深信不疑网络
极限学习机
偏自我相关函数
组分(热力学)
光伏系统
模式(计算机接口)
超参数
计算机科学
深度学习
人工智能
人工神经网络
工程类
机器学习
自回归积分移动平均
时间序列
物理
电气工程
操作系统
热力学
作者
Peng Tian,Yiman Li,Zhigang Song,Yongyan Fu,Muhammad Shahzad Nazir,Chu Zhang
标识
DOI:10.1016/j.jobe.2023.107227
摘要
Accurate prediction of solar radiation is of great significance to improve the utilization of solar energy for photovoltaic power generation on the roofs of urban buildings. Therefore, a hybrid model, namely the OVMD-PACF-ISSA-DBN-OSELM model, is being proposed in this study for solar radiation prediction. Firstly, the data is decomposed by optimal variational mode decomposition (OVMD) technology. Secondly, the partial autocorrelation function (PACF) technology selected features for components decomposed by OVMD technology. Thirdly, a deep belief network-online sequential extreme learning machine (DBN-OSELM) prediction model is constructed, and sparrow search algorithm (SSA) is utilized to optimize the model hyperparameters. Aiming at the shortcomings of SSA algorithm, Tent map and differential evolution strategy are introduced to improve the SSA algorithm. Finally, the model is used to predict each component, and the prediction results of each component are summarized to obtain the prediction results of solar radiation. To verify that the proposed method can obtain accurate prediction results, experiments are conducted using solar radiation data for January, May and October 2022, and compared with other models. Taking January as an example, compared with shallow model BP, the RMSE, MAE and R of the proposed model are increased by 66.83%, 58.65% and 5.33%, respectively. Moreover, the experiments of the proposed model in other months have excellent results. The current findings demonstrate that the suggested model is a strong contender in the arena of solar radiation prediction, and can consequently provide valuable support in enhancing the efficiency of photovoltaic equipment installed on urban building roofs.
科研通智能强力驱动
Strongly Powered by AbleSci AI