Abstract:

Smart community construction in the smart power consumption link of the smart grid can bring real time interaction between the grid and end-users, improve demand response performance, enhance user energy efficiency management, and realize the load peak shaving. A new approach was proposed to recognize the resident user type in the smart community based on the support vector machine (SVM) classification model, some interesting features were discussed, which included the power consumption rate in the peak load period, load rate, user cooperation degree and so on. Experiment data were collected from the users of the smart community. Experimental results show that SVM is effective for the power resident user type.

Key words: business intelligence, support vector machine, feature selection