With a large amount of distributed energy accessing to the distribution network, the randomness of power fluctuations has a serious impact on the new energy consumption and equipment utilization. In this paper, the adjustable characteristics of the load side power are fully utilized, by adjusting the feeder voltage, the distributed energy output can be tracked by the load, thereby the suppression of the power at the point of common coupling (PCC) is achieved. In order to ensure the smooth suppression of fluctuations on the basis of economic operation of distribution network and to eliminate the influence of randomness of distributed energy on control, this paper establishes a source-load tracking model based on a framework of intraday optimization and real-time control. At the intraday optimization level, the optimization model is established with the adjacent period PCC power difference and operation cost as the objective function to solve the optimal PCC power. At the real-time control level, the voltage regulation is solved by rolling optimization based on the model predictive control algorithm and the results of the intraday optimization, so as to realize the source-load tracking. Finally, a case of the measured data from an industrial park in Guangzhou verifies the effectiveness of the model in suppressing the power fluctuation.