Capacity configuration of green hydrogen energy storage system for wind farm group based on correction condition generated adversarial network

Semi supervised learning represented by Conditional Generative Adversarial Networks (CGAN) can take into account wind power fluctuations and generate a set of output scenarios. The generated data can be input into a hydrogen energy storage capacity configuration model to support optimization solutions. To this end, this article designs a Corrective Conditional Generative Adversarial Network (CCGAN) and constructs a conditional corrector based on wind power prediction errors to identify and correct the label information input to the generator under prediction inaccuracy events and wind power climbing events; Then, taking the comprehensive cost of energy storage capacity and the cost of wind abandonment in each wind farm as the objective functions, a multi-objective optimization model for green hydrogen energy storage capacity configuration is constructed.