A Non-intrusive Carbon Emission Accounting Method for Household in the Power System

The escalating environmental issues resulting from excessive greenhouse gas emissions have become increasingly severe, with the proportion of carbon emissions from households on the rise. Current carbon emission accounting methods can’t account for household carbon emission in a refined and timely manner. Therefore, a non-intrusive appliance household carbon emission accounting (NACEA) method is proposed to achieve the detail of household carbon behaviors. First, convolutional neural network (CNN) based sequence-to-sequence (seq2seq) is used to realize household load decomposition. Then, household carbon emission is accounted based on dynamic carbon emission factors and parameter fine tune is used for accounting error compensation. Finally, the load decomposition and NACEA results are validated on the REFIT dataset. The proposed method can accurately realize NACEA.