分散式风电中MCP方法对风速和发电量预测的误差研究

Research on the errors of MCP in predicting wind speed and power generation in decentralized wind power

目前行业内多采用测量相关预测(MCP)方法得到完整的测风数据,但在实际的分散式项目风资源评估业务中无法分辨各种MCP方法的可靠性。随机选取400余个样本,针对年平均风速和年发电量,分析不同的同步时长、目标站和参照站相关性对MCP方法误差的影响,定量给出6种MCP方法存在的误差。研究表明,LLS、VS和BSR算法在年平均风速的预测上表现更好,而WBL、VR、BSR和SS算法则在年发电量的预测上表现更好,其中BSR算法可以兼顾年平均风速和发电量的预测。

At present,measurement correlation prediction(MCP)method is widely used to obtain complete wind measurement data,but in the actual wind resource evaluation business of decentralized projects,it is difficult to find out the reliability of various MCP methods.More than 400 samples were randomly selected,and the effect of different synchronization durations,the correlation between different target stations and reference stations on the error of the MCP method were analyzed for the annual average wind speed and annual power generation,and the errors of six MCP methods were quantitatively given.The results show that LLS,VS and BSR algorithms perform better in the prediction of annual average wind speed,while WBL,VR,BSR and SS algorithms perform better in the prediction of annual power generation.BSR algorithm can both work well in the prediction of annual average wind speed and power generation.