The load decomposition is realized by the underdetermined solution of the current signal in non-intrusive acquisition mode. The complete current of each independent load is obtained

and state identification is realized on the basis of load decomposition. An optimization model is developed for the underdetermined issues difficult to solve by utilizing the load operation habit of residential users

thus transforming them into one-dimensional underdetermined issues

and the solving model is thus established just by separating two lines of signals from the signals collected in a unitary time interval. According to the sparsity of current frequency domain signals

the two-step iterative shrinkage threshold algorithm is used to get the optimal solution

so that each load put into operation can be separately decomposed. A priori approach is used to obtain the characteristic filter set for the characteristic current of various loads on the power network

enabling frequency domain filtering of the decomposed current

and the load identification is thus achieved by quantitatively determining the frequency components after filtering. The power consumption data acquired in the real world is used to prove that the algorithm is able to effectively achieve the load decomposition

and accurately recognize the status of loads.