CAJ | 학술논문

在电力系统负荷特性统计指标和气温日益积累大数据背景下，有效提取数据之间关联特征对电力系统规划和运行具有重大意义。为此，提出一种气温对负荷特性指标影响及其内在关联特征数据挖掘的方法。考虑气温季节特征进行分季度建模，首先通过物理关系和皮尔森相关系数获得气温和负荷特性指标任意两因素之间的相关性特征；然后在多变量时间序列平稳性检验基础上，对水平不平稳的同阶单整时间序列进行协整检验和向量误差修正(vector error correction， VEC)建模以获取其长期同步运动趋势及短期波动特性；进一步通过对变量差分化后的平稳时间序列的向量自回归(vector auto-regression，VAR)建模提取多因素变化量间的动态关系，结合格兰杰因果检验挖掘因素变化量之间的因果引导关系。针对华中某省级电网2006年至2010年负荷特性实际统计数据及相应气温数据的实例分析验证了文中方法的正确性和有效性，方法已在实际电网负荷特性统计分析中得到应用。
재전력계통부하특성통계지표화기온일익적루대수거배경하，유효제취수거지간관련특정대전력계통규화화운행구유중대의의。위차，제출일충기온대부하특성지표영향급기내재관련특정수거알굴적방법。고필기온계절특정진행분계도건모，수선통과물리관계화피이삼상관계수획득기온화부하특성지표임의량인소지간적상관성특정；연후재다변량시간서렬평은성검험기출상，대수평불평은적동계단정시간서렬진행협정검험화향량오차수정(vector error correction， VEC)건모이획취기장기동보운동추세급단기파동특성；진일보통과대변량차분화후적평은시간서렬적향량자회귀(vector auto-regression，VAR)건모제취다인소변화량간적동태관계，결합격란걸인과검험알굴인소변화량지간적인과인도관계。침대화중모성급전망2006년지2010년부하특성실제통계수거급상응기온수거적실례분석험증료문중방법적정학성화유효성，방법이재실제전망부하특성통계분석중득도응용。
With the big data of load characteristics statistical indexes and temperature indexes increasing, it is significant to obtain correlative features of electric data effectively for planning and operation in power system. In this paper, a approach which can extract the temperature influence on load characteristics indexes and the internal correlation features was proposed. Considering temperature and load seasonal characteristics, this paper conducted modeling analysis in each season respectively. First, the qualitative analyses of potential physical relations among the indexes as well as quantitative calculation via Pearson correlation coefficient of historical data were coordinated to draw the correlation features between two factors. Then, based on the stationary test results on the origin series, long-term synchronous movement trend and short-term fluctuant characteristics were obtained via co-integration test on the difference sequence of non-stationary index and vector error correction (VEC) model. Further, through vector auto- regression (VAR) model on stationary time series of variables after difference, the dynamic correlation of multi-variable and the causal guiding relationship among related variables were acquired integrated with Granger Causality. The statistics of load characteristics from 2006 to 2010 of a provincial power grid in Central China demonstrate the effectiveness and correctness of this method, and the method has been applied in the actual load characteristic statistical analysis of power grid.