Abstract: Aiming at the state of charge (SOC) imbalance of battery units and neglecting the wind-storage net income in the process of smoothing wind power fluctuations by battery energy storage system (BESS), we proposed a power allocation strategy for BESS where the SOC balance and the income are taken into account. Firstly, an optimization model for grid-connected wind power signal was established comprehensively by taking into account multiple factors of the electricity sale revenue, wind curtailment penalty, electricity shortage penalty, and operation cost of BESS. Meanwhile, multiple factors of grid-connected power signal volatility, standard deviation of SOC for battery groups, and so on, were selected as the constraints. An improved arithmetic optimization algorithm (IAOA) was proposed to solve the optimization model. Then, the BESS was divided into two battery groups, and a double-layer power allocation method (DPAM) for BESS was designed. The upper layer assigns the charging/discharging power signal of BESS to the two battery groups, and the lower layer allocates the charging/discharging power signal of each battery group into its battery units with the power allocation principle of maximum charging/discharging power or novel SOC balance. Finally, the proposed strategy was verified by simulation. The results show that IAOA speeds up the optimization speed and improves the optimization accuracy. DPAM enhances the speed of SOC balance for battery units in battery groups and improves their balance degree. The proposed power allocation method can be adopted to further reduce the volatility of grid-connected wind power and improve the net income of wind-storage system.