2022, issue 1, p. 11-18

Received 04.05.2022; Revised 26.05.2022; Accepted 28.06.2022

Published 30.06.2022; First Online 03.08.2022

https://doi.org/10.34229/2707-451X.22.1.2

UDC 004.94

On Investigation of Natural Algorithms and Their Complex Application for Optimization of Logistics Tasks

Victor Andriichuk,   Violeta Tretynyk *

The National Technical University of Ukraine "Igor Sikorsky Kyiv Polytechnic Institute"

Introduction. At present, with the increase in the number of data and the complexity of various processes in many branches of the national economy, the classical methods of answering have ceased to meet the needs. So, for example, if you consider the logistics industry (this and transportation, and various cartographic services for consumers), we see how dynamically developing these areas. To construct optimal routes, if we continued to use old algorithms, we would have to spend a huge amount of computer resources and time to get results. Instead, the study finds new methods that make it possible to speed up and simplify the solution of problems. So, for example, for logistics problems, instead of classical combinatorial problems, algorithms are being replaced by environmental scientists. They were called natural algorithms. For example, scientists have noticed that ants find the shortest route to food using the iteration and intensity of a pheromone trace that leaves previous ants who are moving along the path. This idea formed the basis of the algorithm of the ant colony. Even if in reality ants find the optimal way more difficult than the simplified idea of researchers, it turned out that this simplification was enough to find an optimal route between some points of the type like ants looking for an optimal route food.

To develop a complex meta-algorithm that would solve the problem of logistics network, the following methods are taken into account: natural algorithms, salesman problem, dynamic programming methods, combinatorial approaches, algorithms for complex data analysis.

This paper considers the application of methods of natural algorithms to solve the problem of coordinated logistics.

The purpose of the paper. The aim of the work is to increase the efficiency of using natural algorithms in logistics problems. In general, the main tasks of logistics are to forecast the amount of costs, products and resources under certain circumstances. However, the required amount of material, information, financial, service and other data flows is not always available in logistics tasks, many variables are unknown. Then heuristic methods of algorithms come to the rescue to solve such problems of an applied nature.

Results. The software implementation of the model in the Python programming language was performed. Some methods of dynamic programming, fuzzy logic were used and the hyperopt library was used to implement the script.

Keywords: natural algorithms, logistics, ant algorithm, meta-algorithm.

Cite as: Andriichuk V., Tretynyk V. On Investigation of Natural Algorithms and Their Complex Application for Optimization of Logistics Tasks. Cybernetics and Computer Technologies. 2022. 1. P. 11–18. (in Ukrainian) https://doi.org/10.34229/2707-451X.22.1.2

References

1.     Dorigo M. “Ottimizzazione, apprendimento automatico, ed algoritmi basati su metafora naturale (Optimization, Learning, and Natural Algorithms)”, thesis for attainment a scientific degree “Doctorate in Systems and Information Electronic Engineering”, Politecnico di Milano, 1992.

2.     Stützle T., López-Ibáñez M., Pellegrini P., Maur M., M. de Oca, Birattari M., Maur M., Dorigo M., “Parameter Adaptation in Ant Colony Optimization”. Technical Report, IRIDIA, Université Libre de Bruxelles, 2010. https://doi.org/10.1007/978-3-642-21434-9_8

3.     Hutter F., Hoos H.H., Leyton-Brown K. Sequential Model-Based Optimization for General Algorithm Configuration. Proceedings of the conference on Learning and Intelligent OptimizatioN (LION 5). Rome, Italy: Springer-Verlag, 2011. https://doi.org/10.1007/978-3-642-25566-3_40

4.     Bergstra J., Bardenet R., Bengio Y., Kegl B. Algorithms for hyper-parameter optimization. Advances in Neural Information Processing Systems. 2011.

5.     Snoek J., Larochelle H., Adams R. Practical Bayesian Optimization of Machine Learning Algorithms. Advances in Neural Information Processing Systems. 2012. Bibcode: 2012arXiv1206.2944S. arXiv:1206.2944.

6.     Thornton C., Hutter F., Hoos H., Leyton-Brown K. Auto-WEKA: Combined selection and hyperparameter optimization of classification algorithms. Knowledge Discovery and Data Mining. 2013. Bibcode: 2012arXiv1208.3719T. arXiv:1208.3719. https://doi.org/10.1145/2487575.2487629

ISSN 2707-451X (Online)

ISSN 2707-4501 (Print)