Accurate forecasting of photovoltaic (PV) power generation is crucial for conventional power dispatch planning and system operation coordination.To this end

this paper proposes a forecasting method using a combined autoregressive moving average-back propagation (ARMA-BP) neural network model.First

an analysis reveals that PV generation data exhibit both linear and nonlinear characteristics.To address this

the ARMA model is employed to capture linear trends

while the BP neural network is used to model nonlinear patterns.A hybrid ARMA-BP model is thus constructed to leverage the strengths of both approaches.The effectiveness of the proposed model is verified through case studies at an actual PV power station.Compared with standalone ARMA and BP models

the combined ARMA-BP model achieves significantly improved prediction accuracy.The results offer a valuable refe rence for grid operators in optimizing generation planning.