Extraction and Analysis of Heat Storage Load Characteristics for Power Quality Improvement

Abstract

With the rapid increase of grid-connected capacity of wind turbines in China, wind power consumption is challenged. Especially, the three northern areas' wind curtailment is up to 95% of the whole country. In order to improve the capacity of wind power consumption, our country encourage the technology of clean energy heating. Electrical heating with heat storage is one of the most popular methods of clean energy heating. However, the analysis of the heat storage load is not adequate by the method, because of the huge power grid data, poor data quality and the inaccurate extraction of heat storage load. Aiming the problems, a method based on K-means, K-nearest neighbor algorithm (KNN), and support vector machine algorithm (SVM) is proposed for characteristic extraction of heat storage load in power system. K-means is used to cluster data and extract heat storage load as the training cluster. KNN and SVM are used to train the training cluster and extract the heat storage load. Take the heat storage load in a certain area in Liaoning province as an example. Compared with the traditional methods, the extraction method is more accurate. Then the extracted heat storage data are classified and analyzed. It is of great practical significance to improve the consumption of wind power and improve power quality reasonably.