The decomposed information of power consumption of household appliances is meaningful for scheduling the appliances and the reduction in home energy use. This paper proposes a novel load identification method based on quadratic 0-1 programming after a brief analysis on the residents load signature, and transforms the discrete model into continuous quadratic 0-1 programming model for overcoming the defects of complicated solving process. The power consumption of individual household could be decomposed into load level using the proposed algorithm. The recognition accuracy of load identification algorithms based on continuous quadratic 0-1 programming algorithm is tested, the results reveal that the algorithm could achieve high identification accuracy with appropriate load identification index, suitable for the NILM system of residential household.