基于时间序列模式匹配的电能质量区域化治理

Regionalization Control for Power Quality Based on Time Series Pattern Matching

针对以往基于机理建模进行稳态电能质量污染治理的局限性,该文采用数据驱动的方式,提出一种基于时间序列模式匹配的电能质量污染区域化治理策略。依据配电网各节点指标监测数据建立较长时间尺度的电能质量时间序列,采用考虑关键趋势转折点的分段线性表示提取时间序列主要局部趋势特征及其时间特征,构建时间序列模式特征矩阵。提出一种基于动态时间弯曲的趋势–时间距离方法对各节点时间序列实施模式匹配,测量节点间电能质量的长期耦合度。定义耦合度分区判则,划分电能质量治理区域,确定各区域主导控制节点。以IEEE 14节点配电系统进行算例仿真分析,表明了所提策略的合理性和有效性,验证了所提方法的准确性和适用性。

Focusing on the limitations of steady-state power quality controls based on the mechanism modeling, the data-driven approach was used, and a regionalization control strategy based on time series pattern matching was proposed for power quality. Relatively long time-scale power quality time series were constructed from the monitoring data at nodes in a power distribution network. A piecewise linear representation taking into account critical turning points of trends was employed to extract major local trend features and their time features, thereby establishing pattern feature matrices of the time series. A trend-time distance (T-TD) method based on dynamic time warping (DTW) was developed to implement the pattern matching on the time series to measure the long-term coupling degrees of power quality among nodes. A partition criterion according to the coupling degrees was defined to divide control areas of power quality. Dominant control nodes were determined in the identified areas. Case simulation analyses on the IEEE 14-node power distribution system were conducted to indicate the rationality and effectiveness of the proposed strategy and to verify the accuracy and applicability of the proposed method.