Сибирский федеральный университет

Тип публикации: доклад, тезисы доклада, статья из сборника материалов конференций

Конференция: International Joint Conference on Computational Intelligence, IJCCI 2016; Porto; Porto

Год издания: 2016

Ключевые слова: algorithm control, dynamical system, evolutionary strategies, linear differential equation, Meta-heuristic, Restart operator, system identification

Аннотация: In this paper a meta-heuristic for improving the performance of an evolutionary optimization algorithm is proposed. An evolutionary optimization algorithm is applied to the process of solving an inverse mathematical modelling problem for dynamical systems. The considered problem is related to the complex extremum seeking problem. The objective function and a method of determining a solution perform a class of optimization problems that require specific improvements of optimization algorithms. An investigation of algorithm efficiency revealed the importance of designing and implementing an operator that prevents population stagnation. The proposed meta-heuristic estimates the risk of the algorithm being stacked in a local optimum neighbourhood and it estimates whether the algorithm is close to stagnation areas. The metaheuristic controls the algorithm and restarts the search if necessary. The current study focuses on increasing the algorithm efficiency by tuning the meta-heuristic settings. The examination shows that implementing the proposed operator sufficiently improves the algorithm performance.

Журнал: IJCCI 2016 - Proceedings of the 8th International Joint Conference on Computational Intelligence

Выпуск журнала: 1

Номера страниц: 178-185

• Ryzhikov I. (Department of System Analysis and Operations Research,Siberian State Aerospace University)
• Semenkin E. (Department of System Analysis and Operations Research,Siberian State Aerospace University)
• Sopov E. (Department of System Analysis and Operations Research,Siberian State Aerospace University)

• РИНЦ (eLIBRARY.RU)