Search

Battery energy storage system (BESS) plays an important role in the grid-scale application due to its fast response and flexible adjustment. Energy loss and inconsistency of the battery will degrade the operating efficiency of BESS in the process of power allocation. BESS usually consists of many energy storage units, which are made up of parallel battery clusters with a cell-pack-cluster hierarchical structure. This paper presents a power allocation strategy based on cluster switching to relieve the stated problem in two levels. Cluster switching is identified as a new control approach to eliminating the imbalanced state-of-charge (SOC) in the cluster level. In the unit level, an optimization model is constructed for power allocation, where the objective function consists of two aspects: minimizing battery energy loss and maximizing SOC consistency. Finally, the effectiveness of the power allocation strategy for BESS in peak shaving and frequency regulation is validated by simulation and scaled-down experiment. The results prove that the power allocation strategy can reduce the battery energy loss and prevent from overcharging/overdischarging to extend the battery lifetime.

The safety of battery is of great importance in retired battery based energy storage system (ESS). When the state-of-health (SOH) of battery reaches the lower limit, the battery should be replaced to prevent accidents. In the energy storage power station, all the batteries are expected to be replaced at the same time so that the repair time and the maintenance cost can be reduced. This paper proposes a coordination strategy to equalize SOH among energy storage units and relieve inconsistency of batteries. Firstly, based on lithium-ion battery life cycle model, a method for SOH equalization control is derived. Secondly, an optimization model is constructed for the coordinated control in energy storage power station. Finally, the simulation shows the effect of the coordinated control strategy in this paper. This strategy can be applied to the energy management in retired battery energy storage power station for on-site control.

With the development of information society, Recommendation System has been an import tool to help users filter information and create more economic value for enterprises. However, it is difficult for traditional recommendation systems to interpret recommendation results. In order to improve users’ trust in recommendation results, interpretable recommendation models have attracted more and more attention. In this paper, we present the Explanation Chains Model based on the Fine-grained Data (F-ECM) to enhance the effectiveness of recommendation while achieving the interpretability of recommendation. First, we generate parsing trees from user comments and extract three key sentence structure information (i.e., aspects, features and sentiment tendency) from those generated parsing tree. The fine-grained similarity is computed based on the aspects and features of the products to be recommended, and users’ product satisfaction is predicted by combining sentiment tendency. Then the recommendation chain will be constructed according to the satisfaction degree in the recommendation list. Finally, we calculate recommendation chain scores of all the items to be recommended to the target user, generate the recommendation results and personal explanation for the user by the recommendation chains. Experiments in the Amazon data set show that the Explanation Chains Model based on the Fine-grained Data achieve better interpretability and performance of product recommendation systems.