In recent years a large number of photovoltaic power generations were applied in power distribution network, but the photovoltaic power station has the random output, leading to the inaccurate power flow optimization results. In order to be able to optimize the power flow of distribution network considering photovoltaic power generations with the change in load, in the paper we will divide the photovoltaic power and changeable load into 24 periods, use the genetic algorithm(GA) calculating the static reactive power optimization in the each period for getting power loss,utilize dynamic programming(DP) processing the ability of solving the problem of time sequence optimization, consider the minimum of power loss summation each period as the objective function, achieve the action time table of the control equipment. In order to avoid complex calculations, simplify the operation,ignoring the invalid search path but also ensuring voltage stability. The method is simple and effective model, satisfies the constraints the numbers of capacitors actions under the conditions of reducing the power loss, increases the life of the capacitors. Through numerical results show that the proposed method is feasible and effective operation, achieved the intended purpose.