Climate‐Driven Low‐Carbon Dispatch Strategy for Photovoltaic‐Storage‐Charging Microgrids

ABSTRACT The operation of photovoltaic‐storage‐charging microgrids is increasingly challenged by the volatility of renewable resources and demand patterns driven by dynamic climate conditions. To address these challenges, this paper proposes a climate‐driven low‐carbon dispatch strategy for photovoltaic‐storage‐charging microgrids. First, to accurately capture system uncertainties associated with fine‐scale atmospheric dynamics, a convolutional neural network–long short‐term memory (CNN–LSTM) deep learning model is constructed based on a 15‐min dispatch resolution. Simultaneously, a multifeature long short‐term memory (LSTM) model, integrating temporal periodicity and historical dependencies, is developed to predict weather‐sensitive electric vehicle charging loads. Second, to enhance system resilience against climate‐induced power fluctuations, the operational constraints of the energy storage are modelled explicitly, incorporating a battery degradation cost model to quantify asset lifespan loss. Building on this, a dual‐objective optimisation framework is formulated that internalises dynamic carbon trading costs into the economic dispatch. This model coordinates the power interaction among the photovoltaic system, energy storage and the main grid to maximise total operating revenue while minimising carbon footprints. Simulation results verify that the proposed strategy effectively balances economic efficiency and carbon reduction, achieving a system carbon emission reduction of 39.82%, demonstrating superior adaptability to climate‐driven variations in renewable generation and load demand.