Non-invasive Load Identification Method Based on the Characteristics of Residential Electrical Appliances

Aiming at the non-intrusive load monitoring technology supported by low frequency data, a non-intrusive load identification method with the characteristics of residential electrical appliances is proposed, which improves the practicability of the current identification algorithm in the intelligent electricity meter installed on a large scale in Jiangsu Province. Select the active power of resident electrical appliances as a steady-state electrical characteristic, introduce the behavior characteristics such as the running time of electrical appliances, the number of switches and so on, detect load switch events by di-pushing SVD decomposition method, and extract the characteristic data to establish the load identification neural network training database, through THE neural network learning and training, form a non-invasive load BP neural network identification model; The results show that the method has good identification accuracy and practicability without introducing current characteristics.