论文部分内容阅读
为提高搜索精度和解决样本数据高维性问题,以数值天气预报为基础,提出一种基于投影寻踪和改进状态转移算法优化的风电功率预测模型.该方法首先选取风电场周围多个位置多个高度的气象数值信息,采用投影寻踪主成分分析方法将高维的样本数据投影到低维空间,提取主成分,再建立投影寻踪耦合模型;同时通过加入正交变换的状态转移算法优化最佳投影方向、多项式系数和阈值项,确定网络结构以确保得到最佳模型.以某风电场为实例研究,表明基于投影寻踪和改进状态转移算法的方法可靠性高,能有效解决风电功率预测中存在的预测精度低、数据非线性和高维性等实际难题.“,”In order to improve prediction accuracy and solve the problem of high dimensional sample data,a wind power predicting model based on projection pursuit and improved state transition algorithm was proposed based on numerical weather forecast.Firstly,the numerical weather information of multiple heights and locations around the wind farm were chosen,the projection pursuit principal component analysis method was used to project high dimensional sample data into low dimensional space,extract principal component,then establish the projection pursuit coupled model. Meanwhile,the state transition algorithm added orthogonal transformation was used to optimize the best projection direction,polynomial coefficient and threshold,and determine the network structure to ensure the best model.The case study of certain wind farm shows that the model based on projection pursuit and state transition algorithm has high reliability and can effectively solve the practical problems of low prediction accuracy,data nonlinearity and high dimensionality in wind power prediction.