Appliance load identification is an important part of intelligent power consumption. Traditional intrusion load monitoring has the drawbacks of high cost

complex installation and maintenance

hence the need of a load identification algorithm based on non-intrusive load monitoring. According to the principles and method of system identification

a load identification algorithm based on local mean decomposition(LMD)and model matching is proposed

which is characterized by steady-state current and voltage. In order to construct the linear and nonlinear model libraries

the steady-state data of each load in the power network is collected in advance. Then

the LMD algorithm is used to decompose the mixed signals into electricity consumption data of single load. By pre-screening

these separated data are categorized into the model library that they belong to

and finally these loads are recognized according to the model matching principles. Simulation results show that the proposed algorithm can accurately identify the operating status of each load and has high computational efficiency. Furthermore

it can effectively deal with the situation when a new load joins the power network.