The Method of Household Adjustable Electricity Prediction Based on Maximum Mean Discrepancy

Abstract As electricity market reforms continue and households consume more electricity, Household Electricity Response (HER) becomes an essential part of Demand (DR). The ability to adjust their demand is a critical factor in evaluating the demand-side response capacity power systems. Therefore, prediction method household adjustable (HAE) based on cross-domain generalization proposed solve problem. In our method, convolutional neural networks (CNNs) model with Maximum Mean Discrepancy (MMD) loss term constructed realize goal, which improves accuracy HAE different regions provides technical support for DR.