基于机器学习的光伏输出功率预测方法研究

光伏发电机组容量在电力系统中的比重日益增大,预测光伏出力对电力系统调度具有极其重要的意义。因为影响光伏发电系统的许多因素随机性较高,使得预测工作难度加大。传统的预测方法对数据的依赖性较强,数据的完整性对预测过程影响很大,因此需要更严谨、便捷的方法使光伏功率的预测工作更加准确、实用。通过对光电站历史数据的探索性分析,对比多种回归预测模型,对影响功率的因素建立神经网络与非线性拟合的组合预测模型。仿真结果表明,组合分步法可以显著降低预测误差,对电网规划、提升新能源发电竞争力、优化调频具有一定的意义。

Research on Photovoltaic Output Power Prediction Method Based on Machine Learning

Nowadays, the capacity of photovoltaic generator set is increasing in power system. It is very important to predict PV output for power system dispatching. Because of the high randomness of many factors affecting photovoltaic power generation, it is difficult to predict. The traditional prediction method has a strong dependence on the data, and the integrity of the data has a great impact on the prediction process. Therefore, more rigorous and convenient methods are needed to make the photovoltaic power prediction more accurate and practical. Based on the exploratory analysis of the historical data of the photoelectric station, several models built up with regression method were compared, and the combined prediction model of neural network and nonlinea fitting were established according to the factors affecting the power. The simulation results show that the combined stepwise method can significantly reduce the prediction error, which has a certain significance for power grid planning, enhancing the competitiveness of new energy generation, and optimizing the frequency modulation.