MCMC Wind Power Sequence Modeling Method Considering Climbing Direction

Abstract

Since the traditional Markov Chain Monte Carlo (MCMC) method has the problem that the wind power is stuck in a certain state and it is difficult to jump, this paper proposes an improved MCMC method considering the climbing direction, first dividing the state by the size of the output and the climbing direction, and then determining the duration of each state by changing the state, and then sampling and generating the corresponding number of output samples according to the duration of each state, and sorting these samples according to the climbing direction, and the sorted output is the output timing of the state. This method was used to generate wind power sequences from a wind farm in the Northeast China Sea, and the characteristics of a wind power sequence were compared and analyzed with the original wind power sequence, and the results were better than the wind power sequences generated by the traditional MCMC method.