区块链架构下风光储电站超短期功率预测方法
Fusion Method for Blockchain-based Cyber-physical System of Wind-solar-storage Power Stations and Ultra-short-term Power Prediction
风光储电站的功率预测是其自身及大电网调度的基础,其信息物理系统作为保障预测过程的前提则需要更强的安全性。满足调度要求的功率预测需保证高精度、高频更新、短时间尺度及多时间尺度。本文提出了一种基于区块链的风光储电站超短期功率预测方法,基于区块链的风光储电站信息物理系统双层架构,可弱化对储能电站物理拓扑结构的依赖性,消除应对攻击时的脆弱性;改进时空图卷积网络算法可增加参数规模灵活性和参数间相关性,以提高预测精度。将二者在区块链的智能合约中结合,即形成基于区块链的风光储电站超短期功率预测方法,保证了预测过程及预测结果的安全性。最后,通过算例验证了所提出方案的有效性,电负荷预测平均误差至少下降1.22%,热负荷预测平均误差至少下降4.70%,冷负荷预测平均误差至少下降1.60%,可再生能源出力预测平均绝对误差至少下降2.17%。
The power prediction of wind-solar-storage power stations is fundamental to both the dispatch of the stations themselves and the dispatch of the larger power grid. To ensure the prediction process, the cyber-physical system (CPS) that underpins this process requires stronger security. Power forecasts to meet dispatch requirements need to ensure high accuracy, high-frequency updates, short time scales, and multiple time scales. In this paper, a blockchain-based ultra-short-term power prediction method for wind-solar-storage power stations is proposed. The blockchain-based two-layer architecture of wind-solar-storage power station cyber-physical system can weaken the dependence on the physical topology of the storage power station and eliminate the vulnerability when responding to attacks. The improved spatiotemporal graph convolutional network algorithm can increase parameter scale flexibility and inter-parameter correlation to improve prediction accuracy. Combining these two in a blockchain smart contract, forming a blockchain-based ultra-short-term power prediction method for wind-solar-storage power stations, ensures the security of the prediction process and the prediction results. At the end, the proposed method is validated through case studies, demonstrating its effectiveness. The average error in electricity load forecasting is reduced by at least 1.22%, the average error in heat load forecasting is reduced by at least 4.70%, the average error in cooling load forecasting is reduced by at least 1.60%, and the average absolute error in renewable energy output forecasting is reduced by at least 2.17%.
| [1] |
黄碧斌,张运洲,王彩霞. 中国“十四五”新能源发展研判及需要关注的问题[J]. 中国电力,2020,53(1):1-9. |
| [2] |
朱建康. 风力发电和光伏发电并网的问题和对策[J]. 光源与照明,2022(7):201-203. |
| [3] |
薛福,马晓明,游焰军. 储能技术类型及其应用发展综述[J]. 综合智慧能源,2023,45(9):48-58. |
| [4] |
王思文,岳建华,席雪飞. 典型风能参数波动条件下考虑风电消纳路径的技术研究[J]. 内蒙古农业大学学报(自然科学版),2019,40(4):64-69. |
| [5] |
李建林,卢冠铭,游洪灏, |
| [6] |
丁达,曹杰. 信息物理融合系统网络安全综述[J]. 信息与控制,2019,48(5):513-521+527. |
| [7] |
杨挺,刘亚闯,刘宇哲, |
| [8] |
马钊,安婷,尚宇炜. 国内外配电前沿技术动态及发展[J]. 中国电机工程学报,2016,36(6):1552-1567. |
| [9] |
张晶晶,陈博进,吴佳瑜, |
| [10] |
龚立,王先培,田猛, |
| [11] |
昌力,曹荣章,吉斌, |
| [12] |
何云华,罗明顺,胡晴, |
| [13] |
高亚静,孙永健,杨文海, |
| [14] |
刘鸿达,孙旭辉,李沂滨, |
| [15] |
|
| [16] |
|
| [17] |
彭秉刚,潘振宁,余涛, |
| [18] |
朱继忠,董瀚江,李盛林, |
| [19] |
|
| [20] |
史佳琪,张建华. 基于多模型融合Stacking集成学习方式的负荷预测方法[J]. 中国电机工程学报,2019,39(14):4032-4042. |
| [21] |
范斌,李豫明,郭强强, |
| [22] |
|
| [23] |
鲁斌,霍泽健,俞敏. 基于LSTNet-Skip的综合能源系统多元负荷超短期预测[J]. 中国电机工程学报,2023,43(6):2273-2282. |
| [24] |
赖昌伟,黎静华,陈博, |
| [25] |
吕伟杰,方一帆,程泽. 基于模糊C均值聚类和样本加权卷积神经网络的日前光伏出力预测研究[J]. 电网技术,2022,46(1):231-238. |
| [26] |
邢晨,张照贝. 基于改进时间卷积网络的短期光伏出力概率预测方法[J]. 太阳能学报,2023,44(2):373-380. |
| [27] |
王育飞,付玉超,薛花. 计及太阳辐射和混沌特征提取的光伏发电功率DMCS-WNN预测法[J]. 中国电机工程学报,2019,39(S1):63-71. |
| [28] |
王晓霞,俞敏,霍泽健, |
| [29] |
许诘翊,刘威,刘树, |
| [30] |
徐敏,李万伟,张中丹, |
| [31] |
龚钢军,蔡贺,杨德龙, |
国网山东省电力公司科技项目(2024A-027)
/
| 〈 |
|
〉 |