Abstract: To cope with the difficulty of accommodation during complementary operation of multi-renewable energy sources and mitigate the frequent voltage out-of-limit caused by the fluctuation of renewable energy sources in the night, a predictive control based dynamic dispatching model for the power system containing complementary multi-renewable energy sources was proposed. Firstly, the timeliness of both photovoltaic (abbr. PV) power generation and wind power generation as well as the feature of slightly fluctuation of chain relative ratio of household demand and production demand of electricity in a short term were analyzed, and the wavelet filtering method with improved adaptive threshold membership was utilized to denoise the historical data of PV generation and wind generation to restore the data authenticity. Secondly, the denoised data was input into the improved autoregressive integrated moving average model (abbr. ARIMA) to predict the generated energy and electricity consumption in a future time period, and taking the predicted generated energy and electricity consumption as the decomposition function and the augmented function the synchronous alternating direction method of multipliers (abbr. SADMM) was improved. Finally, using the predicted generated energy of renewable energy sources in all time intervals as the fixed value of next iterative computation, the iterated value of SADMM was modified to achieve the purpose of dynamic economic dispatching. Experimental results show that using the proposed model the times of alarm of voltage out-of-limit is reduced by 91%, besides, the economy of the power system is improved by 5% than traditional method.