电网技术
電網技術
전망기술
POWER SYSTEM TECHNOLOGY
2015年
1期
208-214
,共7页
李驰%刘纯%黄越辉%王伟胜
李馳%劉純%黃越輝%王偉勝
리치%류순%황월휘%왕위성
风电波动特性%时间序列%自组织映射聚类%序贯抽样%概率统计
風電波動特性%時間序列%自組織映射聚類%序貫抽樣%概率統計
풍전파동특성%시간서렬%자조직영사취류%서관추양%개솔통계
wind power fluctuation characteristics%time series%self-organizing map (SOM) clustering%sequential sampling%probability and statistics
掌握风力发电的随机、波动与间歇特性,并在此基础上构建风电出力时间序列模型对于电力系统规划与运行具有重要意义。提出了一种构造未来风电出力场景的新方法。研究了风电波动过程特性,在极值点处将历史风电出力时间序列划分为波动,采用自组织映射(self-organization map, SOM)神经网络将波动聚类为大波动、中波动、小波动和低出力波动。波动变化规律可用高斯函数来定量表达。基于风电波动过程特性阐述了建模方法,将月份按波动出力特性进行分类,分别统计波动类间转移概率和类内统计参数的概率分布,按月序贯抽样风电波动类别与各统计参数,计算并模拟得到风电出力时间序列。对中国某省部分风电场进行了仿真模拟,统计特征参数的对比分析结果验证了上述方法的有效性。
掌握風力髮電的隨機、波動與間歇特性,併在此基礎上構建風電齣力時間序列模型對于電力繫統規劃與運行具有重要意義。提齣瞭一種構造未來風電齣力場景的新方法。研究瞭風電波動過程特性,在極值點處將歷史風電齣力時間序列劃分為波動,採用自組織映射(self-organization map, SOM)神經網絡將波動聚類為大波動、中波動、小波動和低齣力波動。波動變化規律可用高斯函數來定量錶達。基于風電波動過程特性闡述瞭建模方法,將月份按波動齣力特性進行分類,分彆統計波動類間轉移概率和類內統計參數的概率分佈,按月序貫抽樣風電波動類彆與各統計參數,計算併模擬得到風電齣力時間序列。對中國某省部分風電場進行瞭倣真模擬,統計特徵參數的對比分析結果驗證瞭上述方法的有效性。
장악풍력발전적수궤、파동여간헐특성,병재차기출상구건풍전출력시간서렬모형대우전력계통규화여운행구유중요의의。제출료일충구조미래풍전출력장경적신방법。연구료풍전파동과정특성,재겁치점처장역사풍전출력시간서렬화분위파동,채용자조직영사(self-organization map, SOM)신경망락장파동취류위대파동、중파동、소파동화저출력파동。파동변화규률가용고사함수래정량표체。기우풍전파동과정특성천술료건모방법,장월빈안파동출력특성진행분류,분별통계파동류간전이개솔화류내통계삼수적개솔분포,안월서관추양풍전파동유별여각통계삼수,계산병모의득도풍전출력시간서렬。대중국모성부분풍전장진행료방진모의,통계특정삼수적대비분석결과험증료상술방법적유효성。
Mastering random, intermittent characteristics of wind power and constructing a wind power time series model is of great significance for power system planning and operation. A new method to simulate future wind power scenarios is proposed. Characteristics of wind power fluctuation process is analyzed, historical wind power time series are divided into wind fluctuations at the minimum-value point, wind fluctuations are clustered within large fluctuation, medium fluctuation, small fluctuation and low output fluctuation by visualization self-organizing map (SOM) clustering algorithm, fluctuation variation laws are quantitative expression with gauss function. Based on above analysis, the modeling method is proposed: months are classified according to monthly characteristics, transfer probability between fluctuation categories and probability distribution of statistical parameters within categories are counted, wind fluctuation category, fluctuation statistical parameters are sampled sequentially on a per-month basis, then calculated and combined to achieve the simulated wind power time series. Several wind farms of a province in China are simulated, and through comparative analysis on statistical feature parameters, the validity of the proposed model is verified.