首页|基于小生境遗传算法与径向基代理模型的短期风电功率预测

基于小生境遗传算法与径向基代理模型的短期风电功率预测

扫码查看
原文链接
  • NETL
  • NSTL
  • 万方数据
为提高短期功率预测精度,以赋予风电被电网资产更大规模消纳的优势,建立一种基于主导特征影响因素和小生境遗传算法改进的径向基代理模型的滚动式短期(0～72 h)风电功率预测模型.首先,基于罚函数和排挤机制的小生境技术对传统基本遗传算法进行改进,以径向基代理模型(RBF)作为建模基础,利用改进后的遗传算法以反传误差极小为目标函数对RBF模型的连接权值进行优化,借助其寻优能力来获取最佳权值,以达成对RBF网络的改进和二次训练;然后,基于主导特征气象因素,结合改进的RBF模型最终建立N-SGA-RBF风电出力预测模型,对风电场连续3日0～72 h输出功率进行预测;最后,对N-SGA-RBF模型、RBF模型以及BP模型做预测结果趋势变化、各采样点绝对/相对误差分布、发电预测预报准确率和合格率的对比.以新疆东部某风电场实测数据进行算例验证分析,仿真结果表明,所建预测模型具有较高的精度.
SHORT-TERM WIND POWER PRIDICTION BASED ON NICHE GENETIC ALGORITHM AND RADIAL BASIS SURROGATE MODEL
In order to improve the forecast accuracy of short-term power and promote more wind power being consumed by power grid,this paper establishes a rolling short-term(0-72 h)wind power prediction model based on dominant feature influencing factors and niche genetic algorithm improved neural network.Firstly,the niche technology based on penalty function and crowding mechanism is used to improve the traditional basic genetic algorithm.With the radial basis surrogate model(RBF)as the modeling basis,the improved genetic algorithm is used to optimize the connection weights of the RBF model with the minimum back-propagation error as the objective function.With its optimization ability,the optimal weights are obtained to achieve the improvement and secondary training of the RBF network;Then,based on dominant meteorological factors and an improved RBF model,an N-SGA-RBF wind power output prediction model is established to predict the output power of the wind farm for 0-72 hours for three consecutive days;Finally,the N-SGA-RBF model,RBF model,and BP model are compared for trend changes in prediction results,absolute and relative error distribution at each sampling point,accuracy and qualification rate of power generation prediction.An example verification analysis is conducted using measured data from a wind farm in eastern Xinjiang.The simulation results show that the established prediction model has high accuracy.

wind powerforecastingradial basis surrogate modelniche genetic algorithmintelligent optimization

刘沛汉、尹翠、贾娜、樊小朝、杨青斌

展开 >

新疆工程学院能源工程学院,乌鲁木齐 830023

国网新疆电力有限公司超高压分公司,乌鲁木齐 830002

中国电力科学研究院有限公司可再生能源并网全国重点实验室,南京 210009

风电功率 预测 径向基代理模型 小生境遗传算法 智能优化

新疆维吾尔自治区自然科学基金(2022)新疆维吾尔自治区高等学校基本科研业务费科研项目(2024)国家自然科学基金新疆维吾尔自治区科技厅重大专项

2022D01A240XJEDU2024P083522660182022A01001-2

2024

10.19912/j.0254-0096.tynxb.2023-0615

太阳能学报
中国可再生能源学会

太阳能学报

CSTPCD北大核心
影响因子：0.392
ISSN：0254-0096
年,卷(期)：2024.45(8)