Greedy heuristic algorithm for solving series of eee components classification problems*

19th International Scientific Conference Reshetnev Readings 2015

Kazakovtsev, A.L.^1,2 ; Antamoshkin, A.N.^1 ; Fedosov, V.V.^3

Siberian State Aerospace University, Academician M.F. Reshetnev, 31 Krasnoyarskiy Rabochiy prospect, Krasnoyarsk

660037, Russia^1

Siberian Federal University, 79 Svobodny Prospect, Krasnoyarsk, Russia^2

TTC - NPO PM JSC, Molodezhnaya ul.20, Zheleznogorsk, Krasnoyarskiy kray

662970, Russia^3

关键词: Algorithm for solving;    Clustering problems;    Computational experiment;    Greedy heuristics;    Number of clusters;    P-median;    Production batches;    SPACE system;   

Others  :  https://iopscience.iop.org/article/10.1088/1757-899X/122/1/012011/pdf
DOI  :  10.1088/1757-899X/122/1/012011

来源: IOP

【 摘 要 】

Algorithms based on using the agglomerative greedy heuristics demonstrate precise and stable results for clustering problems based on k- means and p-median models. Such algorithms are successfully implemented in the processes of production of specialized EEE components for using in space systems which include testing each EEE device and detection of homogeneous production batches of the EEE components based on results of the tests using p-median models. In this paper, authors propose a new version of the genetic algorithm with the greedy agglomerative heuristic which allows solving series of problems. Such algorithm is useful for solving the k-means and p-median clustering problems when the number of clusters is unknown. Computational experiments on real data show that the preciseness of the result decreases insignificantly in comparison with the initial genetic algorithm for solving a single problem.