To address the issues of high lifespan loss and poor state of charge (SOC) consistency during photovoltaic fluctuation suppression in a battery energy storage system (BESS), an improved K-means dynamic grouping control strategy for BESS under photovoltaic fluctuation suppression is proposed. First, the minimum-maximum scheduling method is used to obtain photovoltaic grid connection instructions. Secondly, an improved dwarf mongoose optimizer (IDMO) is designed and used to improve the traditional K-means clustering algorithm, accelerating the clustering speed. Next, the principle of dynamic grouping of battery cells is formulated, and based on the improved K-means of SOC of battery cells, they are divided into three battery groups. Then, a battery cell SOC consistent power distribution method based on the charge and discharge function is designed, and a BESS two-layer power distribution strategy is proposed. The upper layer determines the charge and discharge order and instructions of the battery pack, and the lower layer calculates the charge and discharge instructions of the battery cell. The proposed strategy is simulated and verified, and the results show that the designed IDMO has higher optimization accuracy and faster optimization. The proposed BESS strategy for suppressing photovoltaic fluctuations effectively suppressed fluctuations while reducing BESS operating life loss and improving the balance of the battery cell SOC.