基于混沌时间序列的风电场输出功率预测方法  

The Method of Wind Farm Power Forecast Based on Chaotic Time Series

作　　者：李媛[1] 张鹏飞[1] 邢作霞[2] 

机构地区：[1]沈阳工业大学理学院,辽宁沈阳110870 [2]沈阳工业大学新能源学院,辽宁沈阳110870

出　　处：《控制工程期刊（中英文版）》2015年第2期14-21,共8页Scientific Journal of Control Engineering

基　　金：受辽宁省博士启动基金支持资助（20141069）.

摘　　要：为了提高风电场输出功率目前预测的准确率，建立基于混沌时间序列的预测模型。风电场功率输出除受众多非线性因素的影响之外，还与自身的混沌性相关。本文首先按时间间隔对时间序列p进行重构，再采用wolf法求得各时间序列的最大Lyapunov数，以判断其混沌性，以C—C方法确定最优延迟时间T和嵌入维数m进行相空间重构，并以此确定混沌神经网络拓扑结构；最后采用处理后的实测数据进行神经网络训练，使之具有功率预测功能。以东北地区某风电场为实例，结果表明，基于混沌时间序列的风功率预测方法准确度高且容易实现。The prediction model based on chaotic time series is established to improve the prediction accuracy of wind farm power output. As the wind farm power output is influenced by nonlinear factors, and is also related to the chaotic sex itself, the time series p is refactored and the largest Lyapunov exponent of time series is gained with Wolf method to determine its chaos. The I topology architecture for chaotic neural network is determined by determining the optimal delay time z with C-C method and refactoring the phase space through embedding dimension m. Finally the measured data after processing is of power prediction function through neural network training. Take a wind farm in northeast China as an example, the results show that wind power prediction method based on chaos-neural network is of high accuracy and easy to implement.

关 键 词：风电场 功率预测 混沌时间序列 混沌神经网络 相空间重构 最大LYAPUNOV指数 延迟时间 嵌入维数 

分 类 号：TP[自动化与计算机技术]