Short-period load predicating model based on temperature accumulation effect and gray correlation degree

Abstract

The invention discloses a short-period load predicating model based on a temperature accumulation effect and a gray correlation degree. The method comprises the steps of 1, selecting meteorological indexes with relatively high correlation degree; 2, considering a highest temperature of a previous day of a to-be-predicated day and average temperature data in a predicating process; 3, selecting a data sample; 4, establishing an LSSVM predication model; and 5, performing load predicating by means of trained model. The short-period load predicating model can well improve similar day selection precision and load predicating precision. The short-period load predicating model has advantages of optimizing a model training process, realizing consideration of the temperature accumulation effect, effectively tracking influence of air temperature change to the load according to a method of selecting the similar day according to the gray correlation degree, reducing time length in training, remarkably improving load predicating precision, and verifying feasibility of the presented predicating model.

Description

technical field [0001] The invention relates to the field of power system load forecasting, in particular to a short-term load forecasting model based on air temperature cumulative effect and gray relational degree. Background technique [0002] Power system load forecasting is a scientific method to predict future load based on forecasting models by using historical load data, meteorological data, and daily type data. Short-term load forecasting is related to the safety and stability of power grid operation, and has important reference value for the formulation of dispatching plan and power generation plan. [0003] Meteorological factors are the key influencing factors in short-term load forecasting. The Lincang power grid in Yunnan Province to be studied in this paper is located in an area with changeable weather and more rainfall, which leads to large temperature fluctuations in this area. Therefore, meteorological factors must be considered when predicting the load of ...

AI Technical Summary

Problems solved by technology

Literature Liu Wenying, Men Deyue, Liang Jifeng, etc. "Monthly Load Forecasting Based on the Combination of Gray Relational Degree and LSSVM". The training samples are input into the LSSVM model for load forecasting, but meteorological factors and temperature accumulation effects are not considered; the literature Gong Wenlong "Short-term Load Forecasting Based on Least Squares Support Vector Machine", Hunan "Hunan University", 2014, proposed a particle swarm optimization algorithm The LSSVM model optimizes the key parameters of the model, but only performs simple processing and normalization on the training sample data, without considering similar load days, so it has the disadvantages of large data dimension and long training time

Images