电动汽车的发展对锂离子电池的比功率和比能量提出了更高的要求，除更换电池单体材料等方法外，电极层结构的设计由于成本相对较低、能够使电池的实际性能更加接近理论性能而受到格外关注。产业界进行电极层结构设计仍然以试错法为主，这种方法在设计过程中极度依赖人工经验且设计结果无法达到最优，本文将电极层结构设计问题转化为多目标优化问题，通过多目标优化算法与可制造性分析，设计出性能最优且性能对生产误差最不敏感的电极层结构。本文通过探讨模型、优化变量以及优化形式的确定问题，确定了电极层结构设计问题的数学表达形式。通过对比不同类型的模型，明确修正后的均相电化学模型最适合电极层结构设计问题，用于反映电极层结构与电池性能之间的映射关系；通过对电极结构参数做全局敏感性分析，发现正负极孔隙率、正极厚度和正负极活性物质颗粒半径对电池比能量和比功率都有一定程度的影响，这五个电极结构参数都应该作为数学表达中的优化变量；本文将比能量和比功率数据的Spearman秩相关系数作为判断是否进行多目标优化的依据，确定同时最大化比能量和比功率的优化形式为多目标优化。选用不同类型的多目标优化算法对电极层结构设计问题进行求解，并从计算时间、帕累托最优解的质量、算法原理等不同角度对比多目标优化算法，贝叶斯优化算法由于计算次数最少、帕累托最优解对应的比能量和比功率最大、帕累托最优解之间分散度最高，最适合电极层结构设计问题。多目标优化问题的特性决定帕累托最优解个数不唯一，需要从中筛选出指导生产的最优解，本文将可制造性作为筛选标准并提出相应的量化指标，在设计阶段降低生产误差的影响，提升成品的一致性。利用不确定性传播求解多目标优化得到的多个帕累托最优解的可制造性，对比筛选出可制造性最高的作为最终的设计结果。通过过电压分解以及浓度场信息分析，发现若某最优解放电末期的正极与集流体交界处液相锂离子浓度较小，其可制造性一定很差，不必对其进行不确定性传播，从而缩短可制造性评估的时间。通过以上的研究，最终找到比能量和比功率最优，且比能量和比功率受生产误差影响最小的电极层结构，用于指导锂离子电池的生产。

The booming electric vehicles require higher specific energy and specific power of lithium-ion batteries. In addition to the replacement of cell materials, electrode structure design has received much attention because of its relatively low cost and the ability to make the actual performance of the battery closer to the theoretical performance. The electrode structure design in the industry is still dominated by the trial-and-error method. This method relies heavily on manual experience in the design process, which can only guarantee better but not optimal design result. In this study, electrode structure design problem is transformed into a multi-objective optimization problem. Multi-objective optimization algorithms and manufacturability analysis are applied to design the electrode structure whose performance is optimal and the least sensitive to production errors.In this study, the choices of model, optimization variables and optimization form are analyzed to get mathematical formulation of the electrode structure design problem. Through the comparison of different types of models, it is concluded that for electrode structure design problem, the revised homogeneous electrochemical model is the most suitable to reflect the mapping relationship between the electrode structure and battery performance. Through global sensitivity analysis of electrode structure parameters, it is found that cathode porosity, anode porosity, cathode thickness, cathode particle radius and anode particle radius influence the specific energy and specific power of the battery to some degree. These five electrode structure parameters should all be treated as optimization variables. The optimization form is determined by the Spearman rank correlation coefficient of specific energy and specific power, and the result shows that multi-objective optimization should be taken in this study.By comparing different types of multi-objective optimization algorithms from different perspectives, for example, calculation time, quality of Pareto optimal solutions, algorithm principle, etc., it is found that the Bayesian optimization algorithm has the least number of calculations, the highest specific energy and specific power corresponding to Pareto optimal solutions, and the highest dispersion of Pareto optimal solutions. Bayesian optimization algorithm is the most suitable for electrode structure design problem.The characteristics of the multi-objective optimization problem determine that the number of Pareto optimal solutions is not unique, and it is necessary to screen one among these multiple optimal solutions to guide production. This study proposes the concept of manufacturability and the corresponding quantitative indicator, which aims to reduce the impact of production errors in the design stage and improve the consistency of the finished product. Manufacturability of each Pareto optimal solution is obtained by uncertainty propagation. The one with the highest manufacturability is chosen as the final design to guide production. Through the analysis of overpotential decomposition and concentration field information, it is found that at the interface of cathode and cathode current collector, if the lithium ion concentration in the electrolyte of some optimal solution is small at the end of discharge, it can be deduced that the manufacturability of this optimal solution is poor, so there is no need to conduct uncertainty propagation, which reduces time of manufacturability assessment.The specific energy and specific power of resulting electrode structure are optimal and least affected by production errors, which is most suitable to guide the production of lithium-ion batteries.