If a load device is put into operation

the current obtained when the load is running separately will be integrally included in the mixed sampling current. Considering the current model of non-intrusive load monitoring(NILM)

a frequency analysis and template filtering-based identification algorithm is proposed. The characteristic current of each load is collected via a priori approach in order to form the characteristic filter. Giving 0 or 1 value to the spectral components of the characteristic filter

the corresponding template filter can be formed

which is used to filter the mixed sampling current signal. The running state of the load can be determined by quantitative decision of the filtered frequency component. The algorithm is verified effective by the actual sampling load data

which can accurately determine the load state. Furthermore

being realized by fast Fourier transform(FFT)

the algorithm is easy to implement in hardware. This work is supported by National High Technology Research and Development Program of China(863 Program)(No. 2015AA050203)and Fundamental Research Funds for the Central Universities(No. 2016MS13).