This paper proposes a novel non-intrusive load monitoring (NILM) method which incorporates appliance usage patterns (AUP) to improve performance of active load identification and forecasting. In the first stage, the AUP of a given residence were learned using a spectral decomposition based standard NILM algorithm. Then, learnt AUP were utilized to bias the priori probabilities of the appliances through a specifically constructed fuzzy system. The AUP contain likelihood measures for each appliance to be active at the present instant based on the recent activity/inactivity of appliances and the time of day. Hence, the priori probabilities determined through the AUP increase the active load identification accuracy of the NILM algorithm. Therefore, the prior probability determined by AUP increases the payload identification accuracy of NILM algorithm. Subsequently, the proposed method is applied to several groups of real home databases, which proves its ability to improve active load estimation accuracy. In addition, the forecasting mechanism of residential electricity consumption is successfully developed and implemented by using the proposed method. The experimental results show the effectiveness of the proposed method.