In order to overcome the impact of photovoltaic power fluctuations on the stability of the power system

a capacity configuration strategy using batteries and supercapacitors as complementary energy storage is proposed. Firstly

a time window is constructed based on the national requirements for grid connected fluctuations at two scales

and it is made to follow the characteristics of photovoltaic output. Real time detection of power off-limits data is performed

and the off-limits window is corrected based on the limit requirements to ensure smooth output of photovoltaic output to the grid. Secondly

a non-local mean filter is proposed

and a multi-strategy improved electric eel foraging algorithm is utilized to optimize its initial parameters

thereby achieving optimal frequency band decoupling of the complementary energy storage power output and ultimately obtaining the energy storage capacity configuration. The results indicate that the off-limits correction strategy can minimize energy storage output redundancy and significantly reduce component losses. At the same time

the non-local means (NLM) filter constructed by parameter optimization with minimizing cost as the objective function obtains the optimal decoupling instruction that fully matches the response frequency band of complementary energy storage components. The capacity configuration under this strategy combines component sustainability and economic benefits. The improved electric eel foraging optimization algorithm adopted in the strategy also has good optimization performance. The research paper can provide economic and effective technical solutions for the absorption of clean energy and the improvement of grid connection quality.