In order to overcome the limitation that the Markov chain Monte Carlo (MCMC) method is unable to better retain the inverse peaking characteristics when generating wind power output sequences

a simulation method of wind power output time series based on the division of intraday time periods is proposed. Firstly

K-means clustering is performed on the daily output curves of each month. Secondly

a day is divided into several time periods of equal length. The same time periods of similar natural days are spliced into a new sequence

and the state intervals containing an equal number of samples are divided according to the output values. The state transfer matrix of the new sequence is solved. Thirdly

stochastic and volatility components are added to the state transfer matrix to simulate the fluctuation

and then the daily curves are generated sequentially by time period. Finally

stochastic and volatility components are incorporated into the state transfer matrix to simulate volatility

and daily curves are then generated by time period. Through the use of case comparisons

it is demonstrated that this method outperforms the traditional K-means MCMC method in terms of basic probability density and autocorrelation function indexes. Furthermore

it has been shown to better restore the inverse peaking characteristics

thereby verifying the effectiveness of the method. The research can serve as a reference point for the generation of wind power data samples in the context of power system planning.