Accurate Ultra-Short-term Load Forecasting Based on Load Characteristic Decomposition and Convolutional Neural Network with Bidirectional Long Short-Term Memory Model
Mingyue Zhang,Yang Han,Amr S. Zalhaf,Chaoyang Wang,Ping Yang,Congling Wang,Siyu Zhou,Tianlong Xiong
DOI: https://doi.org/10.1016/j.segan.2023.101129
2023-01-01
Sustainable Energy, Grids and Networks
Abstract:Aiming at the continuous, periodic, and nonlinear characteristics of load changes, this paper proposes a combined ultra-short-term load forecasting model based on improved complete ensemble empirical mode decomposition with adaptive noise (ICEEMDAN), convolutional neural network (CNN) and bidirectional long short-term memory (BiLSTM) neural network. Firstly, the ICEEMDAN algorithm is utilized to decompose the non-stationary load sequence into different intrinsic mode function (IMF) components, thus reducing the complexity and non-stationarity of the original load sequence. Subsequently, the power load series is divided into trend term, periodic term, and random term subsequences according to the analysis principle of time series. Also, the correlation between the influencing factors and the load is analyzed using grey relational grade (GRG), and the influencing factors with higher correlation are screened to construct different feature sets of load influencing factors. Finally, the historical load data and related impact factor data sets are input to CNN for feature depth extraction. The feature vectors are then constructed in a time series sequence and used as the input data to the BiLSTM network for multi-data-driven dynamic prediction. To verify the effectiveness of the proposed prediction model, the user-side load data of a micro-grid community system in Sichuan Province, China, is used as an example in this paper. The simulation analysis results demonstrate that the proposed prediction model effectively captures the ultra-short-term dependence between data. Compared with the support vector machine (SVM) model and autoregressive integrated moving average (ARIMA) model, the CNN-BiLSTM model exhibits higher prediction accuracy, with a MAPE reduced by 3.547% and 4.222%, respectively. Additionally, the RMSE was reduced by 0.421 kW and 0.554 kW, respectively. In conclusion, this model not only achieves high prediction accuracy but also demonstrates good stability. Therefore, it can provide a theoretical basis and technical support for accurately predicting actual power load.