Linear Regression Based Power Flow Analysis for Multi-Area Networks with Data Partition

Power grid interconnection is an effective way for energy interchange among subareas to improve energy utilization. The conventional integrated Power Flow (PF) method is effective to analyze system operation state under the circumstance of complete power network information. However, for multi-area interconnected networks, the independent regulator for each subarea is generally incapable to access the private system information from others, which makes the information barriers. To generate the same PF solution as that by integrated method, an optional approach is to separately run PF Calculation (PFC) for each subarea alternated with the adjustment of state variables for boundary buses. Thus, this study proposes a decomposition-coordination PF model with two layers of coordination. Moreover, incorporated with the Ridge Regression, an improved Locally Weighted Linear Regression (LWLR) approach is established in treating the adjustment of boundary variables. Case studies based on IEEE300 and IEEE RTS-1996 test systems demonstrate the efficiency of the proposed method to generate PF results with high accuracy and convergence. Furthermore, the proposed PF method is capable to identify the subarea with potential questionable data i n case of PFC divergence, contributing to analyzing the cause of numerical ill-conditions