基于高斯混合模型的光伏发电功率概率区间预测

随着世界经济的绿色发展,大力发展可再生能源逐渐成为共识。可再生能源中太阳能的开发利用已成为当前能源转型中的重要领域,并在很多科技发达国家得到了较广泛的应用。高精度的光伏发电功率预测对电力系统的优化调度、安全运行十分重要。由于光照强度和能见度等会影响太阳能发电量的随机性,提出一种基于高斯混合模型的光伏发电功率概率区间预测方法,通过利用K-means算法将光伏发电历史数据按天气进行划分,以划分后的预测误差为统计样本,采用高斯混合模型进行拟合并使用期望最大化算法估计模型参数,通过计算指定置信水平下的置信区间进行光伏发电功率概率区间预测。仿真结果表明所提方法在进行光伏发电功率区间预测时的性能评价指标均优于典型单一分布模型,证明了所提方法的准确性和适用性。

Research on prediction of photovoltaic power generation probability interval based on Gaussian mixture model

With the green development of the world economy, it has gradually become a consensus to develop renewable energy. The development and utilization of solar energy in renewable energy has become an important field in the current energy transition, and has been widely used in many technologically developed countries. High-precision photovoltaic power generation prediction is very important to the optimal dispatch and safe operation of the power system. Because the randomness of solar power generation is affected by light intensity and visibility, a method for predicting the probability interval of photovoltaic power generation based on the Gaussian mixture model was proposed in this paper. The K-means algorithm was used to divide historical photovoltaic power generation data by weather, then a Gaussian mixture model was utilized to fit the divided prediction errors, and an expectation maximization algorithm was applied to estimate model parameters. Calculating the confidence interval under the given confidence level was used to obtain the probability interval prediction results of photovoltaic power generation. The results show that the performance evaluation indexes of the proposed method are better than the typical single distribution model when forecasting the photovoltaic power interval, which proves the accuracy and applicability of the proposed method.