CAJ | 학술논문

主动解列最优断面搜索是依据广域测量信息，在大电网遭受大扰动失步崩溃之前，依据实时工况和运行方式，快速准确求取电力孤岛划分的紧急策略。然而，在实际大系统的求解中，计算复杂度呈几何指数增长，是一个 NP 难题。提出了一种半监督谱聚类算法，首先采用最小复合有功潮流冲击的目标函数和机组同调/分离等相关约束构建详细解列断面搜索模型，然后将最优断面搜索的优化求解过程，映射为约束谱聚类对静态图分割的松弛解求取过程，最后通过改进的PAM聚类算法选择最优主动解列断面。上述过程，在不丢失全网信息前提下，降低了时间复杂度，IEEE 118标准算例和四川电网实际系统的仿真验证，证明了该算法的正确性、有效性和快速性。
주동해렬최우단면수색시의거엄역측량신식，재대전망조수대우동실보붕궤지전，의거실시공황화운행방식，쾌속준학구취전력고도화분적긴급책략。연이，재실제대계통적구해중，계산복잡도정궤하지수증장，시일개 NP 난제。제출료일충반감독보취류산법，수선채용최소복합유공조류충격적목표함수화궤조동조/분리등상관약속구건상세해렬단면수색모형，연후장최우단면수색적우화구해과정，영사위약속보취류대정태도분할적송이해구취과정，최후통과개진적PAM취류산법선택최우주동해렬단면。상술과정，재불주실전망신식전제하，강저료시간복잡도，IEEE 118표준산례화사천전망실제계통적방진험증，증명료해산법적정학성、유효성화쾌속성。
The optimal controlled partitioning is such an emergency strategy to split power grid into isolated islands rapidly and accurately according to real-time operating conditions and operation modes of power grid before its collapse due to severe disturbance. However, during the solution of actual large power grid the computation complexity increases in geometric-exponent regularity, so it is an NP-hard problem and to solve this problem a semi-supervised spectral clustering algorithm is proposed. Firstly, taking the minimal composite power-flow disruption is taken as the objective function and the unit coherence, unit separation, active power and reactive power as constraints a detailed partitioning search model is constructed; secondly, the optimization solution of the optimal controlled partitioning is mapped to the process of achieving the relaxation solution of the constraint spectral clustering aimed at the partitioning of the static graph; finally, based on the improved partitioning around medoid (PAM) clustering algorithm the optimal controlled partitioning sections can be selected. Without losing the information of whole power grid, above-mentioned process can decrease the time complexity. The correctness, effectiveness and rapidity of the proposed algorithm are validated by simulation results of IEEE 118-bus standard system and actual Sichuan power grid.