Electric vehicle load prediction based on modal decomposition CNN-BiLSTM model

With the rapid rise of the electric vehicle industry, the huge and volatile electricity demand generated by it poses a significant challenge to the security and stability of the grid. In view of the nonlinearity and characteristic fuzziness of electric vehicle load data, this study innovatively proposes a load forecasting strategy for electric vehicles that integrates mode decomposition technology. Firstly, Pearson correlation coefficient is used to efficiently screen load data to eliminate redundant information. Then, the adaptive noise complete set empirical mode decomposition (CEEMDAN) method is introduced to finely divide the complex load sequence into high-frequency and low-frequency components, aiming to capture the load variation characteristics at different time scales. At the level of model construction, the decomposed load components are taken as inputs, and a composite prediction model is constructed by combining the advantages of convolutional neural network (CNN) and BiLSTM. The model can make full use of the high efficiency of CNN in feature extraction and the powerful ability of BiLSTM in processing time series data. In order to further improve the performance of the model, the improved Dung Beetle Optimization Algorithm (IDBO) was also adopted in this study to perform intelligent optimization of key hyperparameters of the model, which significantly improved the prediction accuracy and optimized the computational efficiency. The experimental results show that the prediction method proposed in this study not only shows significant advantages in the prediction speed, but also reaches a high level in the prediction accuracy.