Improved niche genetic algorithm based parameter identification of excitation system considering parameter identifiability
| Autor: | Jinxing Jie, Liquan Zhao, Xingjie Liu, Liang Yan, Yadong Liu |
|---|---|
| Jazyk: | angličtina |
| Rok vydání: | 2018 |
| Předmět: |
Fuzzy clustering
Computer science Niche pattern clustering Energy Engineering and Power Technology excitation system model System model genetic algorithms improved niche genetic algorithm based parameter identification Genetic algorithm fuzzy set theory fitness sharing mechanism parameter identifiability Ecological niche convergence General Engineering identifiability analysis power system identification Local convergence Identification (information) lcsh:TA1-2040 Identifiability fuzzy clustering partial parameters lcsh:Engineering (General). Civil engineering (General) Algorithm Software |
| Zdroj: | The Journal of Engineering (2018) |
| DOI: | 10.1049/joe.2018.8816 |
| Popis: | In order to obtain the parameters of excitation system accurately, and improve the accuracy and efficiency of parameter identification much further, an improved niche genetic algorithm was adopted, which can overcome local convergence of genetic algorithm by the fitness sharing mechanism that could increase population diversity. Moreover, the proposed algorithm could avoid the disadvantage of usual niche genetic algorithm that is difficult in determining the niches. By using the method of fuzzy clustering, niche groups can be created dynamically. In addition, regarding the problem that the partial parameters of excitation system cannot identify stably because of the relevance among these parameters, this study analysed the identifiability of excitation system model primarily, separated associated parameters into several parameter sets then. This type of relevance would be removed by assigning typical values to representative parameters, which belong to different sets. Identifiability analysis can not only avoid identifying parameters blindly, but also increase the credibility of the results. The test study shows that the proposed method can acquire accurate and reliable parameter values. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|