基于新能源发电能力预测高效并行计算技术的就地消纳方案优化探究  

Optimization of on-site consumption scheme based on efficient parallel computing technology for predicting new energy generation capacity

作　　者：牟雪鹏 王斌 陈巨龙 杨世平 王伟 MOU Xuepeng;WANG Bin;CHEN Julong;YANG Shiping;WANG Wei(Guizhou Power Grid Co.,Ltd.,Power Grid Planning and Research Center,Guiyang 550000,China)

机构地区：[1]贵州电网有限责任公司电网规划研究中心,贵阳550000

出　　处：《自动化与仪器仪表》2025年第3期269-273,共5页Automation & Instrumentation

摘　　要：由于新能源发电具有不稳定性的特点,因此,研究以长短期记忆网络算法、K-means算法和位置尺度t分布共同实现风光发电预测,达到高效并行计算目的。在Gaussian Copula函数拟合风光出力避免弃光、弃风后,采用优化的离散粒子群算法模型实现风光消纳。结果表明,该方法在两个场景下的成本均不超过32.50万元,风光消纳为293.73 kW·h,对比优化前的281.35 kW·h,风光消纳率提升了4.40%。同时预测风电和光伏的平均绝对误差最小,分别为0.05和0.25,模型准确率高。该方法能够最大化利用风能和光伏发电,不仅提高了发电预测的精度,还降低了系统的运行成本。这为解决新能源发电的不稳定性和提高风光消纳提供了理论依据和实用方法。Due to the instability of new energy generation,this study aims to achieve efficient parallel computing by using long short-term memory network algorithm,K-means algorithm,and location scale t-distribution to jointly predict wind and solar power generation.After fitting the Gaussian Copula function to avoid abandoning light and wind,an optimized discrete particle swarm optimization algorithm model is used to achieve wind and solar energy consumption.The results show that the cost of this method in both scenarios does not exceed 32500000 yuan,and the wind and solar consumption is 293.73 kwh.Compared with the optimized 281.35 kwh,the wind and solar consumption rate has increased by 4.40%.The average absolute error of predicting wind power and photovoltaic simultaneously is the smallest,with values of 0.05 and 0.25,respectively,indicating high accuracy of the model.This method can maximize the utilization of wind and photovoltaic power generation,not only improving the accuracy of power generation prediction,but also reducing the operating cost of the system.This provides theoretical basis and practical methods for solving the instability of new energy generation and improving wind and solar energy consumption.

关 键 词：风光出力 就地消纳 LSTM 离散粒子群算法 新能源发电 

分 类 号：TM73[电气工程—电力系统及自动化]