind it is the minimization of the risk, i.e. the possible regret associated to an expansion plan, given a set of uncertainties in the future. General future scenarios are organized in a tree of futures. A diversity of uncertain data are represented by fuzzy values, and fuzzy power flow or reliability models are adopted. The general solution techniques are based on GAs, and decisions are guided through a multicriteria approach. As a result, one may obtain the definition of not a single plan but a strategy for the development of the network.
To ensure a given level of reliability of energy supply, distribution networks should be configured in such a way that each load point may be supplied from alternative sources. The method proposed in the paper [PS155] is aimed at designing such distribution systems with minimal feeder length, energy losses and load imbalance between transformers, subject to voltage drop and capacity constraints. The method is based on the biologically inspired GA. Basic GA procedures adapted to the given problem are presented and five versions of the GA are compared. Test results are reported which demonstrate that the chosen version of the proposed algorithm outperforms a heuristic procedure proposed previously.
The papers [PS156, PS157] presents an application of parallel GAs (PGA) to the optimal long-range GEP. The proposed method considers introduced power limits of each technology, maximum loads at each interval, and load duration curves at each interval. Appropriate string representation for the problem is presented. Binary and decimal coding, and three selection methods are compared. The method is developed on a transputer that is one of the parallel processors. The feasibility of the proposed method is demonstrated using a typical expansion problem with four technologies and five intervals and compared with the conventional dynamic programming and simple GA with promising results.
The fuel cell has been expected to be a very energy efficient and clean device and has now matured technologically to a stage of practical use. In consideration of its relatively small generation capacity, a fuel cell may be installed in a distribution network. The paper [PS158] presents the framework of a method of introducing fuel cells into a radial distribution system. An optimal deployment problem of fuel cells is formulated as a mixed integer programming method. Since this problem has a nonlinear objective function to be minimized, an exhaustive search is required for solution. However, the number of searches causes a combinatorial explosion for large scale systems. Hence, a GA is selected as a solution algorithm to obtain a solution within reasonable computation time. A proposed method is applied to a test distribution system of 69 nodes and the results are examined.
The paper [PS159] presents a GA for large scale and long term scheduling problems. The proposed method shows a new genetic operation that finds the local optimum faster than the simple GA with a low possibility of prematurity and with an efficient encoding/decoding technique. The acceptance probability of a simulated annealing method is included in the algorithm as a criterion for the survival of individuals during the evolution process. The target of the study is to reduce the computing time of the simulated annealing based method and make the solution more accurate than that of the simple GA. In the proposed method, a part of the proposed method is implemented for real scale thermal unit maintenance scheduling which covers several consecutive years.
The paper [PS160] presents an approach for solving transmission expansion planning based on neurocomputing hybridized with GAs. This approach generates suitable initial states, which include past information, of ANNs utilizing GAs. Mingling neurocomputing and GAs, the proposed approach finds many good solutions in a reasonable time making full use of their merits. Computational examples show the effectiveness of the proposed approach by comparison with conventional approaches.
The paper [PS161] presents an economics-based model of sectionalizer allocation in single radial feeder distribution systems. The model considers both cost of energy losses and capital investment in the sectionalizer installation. The cases when sectionalizers are not fully reliable and when they may cause additional short-circuits are investigated. To solve the problem of optimal sectionalizer allocation a GA based procedure is developed. An illustrative example is presented.
Please send Questions, Comments, Suggestions to bencr@hotmail.com