This study focuses on the forecasting of renewable energy generation and electricity consumption, exploring how evolutionary algorithms and hyper-heuristic algorithms can be employed for automatic parameter optimization and model selection in time series forecasting to enhance accuracy and adaptability. It begins by reviewing the challenges faced by traditional forecasting models when dealing with the stochastic and nonlinear characteristics of energy data. Then, it systematically introduces the principles and application processes of metaheuristic algorithms and hyper-heuristic algorithms, highlighting their performance and advantages in various scenarios such as seasonal exponential smoothing, hybrid short-term load forecasting, and renewable energy generation prediction. Empirical results based on real electricity and renewable energy data demonstrate that these evolutionary computation approaches, including metaheuristics and hyper-heuristics, can effectively optimize model parameters, improve forecasting accuracy, and provide more precise and robust forecasting support for power system planning and smart grid management.