Resident load prediction method based on LSTM-SAM model and pooling

Abstract

The invention discloses a resident load prediction method based on an LSTM-SAM model and pooling, and belongs to the technical field of power systems, and the method comprises the steps: obtaining historical load data and numerical weather forecast data of a plurality of resident users, and randomly selecting a certain user as a target user; preprocessing the data of each user by adopting two-stage feature engineering; sorting the non-target users, selecting different numbers of non-target users as interconnected users, forming different pooling combinations together with the target users, constructing a training data pool based on pooling, and retaining test set data of the target users; and inputting the training data pool and the test set data of the target user into the LSTM-SAM hybrid model, obtaining the predicted value of each load component, adding the predicted values, and outputting the day-ahead load prediction result of the target user at the to-be-predicted moment and the optimal number of pooling users. According to the method, the prediction precision of the resident load is improved, guidance is provided for system scheduling and demand response implementation, and safe, stable and economical operation of a power system is guaranteed.

Description

technical field [0001] The invention belongs to the technical field of power systems and relates to a residential load forecasting method based on LSTM-SAM models and pooling. Background technique [0002] The power system needs to maintain a dynamic balance between power supply and power demand, and load forecasting has very important practical significance in maintaining the stable operation of the power system and guiding power dispatching. Due to the influence of external factors, the power load has certain fluctuations and uncertainties. Compared with the total system load, the user-level residential load is more difficult to predict due to the lack of load smoothing. In addition, power users voluntarily participate in demand response, which makes residents' load forecasting more complicated. Therefore, it is very important to study accurate residents' load forecasting methods. [0003] Existing load forecasting methods can be divided into traditional statistical met...

AI Technical Summary

Problems solved by technology

[0004] Purpose of the invention: The present invention aims at the deficiencies of current user-level resident load forecasting methods, including problems such as low forecasting accuracy and failure to utilize useful information other than target user data, and proposes a resident load forecasting method based on LSTM-SAM model and pooling, based on LSTM-SAM hybrid model, learning the hidden information in the training data pool, further improving the prediction accuracy

Images