An Improved Non-invasive Multi-label Load Disaggregation Method Based on Appliance Behavior Correlation Graph
As the main means of power consumption information monitoring,non-invasive load disaggregation technology is of great significance to the improvement of energy efficiency and demand-side optimization management.In view of the fact that the current load disaggregation model relies too much on the power consumption characteristics of electrical appliances and ignores the information provided by users'electricity consumption habits,it is difficult to improve the disaggregation effect.In this paper,an improved method of multi-label energy disaggregation considering users'electrical behavior is proposed.The improved model is a serial architecture of two networks.The first network combines the electricity consumption behavior of users to realize multi-label type recognition.The second network completes the energy disaggregation of each online appliance based on the recognition results.In this paper,the user's electricity consumption habits are represented by the appliance behavior correlation graph.The model continuously updates the behavior with the electricity consumption of the users,and gradually generates a unique network graph for the user,which provides the behavior basis for the load disaggregation.Finally,the open dataset REDD and REFIT are used to simulate and evaluate the method.The experimental results show that the proposed method can accurately obtain the power consumption information of each electrical appliance,and has obvious improvement compared with the existing advanced methods.It is proved that the multi-label method considering the power consumption behavior of users is an effective and feasible idea of load disaggregation.
国家科技期刊平台
NSTL
万方数据
维普
