基于深度学习的光伏发电功率短期预测

Short-term Prediction ofPhotovoltaic Power Generation Power Based on Deep Learning

光伏发电受到气象状况及环境因素的影响,具有很强的随机性与间歇性,给电网的安全运行带来了电压波动和闪变、孤岛效应等问题.为了准确提高光伏发电系统短期输出功率预测,建立了改进野狗算法优化GRU(IDOA-GRU)的预测模型,在野狗算法中引进莱维飞行机制和自适应因子加强算法的寻优能力.通过某光伏发电站实测数据进行仿真实验,结果表明,改进野狗算法优化GRU神经网络的预测模型相比传统的GRU预测模型和野狗算法优化GRU(DOA-GRU)具有更高的预测精度.

Photovoltaic power generation is affected by meteorological conditions and environmental factors,with strong randomness and intermittency,which brings serious problems to the safe operation of the power grid.In order to accurately and improve the short-term output power prediction of photovoltaic power generation system,a prediction model based on the improved Dingo algorithm Optimized GRU(IDOA-GRU)is established.Levy flight mechanism and adaptive factor are introduced into the Dingo algorithm to enhance the optimization ability of the algorithm.Through the simulation experiment on the measured data of a photovoltaic power station,the results show that the prediction model of IDOA-GRU neural network optimized by the improved Dingo algorithm has higher prediction accuracy than the traditional GRU prediction model and Dino algorithm optimized GRU(DOA-GRU).