| Autor: |
Vladimir Sudakov, Alexander Batkovsky, Yuri Titov |
| Jazyk: |
ruština |
| Rok vydání: |
2020 |
| Předmět: |
|
| Zdroj: |
Современные информационные технологии и IT-образование, Vol 16, Iss 2, Pp 338-350 (2020) |
| Druh dokumentu: |
article |
| ISSN: |
2411-1473 |
| DOI: |
10.25559/SITITO.16.202002.338-350 |
| Popis: |
The approach that uses fuzzy time to determine the lead time of a knowledge-intensive project requires solving the problem of assigning employees to tasks. In this case, for each employee and each task that the employee can perform, a fuzzy task execution function is assigned. The distribution of employees by tasks is successfully solved by a modification of the Ant Colony Method that works with a decision graph. But the speed and accuracy of the method depends on the optimality of its parameters, and the algorithm is subject to "looping", a situation when all agents move along the same path in the decision graph, putting many weights on it and not being able to choose another route in the graph at the next iterations solutions. It is proposed to solve these problems by resetting the decision graph with different methods of determining the moment of "looping". The looping moment is proposed to be determined by the statistical parameters calculated at one iteration of the algorithm. The loop detection algorithm, in which it is determined whether new solutions were found during iteration, showed better performance. But this approach requires storing all found solutions, which works well if the calculation of the criterion takes a lot of simulation time. In addition, the idea of resetting the decision graph allows us to solve some problems of setting ineffective parameters of the Ant Colony Method. To speed up the process of finding rational paths in the work, the possibility of entering various initial weights after resetting the decision graph is considered. The most effective way will be to add weights from 2 to 5 of the best paths found by the agents during the operation of the Ant Colony Method. In the future, it is proposed to consider the multicriteria assignment problem and various algorithms for using fuzzy sets in the scheduling of tasks. |
| Databáze: |
Directory of Open Access Journals |
| Externí odkaz: |
|
