基于长短期记忆及自适应Kriging子集模拟优化的风电功率预测方法
A Wind Power Prediction Method Based on Long Short-Term Memory and Adaptive Kriging-Subset Simulation Optimization
付振宇 1王文胤 1凌小明 1张文坤 1陈恒1
作者信息
- 1. 广东电网有限责任公司湛江供电局,广东 湛江 524000
- 折叠
摘要
风电功率的准确预测是目前践行各类监管和运营战略,发展智能电网和先进控制系统等重要技术的前提.提出一种基于长短期记忆及自适应Kriging子集模拟优化的风电功率预测方法.在风电场数据特征提取的基础上,以长短期记忆预测均方根误差为目标,超参数为设计变量,通过基于期望改善的深度耦合自适应Kriging子集模拟优化,进行超参数的高效优化设计,输出最优预测功率.最后,通过案例验证所提方法的预测性能.
Abstract
Accurate prediction of wind power is a prerequisite for implementing various regulatory and operational strategies and developing important technologies such as smart grids and advanced control systems.This paper presents a novel prediction method for wind power based on the long short-term memory and the adaptive Kriging-subset simulation optimization.Based on the extraction of data features from wind farms,with the goal of minimizing the root mean square error of long short-term memory predictions,and with hyperparameters as design variables,efficient hyperparameter optimization design is carried out through deep coupled adaptive Kriging-subset simulation optimization based on the expected improvement,and the output is the optimal predicted power.Finally,the prediction performance of the proposed method is verified through a case study.
关键词
风功率预测/长短期记忆/自适应Kriging/超参数设计/子集模拟优化Key words
wind power prediction/long short-term memory/adaptive Kriging/hyper-parameter design/subset simulation optimization引用本文复制引用
出版年
2024
NETL
NSTL
万方数据