Abstract:Abstract To improve the forecasting accuracy of power load, the forecasting model based on sparrow search algorithm (SSA), variational mode decomposition (VMD), attention mechanism and long short-term memory (LSTM) was proposed. Firstly, SSA is used to optimize the number of decomposition and penalty factor in VMD and realize the decomposition operation of the initial data. Then, LSTM is used to predict each component, and on this basis, feature and temporal attention mechanisms are introduced. Feature attention mechanism is introduced to calculate the contribution rate of relevant input features in real time, and the feature weights are modified to avoid the limitations of traditional methods relying on the threshold of expert experience association rules. Temporal attention mechanism is applied to extract the historical key moments and improve the stability of the time series prediction effect. Finally, the final result is obtained by superimposing the prediction results of each component to complete the power load prediction. Practical examples show that, compared with other methods, the proposed model achieves the highest prediction accuracy, with an RMSE of 1.23, MAE of 0.99 and MAPE of 11.62%.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is to improve the accuracy of power load forecasting. Specifically, the article proposes a prediction model based on the Sparrow Search Algorithm (SSA), Variational Mode Decomposition (VMD), attention mechanism and Long - Short - Term Memory network (LSTM), aiming to overcome the deficiencies in traditional methods. The following are the main problems in this research and their solutions: 1. **Problem of traditional VMD parameter optimization**: - Traditional VMD cannot adaptively determine the number of decomposition layers and the penalty factor, resulting in under - decomposition or over - decomposition of the signal. - **Solution**: Use the Sparrow Search Algorithm (SSA) to optimize VMD parameters. By using envelope entropy as the fitness function, find the optimal number of decomposition layers and the penalty factor. 2. **Problem of feature correlation mining**: - In the LSTM prediction process, the correlation between power load and input features is not fully considered. - **Solution**: Introduce the feature attention mechanism, calculate the contribution rate of each input feature to the power load in real - time, and dynamically adjust the feature weights to avoid the limitations of relying on expert - experience - based association rules. 3. **Problem of time - series dependence**: - Setting an overly long time - series length in the LSTM training process may affect its stability, and the correlation between the current power load and historical moment information is not fully mined. - **Solution**: Introduce the time attention mechanism to extract key historical moment information and improve the stability of time - series prediction. 4. **Problem of non - stationary data processing**: - The power load time - series has the characteristics of non - stationarity and non - linearity, and a single model is difficult to achieve a satisfactory prediction effect. - **Solution**: Decompose the original load data into different frequency components through VMD, and then combine models such as LSTM for prediction, effectively weakening non - linearity and improving prediction accuracy. Finally, this model completes the power load prediction by superimposing the prediction results of each decomposed component. The experimental results show that, compared with other methods, this model achieves the highest prediction accuracy, with RMSE of 1.23, MAE of 0.99, and MAPE of 11.62%.

Short-term power load forecasting based on sparrow search algorithm-variational mode decomposition and attention-long short-term memory

Power Load Prediction Model Based on Long Short Term Memory and Sparrow Search Algorithm

Short-Term Power Load Forecasting Method Based on Improved Sparrow Search Algorithm, Variational Mode Decomposition, and Bidirectional Long Short-Term Memory Neural Network

Load Forecasting of Sparrow Search Algorithm Optimization Double BIGRU

Research on Optimization of Improved Short-Term Load Composite Forecasting Model Based on AM–CNN–Bi–LSTM

Short-Term Power Load Forecasting Model Based on t-SNE Dimension Reduction Visualization Analysis, VMD and LSSVM Improved with Chaotic Sparrow Search Algorithm Optimization

Research on short-term power load forecasting method based on multi-factor feature analysis and LSTM

Study on Short-Term Electricity Load Forecasting Based on the Modified Simplex Approach Sparrow Search Algorithm Mixed with a Bidirectional Long- and Short-Term Memory Network

A Combined Model Incorporating Improved SSA and LSTM Algorithms for Short-Term Load Forecasting

Short term power load forecasting based on BES-VMD and CNN-Bi-LSTM method with error correction

Daily Peak-Valley Electric-Load Forecasting Based on an SSA-LSTM-RF Algorithm

A Short-Term Power Load Forecasting Method Based on SBOA–SVMD-TCN–BiLSTM

Research on Short-Term Load Forecasting Method of Power System Based on Seq2Seq-Attention Model

A Short-term Power Load Forecasting Method Based on An Improved VMD-LSTM Algorithm

Demand-Side Electricity Load Forecasting Based on Time-Series Decomposition Combined with Kernel Extreme Learning Machine Improved by Sparrow Algorithm

A power load forecasting model based on a combined neural network

Research on short-term power load forecasting based on VMD and GRU

Short-Term Power Load Forecasting Based on Secondary Cleaning and CNN-BILSTM-Attention

Research on PCA-LSTM-based Short-term Load Forecasting Method

Long Short-Term Memory for Short Term Load Forecasting with Singular Spectrum Analysis and Whale Optimization Algorithm

Multi-Energy Coupling Load Forecasting in Integrated Energy System with Improved Variational Mode Decomposition-Temporal Convolutional Network-Bidirectional Long Short-Term Memory Model