Energy Management And Optimization Strategies Of Reconfigurable Battery Packs

With the rapid growth in the number of electric vehicles,a large number of retired batteries will be generated in the future years.These retired batteries typically have 60%-80% of their capacity remaining and can be used for grid storage,thereby reducing the life-cycle costs of electric vehicles and relieving environmental pressures caused by dismantling batteries.However,the ”barrel effect” of traditional fixed-connected battery packs can lead to a significant reduction in the effective capacity and cycle life of energy storage systems due to the vast differences in characteristics among retired batteries.Reconfigurable battery packs are particularly suitable for energy storage systems with large differences in battery characteristics,as they can be flexibly switched between series and parallel connections,and overcharged/overdischarged/overheated/faulty batteries can be removed at will.The flexibility of reconfigurable battery packs is determined by the number of switching devices.The greater the number of switches per cell,the more flexible the connections among the cells,and the better it is to overcome the effects of inconsistencies in the batteries.However,the large number of switches makes reconfigurable battery packs suffer from significant shortcircuit risks and makes the optimization of path combinations face a ”dimensional disaster”,which severely limits the application of reconfigurable battery packs in large-scale energy storage systems.This thesis focuses on the above-mentioned bottlenecks of reconfigurable battery packs,with the main research elements including:（1）A path search and analysis method based on sneak circuit theory is proposed for the short-circuit problem of reconfigurable battery packs.Based on the topology of the reconfigurable battery pack,a directed graph is built and its connection matrix is extracted.All potential paths are searched by calculating the corresponding cofactor of the connection matrix.Based on the characteristics of the analyzed paths,false paths,complex paths,short-circuit paths and simplest paths are defined and the elemental composition features of their corresponding mathematical expressions are grasped.The search methods for short-circuit paths and simplest paths are designed,and the short-circuit path table and the simplest path table are established respectively as the basis for path switching control of reconfigurable battery pack.The experimental prototype is built,and the proposed method is proved to be effective in avoiding short-circuit of reconfigurable battery pack through constant voltage discharge experiment and multi-level switching experiment.（2）A circulation suppression strategy based on crisscross optimization is proposed for the circulating current problem of reconfigurable battery packs.With the main objectives of constructing the most parallel paths and suppressing circulating current,the electrical energy optimization model for reconfigurable battery packs is established,taking into account circuit topology,output voltage range,and other constraints.Round binary crisscross optimization algorithm is designed based on the characteristics of optimisation model.Hence,overcoming the ”dimensional disaster” problem in the path combination optimization of the reconfigurable battery pack.On an experimental prototype,the effectiveness of the proposed method for circulation suppression is verified.（3）In order to meet the requirements of estimating the battery state accuracy for the optimal control of reconfigurable battery packs,a method for estimating the state of charge in lithium-ion batteries based on the relaxation effect is proposed.Firstly,by collecting a large amount of charge/discharge experimental data,the battery relaxation effect is analyzed,and the relationship between discharge current,relaxation time and state of charge is established.Based on the characteristics of the relaxation effect data,a convolutional neural network is designed to achieve the estimation of the SOC of battery.The experimental results show that the accuracy of the proposed method to estimate the state of charge is higher than that of the last-minute voltage estimation using the relaxation effect.The work in this thesis reduces the short-circuit risk of reconfigurable battery packs,improves the efficiency of reconfigurable battery packs,and lays a theoretical foundation for the application of reconfigurable battery packs in large-scale energy storage systems.