Load prediction method under extreme scene based on modal decomposition and transfer learning

Abstract

The invention discloses a load prediction method under extreme scenes based on modal decomposition and transfer learning. The method comprises the following steps of: 1) counting power load historicaldata in various extreme scenes, and classifying the power load historical data according to an approximate trend of a load curve; 2) performing frequency division on the data by using an improved aggregation empirical mode decomposition method to obtain a load trend term and a plurality of high-frequency components; 3) performing transfer learning based on attention mechanism model weight transfer by using the trend term of the historical data to obtain a prediction model of the trend term; and 4) respectively carrying out load prediction on the trend term and the intrinsic mode function by using a prediction model and an LSTM network, and superposing prediction results to obtain a load prediction result. According to the method, the power load historical data when various extreme eventsoccur can be fully utilized to predict the power load trend in the same kind of scenes, and the problem that a traditional method is not suitable for power load prediction in extreme scenes is solved.

Description

technical field [0001] The invention relates to the technical field of power system load forecasting, in particular to a load forecasting method in extreme scenarios based on modal decomposition and transfer learning. Background technique [0002] Power system load forecasting is an important means and key link for power system planning and guiding power production. However, uncontrollable factors such as earthquakes, mountain torrents, mudslides, typhoons, frost and other natural disasters or equipment failures in some extreme scenarios will lead to power loads. Mutation occurs. In this case, the historical data relative to the future load curve has no reference value, and it is impossible to calculate a more accurate forecast value through the traditional time series method. Contents of the invention [0003] In order to solve the shortcomings of the above-mentioned prior art, the present invention proposes a load prediction method for extreme scenarios based on modal d...

AI Technical Summary

Problems solved by technology

[0002] Power system load forecasting is an important means and key link for power system planning and guiding power production. However, uncontrollable factors such as earthquakes, mountain torrents, mudslides, typhoons, frost and other natural disasters or equipment failures in some extreme scenarios will lead to power loads. mutation

In this case, historical data has no reference value relative to the future load curve changes, and it is impossible to calculate more accurate forecast values ​​through traditional time series methods

Images