Day-Ahead Forecast of Carbon Emission Factor Based on Long and Short-Term Memory Networks

Abstract

With the increase of climate change and the scarcity of fossil resources, it is urgent to reduce carbon emissions of customers. To encourage users to develop carbon reduction behaviors, a day-ahead carbon emission prediction model is developed for node carbon emission forecast of the regional grid. A long short-term memory network (LSTM) is introduced to learn historical series data of node load and carbon emission factors. Meanwhile, transmission line losses are also used as output indicators to better analyze carbon emissions and power quality in the day-ahead forecast. In the model, load variation on the load side, carbon emission factor on the generation side, loss on the transmission line, and predicted carbon emission factor on the network side nodes are considered. Finally, IEEE 9-bus system is implemented to analysis the 8-hour load fluctuation, carbon emission, and transmission power loss. The prediction experiment verifies the effectiveness and efficiency of the proposed method.