Catálogo de publicaciones - libros

This paper introduces a hybrid metaheuristic, the Variable Neighborhood Particle Swarm Optimization (VNPSO), consisting of a combination of the Variable Neighborhood Search (VNS) and Particle Swarm Optimization(PSO). The proposed VNPSO method is used for solving the multi-objective Flexible Job-shop Scheduling Problems (FJSP). The details of implementation for the multi-objective FJSP and the corresponding computational experiments are reported. The results indicate that the proposed algorithm is an efficient approach for the multi-objective FJSP, especially for large scale problems.

This paper proposes an Adaptive Comprehensive Learning Particle Swarm Optimizer with History Learning (AH-CLPSO) based on the previous proposed Learning Particle Swarm Optimizer (CLPSO) [1], which is good at multimodal problems but converges slow on single modal problems. A self-adaptation technique is introduced to adjust the learning probability adaptively in the search process and the historical information is used in the velocity update equation to search more effectively. The experiment results show that the history learning strategy and the adaptation technique improves the performance of CLPSO on problems which need fast convergence and achieve comparable results on the problems requiring slow convergence.

An innovative long-period fiber grating (LPG) used for erbium-doped fiber amplifier (EDFA) gain flattening synthesized by the particle swarm optimization (PSO) algorithm is demonstrated. In our problem, we used the topological neighborhood local PSO algorithm to improve the performance, in addition, we used the damp boundary conditions to avoid the particles escaping out of the solve space. The simulated results are in good coincidence with design targets, and proved the capability and effectiveness of the algorithm. In addition, this algorithm is general and can be used for other similar synthesis problems of fiber Bragg gratings (FBGs).

This paper gives a mathematical description of the undirected network capacity definitions and theorems of computing capacities towards the network. Prove of the theorem correctness is also given. According to them, the paper proposes an algorithm of computing the undirected network capacities based on spanning tree and even an advanced optimization algorithm. In addition, it discusses network capacity expansion problems with constraints. And then, the optimization algorithm in the scenario of undirected network with limited costs is outlined, whose feasibility and process are illustrated via examples.

The Traveling Salesman Problem (TSP) is one of the classic NP hard optimization problems. The Dynamic TSP (DTSP) is arguably even more difficult than general static TSP. Moreover the DTSP is widely applicable to real-world applications, arguably even more than its static equivalent. However its investigation is only in the preliminary stages. There are many open questions to be investigated. This paper proposes an effective algorithm to solve DTSP. Experiments showed that this algorithm is effective, as it can find very high quality solutions using only a very short time step.

Chaotic neural networks have been proved to be strong tools to solve the optimization problems. In order to escape the local minima, a new chaotic neural network model called Shannon wavelet chaotic neural network was presented. The activation function of the new model is non-monotonous, which is composed of sigmoid and Shannon wavelet. First, the figures of the reversed bifurcation and the maximal Lyapunov exponents of single neural unit were given. Second, the new model is applied to solve several function optimizations. Finally, 10-city traveling salesman problem is given and the effects of the non-monotonous degree in the model on solving 10-city traveling salesman problem are discussed. The new model can solve the optimization problems more effectively because of the Shannon wavelet being a kind of basic function. Seen from the simulation results, the new model is powerful.

There is a class of GA-hard problems for which classical genetic algorithms often fail to obtain optimal solutions. In this paper, we focus on a class of very typical GA-hard problems that we call decomposable black-box optimization problems (DBBOP). Different from random methods in GA literature, two “deterministic” divide-and-conquer algorithms DA1 and DA2 are proposed respectively for non-overlapping and overlapping DBBOP, in which there are no classical genetic operations and even no random operations. Given any DBBOP with dimension and bounded order , our algorithms can always reliably and accurately obtain the optimal solutions in deterministic way using () function evaluations.

Because the efficiency and efficacy in Simulated Annealing (SA) algorithms is determined by their cooling scheme, several methods to set it have been proposed. In this paper an analytical method (ANDYMARK) to tune the parameters of the cooling scheme in SA for the Satisfiability (SAT) problem is presented. This method is based on a relation between the Markov chain’s length and the cooling scheme. We compared ANDYMARK versus a classical SA algorithm that uses the same constant Markov chain. Experimentation with SAT instances shows that SA using this method obtains similar quality solutions with less effort than the classical one.

In this paper, we present an accelerated micro genetic algorithm for numerical optimization. It is implemented by incorporating the conventional micro genetic algorithm with a local optimizer based on heuristic pattern move and Aitken Δ acceleration method. Performance tests with three benchmarking functions indicate that the presented algorithm has excellent convergence performance for multimodal optimization problems. The number of objective function evaluations required to obtain global optima is only 5.4-11.9% of that required by using conventional micro genetic algorithm.

This study considers inventory of reusable items with shortages and develops a replenishing and production policy to satisfy customer need, and to promote the smart use of resources. Distinct from former researches, our study considers reusable items with shortages to derive the best replenishment and production strategy. A hybrid of numerical analysis and search method is used to derive the minimum total cost of the mathematical model. The result is compared with the case when no shortage is allowed.