Space-time combination prediction method based on CNN-LSTM and deep learning

Abstract

The invention discloses a time-space combination prediction method based on CNN-LSTM and deep learning, and the method comprises the steps: S1, carrying out the analysis of the time and space correlation of an original data set through an MI mutual information algorithm, eliminating the redundant information between data, and carrying out the dimension reduction of the original data; S2, establishing an MI-CNN-LSTM model, and predicting the data after dimension reduction, wherein CNN is used for extracting spatial information of each station, and LSTM is used for obtaining dependency information among time sequence data; S3, adding an AT layer between an LSTM layer and an output layer of the LSTM network to obtain an MI-CNN-ALSTM model, and performing model learning on training set data; S4, introducing a PSO algorithm to optimize the training parameters, and obtaining an MI-CNN-ALSTM-PSO model; S5, predicting the test set to obtain a final predicted value. According to the MI-CNN-ALSTM-PSO space-time combination prediction model provided by the invention, the wind power prediction precision and stability are further improved.

Description

technical field [0001] The invention relates to the technical field of wind power forecasting, in particular to a combined spatiotemporal forecasting method based on CNN-LSTM and deep learning. Background technique [0002] In the past few decades, the global wind power installed capacity has grown rapidly. At present, wind energy, as a clean and green renewable energy, is widely used in real life. Therefore, the wind power prediction of wind farms can provide an effective reference for urban construction, power transmission and circuit construction. [0003] At present, wind power forecasting models can be divided into three categories: time series models, machine learning and combination models. Time series models include autoregressive model (Autoregressive Model, AR), moving average model (Moving Average Model, MA), autoregressive moving average model (Autoregressive Moving Average Model, ARMA), autoregressive integrated moving average model (Autoregressive Integrated ...

AI Technical Summary

Problems solved by technology

[0004] However, the above methods are mainly used in the process of wind speed or wind power forecasting, using signal decomposition, weight distribution, deep learning, regression model and parameter optimization and other methods for research, mainly based on the front and rear dependencies of time series for data forecasting, and most wind power wind speed forecasting The method only uses the characteristic data of a single site, does not consider the potential spatio-temporal connection of the characteristics of the adjacent sites of the wind farm, and ignores its spatial feature information, which limits the prediction accuracy.

Images