Probabilistic Prediction Analysis And Research Of Wind Power Based On Bayesian Method

With the increasing maturity and rapid development of wind power technology,the capacity of wind power grid-connected has increased greatly,and wind power grid connection will have a significant impact on the stable operation of power systems.Therefore,high-precision prediction of wind power is of great significance to economic and social development.In order to provide decision makers with more information on decision planning and reliability assessment.Probabilistic interval prediction analysis of wind power based on Bayesian method.The main work of this paper is as follows.Firstly,through the wind power data released by the Global Wind Energy Association,the current status and future development trend of wind power generation are analyzed,and the significance of wind power generation probability research and the probabilistic prediction results of wind power at home and abroadThen,a naive Bayesian wind power probability interval prediction method based on rough set attribute reduction and particle group weight optimization is proposed.The rough set method is used to select the appropriate input of naive Bayes and the particle swarm optimization algorithm is used to optimize the output weight of naive Bayes,thus improving the prediction accuracy.Based on the operational data of a wind farm in the northwest,the simulation is verified.The comparison of various methods shows that it has better engineering use value.Furthermore,by analyzing the probability distribution of point prediction error,the wind power prediction interval is obtained.A cloud model error fitting wind power interval prediction method based on Bayesian estimation ARIMA parameters is proposed.Firstly,the Bayesian estimation method is used to estimate the unknown parameters of the ARIMA model,and then the cloud model theory is used to fit the prediction error probability density to obtain multiple normal concept clouds,Finally,the characteristics of different power segments are different and their corresponding optimal quantitation points are different.The particle swarm optimization algorithm is used to find the best quantile points for each power segment,so that the interval coverage is higher and the average is obtained.Wind power forecasting interval with narrower bandwidth.By comparing with different methods,the method has a superior interval prediction effect.Finally,a Bayesian neural network wind power interval prediction method based on fuzzy C-means clustering analysis is established,and a two-layer wind power interval prediction model is constructed.For the inner layer model,the training data is first classified by fuzzy C-means clustering,and then brought into the Bayesian neural network to train it;The outer model subtracts the predicted power output from the FCM-BNN model with the actual power to obtain a power prediction error sequence,and then fits the power prediction error sequence through non-parametric kernel density estimation to obtain the error sequence probability density function.The upper and lower quantiles satisfying the confidence requirement are selected on the prediction error probability density function,and the predicted power obtained by the FCM-BNN is combined to obtain the predicted power interval.