Embedding Monotonicity in the Construction of Polynomial Open-Circuit Voltage Model for Lithium-Ion Batteries: A Semi-infinite Programming Formulation Approach

作者：He, Yi-Jun; Shen, Jia-Ni; Shen, Ji-Fu; Ma, Zi-Feng^*

来源：Industrial & Engineering Chemistry Research, 2015, 54(12): 3167-3174.

DOI：10.1021/ie5044049

• STATE-OF-CHARGE
• INCREMENTAL CAPACITY ANALYSIS
• GLOBAL OPTIMIZATION
• CONTINUOUS MINIMAX
• PRIOR-KNOWLEDGE
• ALGORITHM
• HEALTH
• REGRESSION
• VEHICLES
• SIDE

Accurate modeling of open-circuit-voltage (OCV) plays important roles both in state-of-charge (SOC) estimation and state-of-health (SOH) monitoring for lithium-ion batteries (LIBs). Monotonicity violation in OCV model would lead to inaccurate SOC estimation and ineffective of incremental capacity analysis (ICA) for SOH monitoring. In this study, first-order derivative of OCV, with respect to SOC is introduced to theoretically ensure the satisfaction of monotonicity and a nonlinear semi-infinite programming (NSIP) problem is constructed to parameter estimation. A global optimization approach via restriction of the right-hand side is used to efficiently and globally optimize the NSIP. Both fitting and ICA results demonstrate the effectiveness of the proposed method. Moreover, in comparison to the traditional polynomial and sigmoid models, the NSIP polynomial model is the best choice for performing further SOC estimation and SOH monitoring. The results thus indicate that a NSIP framework for embedding prior knowledge not only provides a promising approach to automatically capture OCV-SOC monotonicity constraint in LIBs, but also serves as a universal methodology for process modeling with the requirements of embedding derivative constraints.

• 出版日期2015-4-1
• 单位上海交通大学