A physical constraint-oriented electric vehicle charging curve generation method and system

Abstract

The application discloses a physical constraint-oriented electric vehicle charging curve generation method and system, which comprises the following steps: collecting historical real electric vehicle charging power data containing time series and performing normalization processing; constructing and training a conditional generation model; after obtaining the normalized synthetic charging power time series, performing reverse normalization operation on the normalized synthetic charging power time series to obtain a candidate electric vehicle charging power curve; performing multi-dimensional physical property checking on the candidate electric vehicle charging power curve through a composite physical loss function; combining the composite physical loss function with a basic generation loss function to generate a total loss function; taking the total loss function as a training optimization target to perform end-to-end optimization on the conditional generation model; and generating an electric vehicle charging power curve meeting the constraints by using the optimized conditional generation model. The application can synthesize high-fidelity charging data which is highly consistent in statistical characteristics, behavior patterns and dynamic processes.

Images