基于IPSO-BPNN的楼宇屋顶光伏出力功率超短期预测
Ultra-Short-Term Prediction of Building Rooftop Photovoltaic Output Power Based on IPSO-BPNN
鲁娟 1何鑫 2李明海 2邓琨升3
作者信息
- 1. 西安建筑科技大学设计研究总院有限公司,陕西 西安 710055
- 2. 西安建筑科技大学 机电工程学院,陕西 西安 710055
- 3. 西安建筑科技大学 信息与控制工程学院,陕西 西安 710055
- 折叠
摘要
在建筑光伏一体化技术的背景下,准确预测屋顶光伏输出功率对于优化建筑能源管理和确保光伏电力的稳定并网至关重要.提出了一种基于 IPSO-BPNN的楼宇屋顶光伏出力功率超短期预测模型,该模型引入Sine混沌序列初始化和精英粒子反向学习策略,改进了基本的 PSO 算法,并利用此算法对基本BPNN模型的超参数进行优化,从而实现了对屋顶光伏出力功率更加准确的预测.预测模型性能测试实验表明,所提出的IPSO-BPNN预测模型在不同季节的预测准确性和稳定性都有显著提高.该模型能够准确预测屋顶光伏发电功率,为建筑光伏一体化系统的稳定运行和能源管理提供切实可行的解决方案.
Abstract
In the context of building-integrated photovoltaic technology,accurate prediction of rooftop photovoltaic output power is crucial for optimizing building energy management and ensuring the stable grid connection of PV electricity.Based on this,this paper proposes a rooftop PV output power ultra-short-term prediction model based on improved particle swarm optimization and backpropagation neural network(IPSO-BPNN).This model improves the basic particle swarm optimization(PSO)algorithm by introducing Sine chaotic sequence initialization and elite particle reverse learning strategy,and utilizes this algorithm to optimize the hyperparameters of the basic BPNN model,thereby achieving more accurate prediction of rooftop PV output power.Performance testing experiments of the prediction model demonstrate significant improvements in prediction accuracy and stability across different seasons.The proposed IPSO-BPNN model accurately forecasts rooftop PV electricity generation,providing a practical solution for the stable operation and energy management of building-integrated photovoltaic systems.
关键词
建筑光伏一体化/屋顶光伏/反向传播网络/粒子群算法/光伏出力功率预测Key words
building-integrated photovoltaics/rooftop photovoltaics/backpropagation neural network/particle swarm optimization algorithm/photovoltaic output power prediction引用本文复制引用
出版年
2024
NETL
NSTL
万方数据