Fast equalization of lithium battery energy storage system based on large-scale global optimization

The growing emergence of electric vehicles brings the problem of retired lithium-ion batteries (LiB) proliferation, so the retired LiB with different state-of-health (SOH) values are urgent to be employed for the second-life application. Due to the Matthew's effect caused by SOH difference, effective SOH equalization is required to achieve stable performance. In this study, the SOH equalization for large LiB system is established as large-scale global optimization problem, and the model predictive control (MPC) is introduced to control the depth of discharge (DOD) dynamically. In order to overcome the "curse of dimensionality" problem, a novel algorithm namely GA LSE is proposed, in which the solution space segmentation and reorganization mechanism, and the improved selection, crossover and mutation operations are introduced to dispatch the power flows to achieve fast equalization speed. Experimental results show that with the utilization of GA LSE algorithm, the high-dimensional equalization model with up to 1000 variables can be effectively optimized, the convergence speed and accuracy are significantly better than that of the state-of-the-art algorithms. In addition, when the GA LSE algorithm is further integrated with MPC-based DOD control mechanism, the SOH values of large retired LiB packs can be effectively equalized with high accuracy and fast response speed.