Abstract:Demand response is significant in alleviating power supply shortage, promoting new energy consumption and saving resource investment on the generation side, while demand response capability assessment can provide a basis for the formulation of demand response strategies. In order to solve the problem of difficult demand response capacity assessment, this paper proposes a method for regional residential response capacity assessment based on non-intrusive load decomposition. First, a non-intrusive load decomposition model based on random forest algorithm is established, and then the interruptible load of individual customers is decomposed by this model, and then the responsiveness of individual customers at each moment is calculated based on the interruptible load usage, and finally the total responsiveness of area residents is obtained by aggregating all customers in the area. The results show that the proposed random forest-based non-intrusive load decomposition model can obtain more accurate active power values of interruptible loads of customers, and the regional active power aggregation error is also smaller, which can be better for demand responsiveness assessment.