Optimal blocking schemes for 2n and 2n−p designs
Technometrics
Reducing Local Optima in Single-Objective Problems by Multi-objectivization
EMO '01 Proceedings of the First International Conference on Evolutionary Multi-Criterion Optimization
The measure of Pareto optima applications to multi-objective metaheuristics
EMO'03 Proceedings of the 2nd international conference on Evolutionary multi-criterion optimization
Two-phase Pareto local search for the biobjective traveling salesman problem
Journal of Heuristics
A hybrid TP+PLS algorithm for bi-objective flow-shop scheduling problems
Computers and Operations Research
Improving the anytime behavior of two-phase local search
Annals of Mathematics and Artificial Intelligence
Performance assessment of multiobjective optimizers: an analysis and review
IEEE Transactions on Evolutionary Computation
Model-robust designs for split-plot experiments
Computational Statistics & Data Analysis
Editorial: Special issue on algorithms for design of experiments
Computational Statistics & Data Analysis
Hi-index | 0.03 |
Many industrial experiments involve one or more restrictions on the randomization. In such cases, the split-plot design structure, in which the experimental runs are performed in groups, is a commonly used cost-efficient approach that reduces the number of independent settings of the hard-to-change factors. Several criteria can be adopted for optimizing split-plot experimental designs: the most frequently used are D-optimality and I-optimality. A multi-objective approach to the optimal design of split-plot experiments, the coordinate-exchange two-phase local search (CE-TPLS), is proposed. The CE-TPLS algorithm is able to approximate the set of experimental designs which concurrently minimize the D-criterion and the I-criterion. It allows for a flexible choice of the number of hard-to-change factors, the number of easy-to-change factors, the number of whole plots and the total sample size. When tested on four case studies from the literature, the proposed algorithm returns meaningful sets of experimental designs, covering the whole spectrum between the two objectives. On most of the analyzed cases, the CE-TPLS algorithm returns better results than those reported in the original papers and outperforms the state-of-the-art algorithm in terms of computational time, while retaining a comparable performance in terms of the quality of the optima for each single objective.