Power load forecasting based on GOM(1,1) model under the condition of missing data

In the actual power load forecasting, there are often missing data in the original data due to many subjective and objective factors. GOM(1,1) model can't be used to predict based on the equidistant sequence data directly. In this paper, it is supposed that the missing data is the objective existence. Minimizing relative error is taken as the objective function. The problem of GOM(1,1) modeling under the condition of missing data is transformed into the problem of solving parameters and based on nonlinear programming with constraints. Through the example analysis, the forecasting result of this method in this paper is superior to GM (1,1) model and GOM (1,1) model based on traditional interpolation method.