Improved niche genetic algorithm based parameter identification of excitation system considering parameter identifiability

• Main Authors: Xingjie Liu, Liang Yan, Yadong Liu, Liquan Zhao, Jinxing Jie
• Format: Article
• Language: English
• Published: Wiley 2018-10-01
• Series: The Journal of Engineering
• Subjects: convergence; genetic algorithms; power system identification; fuzzy set theory; pattern clustering; partial parameters; excitation system model; identifiability analysis; improved niche genetic algorithm based parameter identification; parameter identifiability; fitness sharing mechanism; fuzzy clustering
• Online Access: https://digital-library.theiet.org/content/journals/10.1049/joe.2018.8816

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.