A Multi-modal Time Series Intelligent Prediction Model

Abstract The power load prediction can ensure the supply and dispatch, which will be useful for market participants to plan make strategic decisions enhance reliability, save operation maintenance costs. Short-term data series have obvious approximate periodicity, while long-term show variability dynamic features. In addition, time of various modalities, such as reports production management data, could play a role in prediction. One kind multi-modal CNN-BiLSTM architecture is proposed predict short-term an improved shared parameter convolutional network learn feature representation attention-based BiLSTM mechanism, model features multimodal on data. Experimental results dataset that, compared with other baseline systems, this has some advantages accuracy.