Household load identification is one of the key techniques in demand side management, which is helpful to realize intelligent power utilization. Combined with the basic principle and method of system identification, each appliance load is seen as an independent system, and a new method of appliance load identification based on system model is proposed in this paper, which is characterized by steady-state current and steady-state voltage. The steady state data of each load in the power network is collected via priori approach in order to construct the ARMAX linear model library and the Hammerstein nonlinear model library. Computing the characteristic value of steady current crest, which can prescreen the load to identify which library it belongs to. Then the load identification is carried out by the model matching principle. The effectiveness of the algorithm is verified by the actual sampling load data, which can accurately determine the load state for both linear loads and nonlinear loads. Furthermore, the algorithm is efficient and can effectively cope with the situation when the new load joins the home network.