In order to further enhance the adaptive capacity of the grey forecasting model to original data, a GM(1, 1) power model with time-varying structure parameters is proposed. Polynomial functions are introduced to describe the dynamic changes of the structure parameters with time. According to the different modeling sample sizes, the identification formulas for unknown parameters are given under three kinds of situations. At the same time, the analytical solution to white differential equation of GM(1, 1) power model with time-varying parameters is proposed. The integral compound trapezoid formula is employed to transform the continuous solution into discrete time response that can be used to predict. An application example shows that the GM(1, 1) power model with time varying parameters has higher simulation and forecast precision than that of the GM(1, 1) power model with fixed parameters.