A power rewards system based on data fusion is proposed in this paper. In this system, a grey dynamic model is used to compensate the missing data by building a GM(1,1) forecast model. And a method of clustering fusion for multi-sensors data is introduced to improve the objectivity of data fusion, the method defines the Euclidean distance matrix and determines the orders of sensor fusion according to the minimum distance clustering. The standard value of the unit electricity consumption is obtained on the basis of data fusion, then the power consumption rewards model is used to give rewards or punishment to the corporation workshops. The application results indicate that the system is effective and reliable.