Network reconfiguration technology is one of the most important means to realize the self-healing control function of intelligent distribution network, and the key point of which is avoiding infeasible solution. This paper puts forward the concept of branch groups and analyse the defects of existing network reconfiguration algorithm according to the requirements of the distribution network and the characteristics of the harmony algorithm. The system topology is coded so that it can be effectively integrated with the intelligent optimization algorithm. Taking the harmony algorithm as an example, this paper gives the implementation steps of the algorithm and the four branch break rules. IEEE 33-node system example results show that this paper summarizes the situation of infeasible solutions into five kinds, which improves the search efficiency compared to the existing method. At the same time, it avoids missing viable solutions and ensures the spatial completeness of solutions. It effectively improves the speed of network reconfiguration and the search ability of global optimal solution.

Abstract. In this paper we present a novel method developed from multiple kernel function for short term load forecasting to integrate multi-source heterogeneous load factors in big data. Nine kinds of load factors were selected as multi-source heterogeneous factors. In the proposed method, three algorithms (the sample distribution method, single variable method and rank space diversity method) were adopted to establish the optimal multiple kernel functions to integrate the load factors. Experimental results show that the average relative error of multiple kernel SVM is smaller than single kernel SVM, and the accuracy of multiple kernel SVM model based on double layer multiple kernel learning algorithm and lp norm is the highest. Therefore, multiple kernel SVM can deal with the multi-source heterogeneous data in the load forecasting effectively, and the speed and accuracy of load forecasting can be improved by parallel processing.