Combined Interval Prediction Method Based on Normal Exponential Smoothing and Kernel Density Estimation

Abstract

The invention discloses a combined interval prediction method based on a normal exponential smoothing method and kernel density estimation. The wind power interval prediction method comprises the steps of: introducing the exponential smoothing method into normal distribution estimation, wherein the weight of the old data is exponentially decayed with time when the wind power prediction error distribution at time t+1 is estimated so that the calculation result is more accurate; introducing the ideas of ''mixing'' and ''sliding'' into the kernel density estimation, when the wind power predictionerror distribution is estimated at the time t+1, estimating the wind power prediction error probability density function during the time before the time t+1 by different bandwidths, and then offsetting the estimation error by weighted combination of different bandwidth estimated probability density functions; and generating the final wind power prediction interval by the prediction interval obtained by estimation of reasonable weighted combination of the normal exponential smoothing method by an entropy weight method and the prediction interval obtained by mixing the sliding kernel density estimation, so that the two methods complement each other to a certain extent.

Description

technical field [0001] The invention relates to the technical field of wind power, in particular to a wind power interval prediction method based on a normal exponential smoothing method and a hybrid sliding kernel density estimation combination. Background technique [0002] With the limitation of conventional energy and the increasingly prominent environmental problems, the new energy with the characteristics of environmental protection and renewable has been paid more and more attention by the governments of various countries. As a green renewable energy, wind energy has been widely used in countries all over the world. Due to the randomness and instability of wind, it brings great challenges to the safe and stable operation of power grids. Accurate and effective wind power forecasting will help the power dispatching department to adjust the dispatching plan in time, reduce the risk of wind power being integrated into the grid, reduce the reserve capacity of the system, a...

AI Technical Summary

Problems solved by technology

[0004] There are usually two types of methods to estimate the probability distribution of prediction errors: parametric and non-parametric methods. Commonly used parameter methods include normal distribution, Beta distribution, etc. The parameter method is simple and intuitive to estimate, but the distribution form and parameter selection of the prediction error are sometimes inaccurate

Non-parametric methods usually do not require a priori assumptions on the distribution of prediction errors, and the probability of each point is determined by real data. Kernel density estimation is a common non-parametric estimation method, but the selection of bandwidth parameters has a great impact on the accuracy of kernel density estimation. Accuracy has a big impact

Images