Dynamic time-delay discrete grey model based on GOWA operator for renewable energy generation cost prediction

Renewable energy is vital for environmental protection and economic development. Accurate forecasting of power generation costs is essential for expanding the renewable energy market. However, the nonlinearity, volatility, and lagging characteristics of generation cost data present significant challenges to accurately predicting trends in generation costs. Therefore, this paper proposes a multivariable dynamic time-delay discrete grey model (DTDDGM-GOWA(1, N)) based on the generalized ordered weighted average (GOWA) operator for power generation cost prediction. Firstly, a new dynamic time-delay term is constructed by introducing a dynamic time-delay function, which dynamically selects the optimal form of the model, enhancing its adaptability and effectively utilizing the information in renewable energy generation cost data. Secondly, a new parameter estimation method is proposed based on the GOWA operator, L2 regularization, and genetic algorithm, enabling the model to accurately capture the nonlinear and volatile characteristics of the cost data, thus improving the forecasting accuracy. Furthermore, Monte Carlo simulations and probability density analysis validate the robustness of the model. Finally, the model is applied to forecasting the generation costs of global onshore wind, solar, and bioenergy. Compared to existing models, the proposed model better adapts to the dynamic characteristics of generation cost data, offering superior prediction and robustness.