中国电机工程学报
中國電機工程學報
중국전궤공정학보
ZHONGGUO DIANJI GONGCHENG XUEBAO
2011年
22期
73-79
,共7页
电力系统%参数辨识%可辨识性分析%轨迹灵敏度
電力繫統%參數辨識%可辨識性分析%軌跡靈敏度
전력계통%삼수변식%가변식성분석%궤적령민도
power system%parameter identification%identifiability analysis%trajectory sensitivity
在模型参数辨识中,具有关联性的参数的不同组合可以得到同样的仿真结果,这将导致它们的辨识结果偏离真实值,因而需要对关联性参数进行识别和评估。根据关联性参数的轨迹灵敏度的线性相关性,研究关联性参数的辨识与评估问题。首先,利用参数轨迹灵敏度的拟合识别出哪些参数具有关联性;然后,给部分关联性参数赋默认值,估计出其他参数;最后,根据轨迹灵敏度的拟合系数,评估当赋给部分参数的默认值偏离真实值时,其他参数辨识结果偏离真实值的大小。算例结果表明该方法能对关联性参数进行正确的识别和鲁棒的估计,当默认值与真值相差不大,或者关联性参数之间的线性度较好时,关联性参数能得到有效的评估。
在模型參數辨識中,具有關聯性的參數的不同組閤可以得到同樣的倣真結果,這將導緻它們的辨識結果偏離真實值,因而需要對關聯性參數進行識彆和評估。根據關聯性參數的軌跡靈敏度的線性相關性,研究關聯性參數的辨識與評估問題。首先,利用參數軌跡靈敏度的擬閤識彆齣哪些參數具有關聯性;然後,給部分關聯性參數賦默認值,估計齣其他參數;最後,根據軌跡靈敏度的擬閤繫數,評估噹賦給部分參數的默認值偏離真實值時,其他參數辨識結果偏離真實值的大小。算例結果錶明該方法能對關聯性參數進行正確的識彆和魯棒的估計,噹默認值與真值相差不大,或者關聯性參數之間的線性度較好時,關聯性參數能得到有效的評估。
재모형삼수변식중,구유관련성적삼수적불동조합가이득도동양적방진결과,저장도치타문적변식결과편리진실치,인이수요대관련성삼수진행식별화평고。근거관련성삼수적궤적령민도적선성상관성,연구관련성삼수적변식여평고문제。수선,이용삼수궤적령민도적의합식별출나사삼수구유관련성;연후,급부분관련성삼수부묵인치,고계출기타삼수;최후,근거궤적령민도적의합계수,평고당부급부분삼수적묵인치편리진실치시,기타삼수변식결과편리진실치적대소。산례결과표명해방법능대관련성삼수진행정학적식별화로봉적고계,당묵인치여진치상차불대,혹자관련성삼수지간적선성도교호시,관련성삼수능득도유효적평고。
It is difficult to identify associated parameters for the reason that different combinations of them can produce the same simulation results. Therefore, in order to build confidence for model parameters, it is necessary to recognize which parameters are associated and then evaluate their credibility. This paper presented identification and assessment methods of associated parameters based on linear dependence of trajectory sensitivities. Firstly, a method to identify associated parameters was proposed, which was based on trajectory sensitivities and could separate associated parameters into several affiliated groups. Then associated parameters were estimated, by assigning default value to some of the representative parameters in the same affiliated group. Lastly, the method was also presented to assess how the default values influence the credibility of the identified parameters. Test studies show that the proposed methods can correctly recognize which parameters are associated and can estimate associated parameters with robustness. When the default values slightly differ from the true ones, or good linearity is kept among the associated parameters, distances between the identification results and the true values of the other associated parameters can be correctly assessed.