基于关联挖掘的无功优化关键参数智能辨识方法

针对传统无功优化中关键参数设置过程繁琐且设置结果不合理的问题,首先,给出一种基于斜率分段归并的曲线划分策略,用于对预测区间进行智能划分;其次,采用一种标准化欧式距离—动态时间弯曲(ED-DTW)混合策略,用于不同数据集间相似度的计算;最后,提出一种基于数据关联挖掘的无功优化参数智能辨识框架,用于对数据库内的历史数据进行挖掘。仿真采用实际电网数据对整个挖掘过程进行分析,挖掘结果显示,提出的辨识框架能自动给出参数的时段划分和设置结果,将挖掘得到的参数结果用于实际控制中,表明该方法获得的划分结果符合负荷峰谷特征,且相比传统方法,在减小电压偏差和提高电压合格率上效果更好。

Association Mining Based Intelligent Identification Method of Key Parameters for Reactive Power Optimization

Owing to the problems of tedious setting process of key parameter and unreasonable setting results during the performing of traditional reactive power optimization, firstly, a curve division strategy based on slope segment merging is presented to divide the forecasting interval intelligently. Secondly, a hybrid strategy of standard Euclidean distance-dynamic time warping(ED-DTW)is used to calculate the similar distance of different data sets. Finally, an intelligent identifying framework of reactive power optimization parameter based on data mining association is proposed to mine the data in the database. The simulation is analyzed by adopting real grid data. The mining results show that the parameter identifying framework could automatically give parameter time division and setting results. The mining parameter used in real control shows that the calculation speed of this approach is fast and has better effect on reducing voltage deviation and improving qualified voltage rate.