International Journal of Metaheuristics (6 papers in press)
by Kayode Owa, Lisa Jackson, Tom Jackson
Abstract: A solution approach for many challenging and non-differentiable optimization tasks in industries is the use of non-deterministic meta-heuristic methods. Some of these approaches include Particle Swarm Optimization (PSO), Genetic Algorithm (GA), and Simulated Annealing (SA). However, with the implementation usage of these robust and stochastic optimization approaches, there are still some predominant issues such as the problem of the potential solution being trapped in a local minima solution space. Other challenges include the untimely convergence and the slow rate of arriving at optimal solutions. In this research study, a tripartite version (PSO-GA-SA) is proposed to address these deficiencies. This algorithm is designed with the full exploration of all the capabilities of PSO, GA and SA functioning simultaneously with a high level of intelligent system techniques to exploit and exchange relevant population traits in real time without compromising the computational time. The design algorithm further incorporates a variable velocity component that introduces random intelligence depending on the fitness performance from one generation to the other. The robust design is validated with known mathematical test function models. There are substantial performance improvements when the novel PSO-GA-SA approach is subjected to three test functions used as case studies. The results obtained indicate that the new approach performs better than the individual methods from the fitness function deviation point of view and in terms of the total simulation time whilst operating with both a reduced number of generations and populations. Moreover, the new novel approach offers more beneficial trade-off between exploration and exploitation of PSO, GA and SA. This novel design is implemented using an object oriented programming approach and it is expected to be compatible with a variety of practical problems with specified input-output pairs coupled with constraints and limitations on the available resources.
Keywords: genetic algorithms (GA); particle swarm optimization (PSO); simulated annealing (SA); nonlinear programming (NLP); non-deterministic polynomial time hard (NP-hard).
A multilevel hyper-heuristic for solving MAX-SAT
by Mourad Lassouaoui, Dalila Boughaci, Belaid Benhamou
Abstract: Ahyper-heuristic is a high-level method that manages a set of low-level
heuristics to solve various problems in a problem-independent manner. In this
paper, we propose a new selection hyper-heuristic with the multilevel paradigm.
The multilevel paradigm refers to the process of dividing large problems. Then,
each sub-problem is being solved to reach a complete solution, using the resulting
solution from a previous level as a starting solution at the next level. For the
heuristic selection element, a choice function is combined with simple random
to select the adequate low-level heuristic at any iteration during the search. For
analysis purposes, several variants of hyper-heuristics are implemented. Max-
SAT is used as the test-bed. The experimental results revealed that the multilevel
paradigm together with a new hybrid heuristic selection mechanism provides a
substantial performance improvement. A comparison with two known state of the
art methods that are GSAT and WALKSAT algorithms is given to further show
the efficiency of our method.
Keywords: Hyper-heuristic; Multilevel paradigm and Max-SAT.
A Unified Framework for Routing Problems with a Fixed Fleet Size
by Stefanie Kritzinger, Fabien Tricoire, Karl F. Doerner, Richard F. Hartl, Thomas Stutzle
Abstract: Vehicle routing problems constitute the core of many operations research efforts. Early studies introduced tailor-made solutions for each variant of a vehicle routing problem, but unified frameworks have emerged more recently. These approaches typically generalize across many vehicle routing problems and implement a method for tackling the generalized problem. In line with this, this research proposes a generic method for solving several fixed fleet vehicle routing problems capacitated vehicle routing, open vehicle routing, vehicle routing with soft and hard time windows, open vehicle routing with soft and hard time windows, and time-dependent vehicle routing with soft and hard time windows by transforming them into a time-dependent vehicle routing problem with soft time windows, solved by a variable neighborhood search using a unique parameter setting, regardless of the original problem. Computational tests using standard benchmark instances from prior literature show that genericity does not come at the expense of solution quality. Moreover, the algorithm yields competitive results and some new best known solutions are obtained for the vehicle routing problem with soft time windows, the open vehicle routing problem with hard time windows, and time-dependent vehicle routing problem with hard time windows.
Keywords: Vehicle routing problem; multiple attributes; metaheuristic framework; variable neighborhood search.
Examining the Effects of Construction Heuristics and Problem Structure on Solution Quality of the Vehicle Routing Problem with Split Deliveries and Time Windows
by Marcus McNabb, Jeffery Weir, Shane Hall
Abstract: This paper investigates a practical extension of the vehicle routing problem (VRP): the VRP with split deliveries and time windows (SDVRPTW). While the SDVRPTW has not received much attention in literature, the papers hint that the underlying problem structure and ultimately methods required in generating high quality solutions may differ significantly from the classical VRP. In particular, this paper uses a structured design of experiments to investigate the SDVRPTW, to include testing different construction heuristics, the effect of varying ratios of customer demand to vehicle capacity, and the impact of splitting loads. Results indicate construction method does not substantively impact solution quality while local search operators with faster run times tend to generate higher quality solutions where solution quality is primarily total distance traveled by the fleet of delivery vehicles.
Keywords: heuristic; vehicle routing problem; time windows; split delivery; ant colony optimization; GRASP; construction; local search.
Predicting Nationwide Road Fatalities in the U.S.: A Neural Network Approach
by Gokhan Eglimez, Deborah McAvoy
Abstract: Road crashes are among the top five leading causes of deaths in the U.S. even though the 8 national trend in fatal crashes has reached to the lowest level since 1949. Therefore, this paper introduces a nonparametric 10 prediction model, Artificial Neural Network (ANN), to assist policy makers in minimizing fatal 11 crashes across the United States. Seven input variables from four safety performance input 12 domains while fatal crashes was utilized as the single output variable for the scope of the 13 research. ANN was utilized and the best neural network model 14 was developed out of 1000 networks. The proposed neural network model predicted data 15 with 84 percent coefficient of determination. In addition, developed ANN model was 16 benchmarked with a multiple linear regression model and outperformed in all performance 17 metrics including r, R-square and the standard error of estimate.
Keywords: Artificial Neutral Networks; Multivariate Regression Analysis; Highway Safety; U.S. Road Fatalities; Prediction.
Is the Bee Colony Optimisation Algorithm Suitable for Continuous Numerical Optimisation?
by Milos Simic
Abstract: Bee Colony Optimisation (BCO) is a Nature-inspired swarm metaheuristic for solving hard optimisation problems. It has successfully been applied to various areas of science, industry, and engineering. However, all those cases belong to the field of combinatorial optimisation. This paper is among the first to test BCOs capacities for solving continuous numerical optimisation problems. We found that the performance of the algorithm depended on the settings of its parameters and characteristics of the optimisation problems to which it was applied. We examined for which types of numerical functions our implementation of improvement-based BCO, known as BCOi, performed well and which classes it was not able to handle successfully. Also, following the Design of Experiments (DoE) approach, we analysed how the parameters of the algorithm affected its performance and provided some useful explanations that might hold for other applications of our implementation and other versions of BCO.
Keywords: Nature-inspired metaheuristics; Swarm intelligence; Bees foraging principles; Optimization problems; Numerical optimisation.