With the promotion of lithium-ion batteries and the development of hybrid energy storage technologies, the difference of energy storage units in the energy storage system becomes more and more significant, which puts forward challenges for power allocation. In order to reduce the aging cost of energy storage systems during operation, this paper proposes a power allocation strategy based on the cycle-based aging feature of each energy storage unit to reduce aging costs. An online rainflow counting method is proposed for real-time updating of power-cycle mapping relationships, then the real-time subgradients of the aging cost with respect to charging and discharging power are obtained by using the chain rule of differentiation. The reciprocal of real-time subgradients is used as power allocation weight to achieve equalizing aging costs of energy storage units. A transferrable power allocation algorithm is further developed, which allocates power according to weights while considering power limits and the state of charge (SOC) boundary of energy storage units. In an energy storage system participating in secondary frequency regulation, the proposed strategy can reduce the aging cost by 8.08% and 10.34%, respectively, compared with two commonly used power allocation methods (based on maximum power and residual energy).

随着锂离子电池的商业化推广和混合储能技术的发展，规模化储能系统可能存在异构特性。组成异构储能系统的储能单元之间存在差异，对储能系统内部功率分配提出更高的要求。为降低异构储能系统运行过程中的老化成本，首先，给出基于循环的老化模型用于计算不同储能单元老化成本；然后，提出一种无差在线雨流计数法实现功率与循环映射关系的实时更新，并使用链式法则得到储能系统老化成本关于充放电功率的实时次梯度，基于次梯度计算功率分配权重；最后，提出一种基于分配权重的功率迭代分配算法，考虑储能单元的功率边界和荷电状态边界。对分期建设的储能系统参与二次调频的场景进行仿真，结果表明，所提策略相比常用的两种功率分配方法（基于最大功率和基于剩余能量的分配策略）能够分别降低8.08%和10.34%的老化成本。