At present, wind farms use a single power ramp prediction model, which has poor generalization ability and low prediction accuracy. By analyzing the two single-power ramp-up prediction models of support vector machine and extreme learning machine, the weight selection methods of these two models are studied, and a combined power hill-climbing prediction model is proposed. The model uses the improved algorithm based on particle swarm optimization algorithm to optimize the weights of the above two models, and forms a new high-precision prediction system. Then the system is modeled and simulated. The simulation results verify the effectiveness of the prediction model, and its prediction accuracy is greatly improved compared with the single prediction model. This work is supported by Open Project of Key Laboratory of Control of Power Transmission and Conversion, Ministry of Education (No. 2016AB14) and National Natural Science Foundation of China (No. 51477099).