Abstract: In view of the large dynamic estimation error of loss-reduction effect of smart grid, a smart grid loss-reduction estimation model is proposed in this paper based on big data analysis, which applies the Map/Reduce model to process the big data of smart grid. According to the statistic and analytical results, the smart grid can be divided into different subnets, and different load forecasting models are applied to predict the loads of subnets. The estimated overall load of smart grid is corrected through fussing the predicted loads of subnets. By inputting the load predictions into the smart grid loss probability evaluation model, the smart grid loss-reduction analysis can be realized under the restriction of both single and combined factors, and a dynamic assessment can be made on the loss-reduction effect of the smart grid. The experimental results show that the proposed model is very effective for estimating the loads of smart grid on both weekdays and playdays.

Key words: big data, smart grid, load prediction, loss reduction