For solving the common existence of feeder overload problem, a load balance technology based on multi-source data fusion is proposed. Firstly, a feeder overload quantitative model is proposed according to the dynamic characteristics of distribution load. Then, distribution load sets are constructed for the three typical application scenerios based on multi-source data of dispatching, operation and maintenance distribution and marketing. Finally, the optimal feeder load balance adjustment strategy is determined through trail calculation and comparison of the overload indexs of multiple distribution reconstruct schemes. Example shows the effectiveness of this technology. Follow this technology, a distribution load management and optimization assistant decision-making system is developed, which has been practically applied to Jiangsu distribution dispatching, operation and maintenance.