Short-Term Load Forecasting Model of Electric Vehicle Charging Load Based on MCCNN-TCN

The large fluctuations in charging loads of electric vehicles (EVs) make short-term forecasting challenging. In order to improve the load performance EV load, a corresponding model-based multi-channel convolutional neural network and temporal (MCCNN-TCN) are proposed. (MCCNN) can extract fluctuation characteristics at various time scales, while (TCN) build time-series dependence between forecasted load. addition, an additional BP maps selected meteorological date features into high-dimensional feature vector, which is spliced with output TCN. According experimental results employing urban station data from city northern China, proposed model more accurate than artificial (ANN), long memory (LSTM), networks (CNN-LSTM), TCN models. MCCNN-TCN outperforms ANN, LSTM, CNN-LSTM, by 14.09%, 25.13%, 27.32%, 4.48%, respectively, terms mean absolute percentage error.