摘要
风电的间歇性和不确定性给大规模风电并网带来了挑战,风电功率的单点预测已很难满足电网安全稳定运行的需求。提出了一种基于小波神经网络的风电功率区间预测多目标优化模型,改进了基本多目标人工蜂群算法的概率选择作用和约束删减策略,以优化小波神经网络的伸缩因子、平移因子和权值解决了区间预测单目标优化模型下惩罚系数的不合理选择问题,提高了风电功率区间预测可靠性。通过分析与单目标优化方法、传统多目标优化方法下区间预测指标的对比结果,表明所构建的多目标智能优化模型对风电功率区间预测具有更优越的性能,可为电网调度提供决策依据。
Intermittency and uncertainty of wind power brings great challenge to large-scale integration of wind power, and wind power single point prediction cannot meet need of safe and stable operation of power systems. In this paper, an intelligent multi-objective optimized prediction interval（PI） model for wind power is proposed with wavelet neural network（WNN） as its basic prediction model. Probability selection and constraint pruning strategy of basic multi-objective artificial bee colony algorithm（MOABC） are improved to optimize WNN scaling factor, shifting parameter and weights. The proposed method for wind power prediction intervals can avoid unreasonable choosing penalty coefficient in single objective optimization PI model and enhance PI reliability efficiently. Compared to single objective optimized method and traditional multi-objective optimized method, simulation results show that the proposed intelligent multi-objective optimized PI model reveals more superior performance, and it can provide a scheduling support for power systems.
出处
《电网技术》
EI
CSCD
北大核心
2016年第8期2281-2287,共7页
Power System Technology
基金
国家自然科学基金项目(61573167
61572237)
高等学校博士学科点专项科研基金(20130093110011)~~
关键词
风电功率
小波神经网络
区间预测
多目标人工蜂群算法
人工智能
wind power
wavelet neural network
prediction intervals
multi-objective artificial bee colony algorithm
artificial intelligence
