When power load forecasting grows quick, the traditional gray prediction model GM(1,1) becomes worse. According to the shortcoming, in this article a new improved Nonlinear regression forecast model based on Grey Model is proposed, combining the advantage about Nonlinear Regression with GM(1,1) Model, using GM(1,1) Model to calculate the initial values about the parameter and then analysis it by Nonlinear Regression to forecast the power load value. The power load forecast example indicates that the model gives better precision and has wider application field.