PV Hosting Capacity Evaluation based on Feature Data

The penetration of photovoltaics (PV) in the power system is gradually increasing, and the future power system will face the one-time integration of large-scale PV and the accessible capacity of PV is a difficult issue to determine. For a region, the evaluation of its PV hosting capacity (PVHC) is complicated, including a large amount of data preparation and the establishment of a mathematical model of power grid. Therefore, an efficient and rapid evaluation method is urgently needed which can efficiently use feature data for PVHC calculation with a certain degree of accuracy. To fill the gaps, an efficient, accurate and rapid PVHC evaluation method based on feature data is proposed. Firstly, considering constraints such as reverse power transmission, a PVHC evaluation model is established. Secondly, by generating a large number of boundary values for constraints, a large number of PVHC under different constraints are calculated. After that, based on the BP neural network, the fitting relationship between PVHC and feature data is obtained. When calculating the PVHC of the actual distribution grid, by simply collecting feature data, PVHC can be calculated with a trained BP neural network, greatly reducing workload and computational complexity. Finally, the random forest algorithm is used to extract the importance of each influencing factor, and the transformer capacity is calculated to be the most important factor affecting PVHC.