Active integration of regional integrated energy side data based on multi-scale channel fusion learning

Regional integrated energy systems involve multiple energy subsystems, such as power, heating, cooling, etc., and the data types and formats of each subsystem may be different, resulting in data heterogeneity. It is difficult to integrate heterogeneous data. Because multi-scale channel fusion learning can simultaneously process data of different data types and resolutions, and retain key information at different scales. Therefore, this paper proposes an active integration technology of regional integrated energy data based on multi-scale channel fusion learning. The comprehensive energy label matrix is constructed, and the weight of energy data is calculated to realize the collection of regional comprehensive energy data. A multi-scale channel fusion learning classifier is designed to classify energy side data. Based on this, SOA architecture is used to construct the overall architecture of regional integrated energy side data active integration technology, which includes data source service layer and application service layer. The experimental results show that: for different energy data sets, the research technology can complete the active integration of regional integrated energy data with high precision, and the integration rate can reach $95 \%$.