A TCN-BiGRU-based multi-energy consumption evaluation approach for integrated energy system

Abstract

The increasing demand of energy consumption reduction makes evaluating multi-load a growing and notable challenge. However, if the accuracy of energy evaluation in integrated energy systems is not high enough, the subsequent index evaluation, economic analysis and scheduling cannot be carried out effectively. In this paper, a Bidirectional Gated Recurrent Unit model adapted for multi-load consumption evaluation is proposed, and combines the inputs of evaluation model with Temporal Convolutional Network, allowing the model to obtain a larger receptive field. The resulting evaluation is shown to be applicable to longer series, and more accurate than those generated with other neural networks. In a system multi-energy evaluation exercise combined with local meteorological data, the proposed model is shown to estimate the energy consumption of the system relatively accurately, which can assist in further analysis.

Keywords

• Integrated energy systems
• Energy consumption evaluation
• Multi-feature mining
• TCN-BiGRU