随着可再生能源的快速发展,光伏发电在电力系统中占据了重要地位。然而,由于光伏发电的功率输出受天气条件影响较大,具有显著的间歇性和随机性,导致预测难度较高。为了提高光伏发电功率的预测精度,本文提出了一种基于改进的Informer模型与天气数据相结合的混合预测方法。首先,收集并预处理光伏发电相关数据,包括光伏历史功率数据与关键气象因素(如温度、辐照度、湿度等)。其次,通过引入经验小波变换对数据进行模态分解处理气象数据。然后,利用基于网格划分的聚类方法(Grid-Based Clustering, GBC)改进局部敏感哈希(Locality Sensitive Hashing, LSH),之后使用该方法对informer模型进行改进。在本文中,尝试结合GBC来选择查询向量中的几个关键向量,改进informer模型中的对查询向量的筛选,从而提高模型预测的准确性。最后使用模拟退火优化算法对超参数进行优化选择,通过真实数据集的实验验证,与Informer相比,数据的MSE,MAE,RMSE和分别提高了47.42%、43.37%、27.50%和6.54%。综上所述,所提出的混合模型在预测精度、拟合度等方面均能够实现有效的提高。研究表明,该模型能够为光伏发电的运行调度和电网稳定性提供更加可靠的支持。With the rapid development of renewable energy, photovoltaic (PV) power generation occupies an important position in the power system. However, the power output of PV power generation is highly affected by weather conditions with significant intermittency and randomness, which leads to high prediction difficulty. In order to improve the prediction accuracy of PV power, this paper proposes a hybrid prediction method based on the combination of the improved Informer model and weather data. Firstly, PV power related data are collected and pre-processed, including PV historical power data and key meteorological factors (e.g. temperature, irradiance, humidity, etc.). Second, the meteorological data are processed by introducing an empirical wavelet transform for modal decomposition of the data. Then, the Locality Sensitive Hashing (LSH) is improved by using Grid-Based Clustering (GBC), after which the informer model is improved using this method. In this paper, an attempt is made to combine GBC to select several key vectors in the query vectors to improve the filtering of the query vectors in the informer model so as to improve the accuracy of the model prediction. Finally, the simulated annealing optimization algorithm is used to optimize the selection of hyper-parameters, which is experimentally verified by the real dataset, and the MSE, MAE, RMSE and R^2 of the data are improved by 47.42%, 43.37%, 27.50% and 6.54%, respectively,
暂无评论