摘要
风速的随机性和间歇性等特点使得目前风电场参数预测模型存在较大的预测误差,对此提出了采用马尔科夫链(MC)方法对模型的预测误差进行修正。分别求出参数的实际值与模型预测值之间的误差序列,利用模糊C-均值聚类算法对其进行状态划分;根据各误差状态计算出MC状态转移概率矩阵,进而计算模型预测误差修正值,最终得到精度较高的预测值。采用MC方法分别对广义回归神经网络(GRNN)模型、T-S模糊神经网络模型以及Elman神经网络模型的预测误差进行修正,并应用MC修正后的3种模型对山西某风电场测风塔不同步长风速进行预测仿真实验研究,分析讨论了MC对各预测模型误差的修正效果。仿真结果表明,所提出的误差修正方法能够有效提高测风塔风速预测精度,为预测模型的误差修正提供了一种有效的实用的方法。
Focused on the characteristics of randomness and intermittency of wind speed,there are large forecasting errors of pa-rameter models of wind farm.The method of the Markov Chain( MC) is proposed to correct the forecasting error of models.Firstly,the error series between the actual values and the forecasting values of the parameter are obtained.Secondly,the fuzzy C-average clustering is used to divide transfer states,then,state transition probability matrixs are calculated according to the error states.Fi-nally,the correction values of the forecasting error are calculated,and higher forecasting precision is obtained.MC method is used to correct forecasting errors of GRNN model,T-S model and Elman model respectively,and the three kinds of model revised by MC are applied to the simulation experiment research on the different steps of wind speed forecasting of wind tower in a wind farm in Shanxi province.And the correction effects of MC are emphatically discussed.Simulation results show that the proposed error correction method can improve the accuracy of wind speed forecasting of wind tower effectively,which is an effective and useful method for the forecasting models.
出处
《电子技术应用》
北大核心
2016年第7期114-118,共5页
Application of Electronic Technique
基金
国家自然科学基金(51277127)
关键词
误差修正
马尔科夫链
预测模型
风速预测
error correction
Markov Chain
forecasting models
wind speed forecasting
作者简介
高淑杰(1989-),女,硕士研究生,主要研究方向:复杂系统建模与智能控制。
田建艳(1966-),女,通信作者,教授,博士,主要研究方向:复杂系统建模与智能控制,E-mail:tut_tianjy@163.com。
王芳(1976-),女,副教授,博士,主要研究方向:智能信息研究与模式识别。