Abstract：During the ultrashortterm wind power prediction using Markov chains, to solve the poor prediction accuracy due to the using samestate transition probability matrix in different variation scopes of wind power and without considering wind power realtime trends, an improved prediction method using Markov chains was proposed based on Mycielski algorithm. The Mycielski approach was employed to find the longest repetitive sequence from the historical sequence of wind power.The sequence in state transition probability matrix of Markov chain at every moment was redefined.The ultrashortterm wind power prediction was realized by the state transition probability matrix of the Markov chain at every moment. The ultrashortterm wind power prediction was also implemented in a domestic wind farm.The results show that for rootmeansquare error, the modified prediction using Markov chains based on Mycielski algorithm can effectively increase prediction accuracy by 14.15% with high practical value.
2016, Vol. 37 
