Analysis and evaluation of short-term wind power interval forecast error based on K-means clustering

Abstract

The intermittency, volatility and randomness of wind power will inevitably affect the accuracy of wind power prediction. In order to improve the accuracy of wind power interval prediction and evaluation, a method of wind power interval prediction and evaluation based on K-means clustering is proposed. First, use K-means clustering to classify wind power prediction error data; second, combine the climate type to more accurately divide and process historical data, and apply multiple linear regression models to establish preliminary prediction models corresponding to each subset; Finally, according to the distribution characteristics of forecast errors, the results of wind power forecasting are analyzed and evaluated with interval coverage and interval average bandwidth as indicators. Comparing the K-means clustering algorithm with the BP neural network algorithm, the results show that the method proposed in this paper can better capture the characteristics of wind power error data, and can obtain more accurate wind power interval prediction and evaluation results.