Identifying Factors Contributing to Poor Performance of Near-Real-Time Power Flow

Accurate near real time monitoring is needed for Distribution Management Systems (DMS), and power flow-based methods are commonly used in practice for this purpose. However, near real time power flow results are not always accurate because of the poor load estimation obtained through the bus load allocation (BLA) procedure used in this approach. This paper focuses on this issue and proposes a data analytics-based method to parse through data and identify factors/parameters that are significant and relevant to poor near real time power flow solution. The proposed method uses a Binary logistic regression method to identify these significant parameters for the feeders having poor BLA performance. Test results show that the significant parameters can help to pinpoint the cause(s) for poor BLA performance.