Abstract: In order to apply the non-instrusive load montiring technology into the new generation of smart

meter，a kind of method based on the BP neural networks enhancement algorithm of clustering for load

identification was presented. In this method，each feature was relatived to the corresponding cluster based on the BP neural network. Then，under the framework of Adaboost algorihtm，the cluster was cascaded each other with aiming to built the strong classificaiton method. Considering that the weight of cluster is associated with the BP neural network，it is necessary to ensure the goodness of BP neural networks during its classification，so the mind

evoluationary algorithm was applied into adjusting the parameter of BP neural network. Finally，the weight of

clustering mehtod was optimaized throught the iteration of Adaboost framework，thus building the strong

classification method. The experiemnts on the AMPds dataset were carried out for demonstrating the performance

of proposed mehtod. Meanwhile，the comparative analysis of testing show that our method has desired result on

non-intrusive load identification.

      Key words: smart meter；non-instrusive；mind evoluationary algorithm（MEA）；Adaboost algorithm