Wyniki wyszukiwania - BazTech

Ograniczanie wyników

Powiadomienia systemowe

Sesja wygasła!

Znaleziono wyników: 3

Liczba wyników na stronie

Wyniki wyszukiwania

Wyszukiwano:
w słowach kluczowych: variable neighborhood search

Sortuj według:

Ogranicz wyniki do:

A Meta-Heuristic Algorithm Combining Between Tabu and Variable Neighborhood Search for the Minimum Latency Problem

Ban H. B., Nguyen D. N.

Fundamenta Informaticae

2017

Vol. 156, nr 1

21--41

Minimum Latency Problem (MLP) is a class of NP-hard combinatorial optimization problems which has many practical applications. In this paper, we investigate the global structure of the MLP solution space to propose a suitable meta-heuristic algorithm for the problem, which combines Tabu search (TS) and Variable Neighborhood Search (VNS). In the proposed algorithm, TS is used to prevent the search from getting trapped into cycles, and guide VNS to escape local optima. In a cooperative way, VNS is employed to generate diverse neighborhoods for TS. We also introduce a novel neighborhoods’ structure for VNS and present a constant time operation for calculating the latency cost of each neighboring solution. Extensive numerical experiments and comparisons with the state of the art meta-heuristic algorithms in the literature show that the proposed algorithm is highly competitive, providing the new best solutions for several instances.

Deterministically guided differential evolution for constrained power dispatch with prohibited operating zones

Jasper J., Victoire T. A. A.

Archives of Electrical Engineering

2013

Vol. 62, nr 4

593--603

Department of Electrical Engineering, Anna University Regional Centre, Coimbatore, India This paper presents a new approach to solve economic load dispatch (ELD) problem in thermal units with non-convex cost functions using differential evolution technique (DE). In practical ELD problem, the fuel cost function is highly non linear due to inclusion of real time constraints such as valve point loading, prohibited operating zones and network transmission losses. This makes the traditional methods fail in finding the optimum solution. The DE algorithm is an evolutionary algorithm with less stochastic approach to problem solving than classical evolutionary algorithms.DE have the potential of simple in structure, fast convergence property and quality of solution. This paper presents a combination of DE and variable neighborhood search (VNS) to improve the quality of solution and convergence speed. Differential evolution (DE) is first introduced to find the locality of the solution, and then VNS is applied to tune the solution. To validate the DE-VNS method, it is applied to four test systems with non-smooth cost functions. The effectiveness of the DE-VNS over other techniques is shown in general.

Designing WDM networks by a variable neighborhood search

Melian-Batista B., Höller H., Voss S.

Journal of Telecommunications and Information Technology

2006

nr 4

15-20

With the ever-rising data volume that is demanded by the market, network planning in order to minimize the necessary investment while meeting the demands is constantly an important task for the network providers. Synchronous digital hierarchy (SDH) and wavelength division multiplex (WDM) form the core of many current backbone networks. In order to solve the provisioning and routing problem in such WDM networks, we develop a variable neighborhood search (VNS) metaheuristic. VNS is a metaheuristic that combines series of random and improving local searches based on systematically changed neighborhoods. An integer flow formulation is modeled in AMPL and solved by CPLEX in order to obtain optimal solutions as a reference for the heuristic.