DEVELOPMENT AND APPLICATION OF REFINED LOAD FORECASTING SOFTWARE FOR ACTIVE DISTRIBUTION NETWORK
Study the refined load forecasting method under the background of the new active distribution network:based on the numerical weather forecast,the convolutional neural network Resnet is used to forecast the photovoltaic power,the GRU algorithm is used to forecast the electrical load considering the load characteristics and meteorological factors,and the refined forecasting results of the active load are obtained by accumulating the forecasting results of the photovoltaic power generation component and the electrical load component.In addition,the software architecture and functional modules of refined load forecasting are designed based on the distribution cloud master station.The research on dynamic network reconfiguration of distribution network based on short-term load forecasting is carried out,and the dynamic optimization strategy of 24 hours ahead of the day is given considering the load sequence segmentation.Finally,through the verification of an example in the distribution area of a city,the results show that the refined load forecasting is more accurate than the direct equivalent load forecasting for the source load,and the dynamic network reconfiguration based on the refined load forecasting can reduce the load balance and optimize the photovoltaic consumption.
PV powerdistribution networksdistributed power generationload forecastingpower forecastingdynamic network reconfiguration
NETL
NSTL
万方数据
