Projects & Grants
| An innovative modifications of selected variants of differential evolution algorithm | |
|---|---|
| Project Id | SGS04/ÚVAFM/2019 |
| Main solver | doc. RNDr. Petr Bujok, Ph.D. |
| Period | 1/2019 - 12/2019 |
| Provider | Specifický VŠ výzkum |
| State | finished |
| Anotation | Global optimization is by extension of IT presently required in many fields of science and commercial sector, and also public health sector. A problem of global optimization is possible to solve by various techniques; the most popular are Evolutionary algorithms (EA). These methods are models of nature-systems and one of the crucial representatives is Differential evolution algorithms (DE). An efficiency of DE is shown by a huge amount of newly proposed DE variants and also by real applications of new DE variants. The most significant criterion of EAs efficiency are world-competitions, where variants of DE are frequently in the front positions.A goal of this project is the innovative development of enhanced DE variants and their application. The first part of the goal is the innovation of mechanism controlling the diversity of population in DE, which was developed in previous projects. Furthermore, innovation will be performed by a strict intervention to a current population in order to achieve the required level of diversity.Beside diversity control, the second part of the goal is the innovative development of cooperative (parallel) variants of DE algorithm to study the parameters settings. Models developed in previous projects found out as successful, therefore the control parameters will be studied more deeply.A part of research of the project will be also focused on development of new adaptive approach for settings of F and CR parameters in DE algorithm. A methodology of development of new DE variants is based on the comparison of new methods with existing state-of-the-art algorithms. A criterion for comparison will be the application of algorithms on test problems and also on real-world problems. The project is focused mainly on a real application because this is the main purpose of the development of new optimization techniques. |
