Method for Generating Massive New Energy Output Scenarios Based on Variational Autocoding and Hidden Variable Modeling
In recent years,the access rate and frequency of various new energy sources such as wind power and photovoltaic in the power system have been continuously increasing,making it increasingly important to implement more accurate modeling for the uncertainty of new energy output.In order to simplify the generation steps of random scenes and improve the efficiency and accuracy of scene generation,a data-driven modeling approach was adopted,based on the unsupervised variational automatic encoding Gaussian mixture model,to create a new method for generating massive new energy output random scenes.The high-dimensional training set data was mapped into a well-structured low dimensional hidden variable space through an encoder for probability modeling,sampling.And then restore it back to the original dimension through a decoder to obtain the scene set.Compared with existing probability methods,the method could complete the learning of spatiotemporal and volatility features of wind power and photovoltaic training data without supervision,extract typical output curves from them,and quickly form a dataset that matched the observation characteristics without the need for scene reduction.The proposed algorithm has been verified to be reliable,reasonable,and effective through the actual historical new energy output calculation of a certain southern province's power grid city.
scenario analysis methodvariational automatic encoderdeep learningscene generationnew energy power system
NETL
NSTL
万方数据
