Short-term load prediction method, device and system

Abstract

The embodiment of the invention discloses a short-term load prediction method, device and system. The method includes the steps that pretreated historical load data and weather data are acquired; the historical load data is decomposed into multiple component data in different frequency bands by means of a wavelet packet decomposition method; single branch construction is conducted on each piece of the component data respectively by means of a wavelet packet reconstruction algorithm so as to obtain each first subsequence; the weather data is added into each first subsequence respectively, and second subsequences are obtained; the second subsequences are input to an extreme learning machine optimization model built in advance for prediction, and sub-prediction results corresponding to the second subsequences one by one are obtained; the sub-prediction results are overlaid, and a short-term load prediction result is obtained; the extreme learning machine optimization model is obtained by optimization of an extreme learning machine on the basis of training sample data and a particle swarm gravitational search hybrid algorithm. In the using process of the method, device and system, the global convergence accuracy is improved, so that the prediction result is more accurate.

Description

technical field [0001] Embodiments of the present invention relate to the technical field of power systems, and in particular to a short-term load forecasting method, device and system. Background technique [0002] In the power system, accurate short-term load forecasting is extremely important in arranging power purchase plan and operation plan. With the continuous development of social economy, the load is becoming more and more complex. Accurate load forecasting can ensure the stability and safety of the power system and improve the economic and social benefits of the power grid. [0003] At present, in the research of short-term load forecasting, traditional methods such as Fourier expansion method, time series method, multiple linear regression method and neural network forecasting method are used to predict short-term load, but due to the power load and weather conditions, Various factors such as holidays are closely related, and the load curve often changes randomly...

AI Technical Summary

Problems solved by technology

[0003] At present, in the research of short-term load forecasting, traditional methods such as Fourier expansion method, time series method, multiple linear regression method and neural network forecasting method are used to predict short-term load, but due to the power load and weather conditions, Various factors such as holidays are closely related, and the load curve often changes randomly. Traditional forecasting methods are difficult to predict load changes, and the global convergence accuracy is low, making it difficult for the forecast results to meet the current demand for load forecasting accuracy.

Images