2019 International Seminar on Electron Devices Design and Production (SED) | 2019
Hybryd Approach for Computer-Aided Design Problems
Abstract
The paper discusses the solution of optimization problems at the design stage. The formulation of the multi-objective optimization problem and a model based on the bee colony behavior are described. To solve this problem, the authors proposes a modified hybrid approach based on two methods: modeling of the bee colony behavior and differential evolution. Also, the global improvement operation is introduced as a modification of the proposed approach. In the developed hybrid algorithm, the search is performed in the neighborhood of different solutions, which allows to avoid falling into local optima. A software is developed to carry out a computational experiment on benchmark. The experimental results show that the developed hybrid approach allows you to get more effective solutions, because the obtained results are on average better by 7% than a simple genetic algorithm and a standard bee algorithm, that indicates the effectiveness of the developed approach. The time complexity of the algorithm lies within O(n log n) − O(n2).