基于参数敏感度和数据驱动的锂电池老化模型建模及剩余寿命预测方法研究

Lithium battery lifetime (durability) is one of the major bottlenecks of the battery storage system. Frequent charge/discharge will lead to accelerated aging of the battery, so that the consistency between the cells is getting worse. The aging process is related to the internal battery chemical reactions which are difficult to describe with circuit model parameters. In addition, with the widespread use of lithium batteries, it is necessary and urgent to make accurate judgments of the state of the batteries (SOH) and remaining useful life (RUL), especially on batteries decommission. Based on the analysis of the existing state identification methods, aiming at the characteristics of large quantity of cells and high reliability of lithium battery energy storage system, a new idea for online identification of battery internal state is put forward. Three aspects of the technical research route, analytical modeling, analysis of sensitive parameters and compound estimation of model parameter, are determined to carry out research, which are focusing on the aging mechanism and the battery internal physical and chemical reactions and thermal effects modeling, analysis of parameter sensitivity of fusion mechanism model and compound estimation algorithm of SOH and SOC model parameter. The study is expected to realize the breakthrough of the key theoretical and technical problems in the three aspects of modeling, identification and estimation. This project is an interdisciplinary field across electrical engineering, electrochemistry and detection technology.

锂电池组的寿命和耐久性是电池储能系统的主要瓶颈之一，频繁的充/放电导致电池加速老化，使电池单体间的一致性越来越差。电池老化过程与电池内部的化学反应相关，很难用电路模型参数描述，而电池的电化学模型又不能直接应用在储能系统的参数估计上。此外，随着锂电池的广泛应用，大量的退役电池鉴别工作也亟待解决，需要对电池的健康状态（SOH）和剩余寿命（RUL）做出准确的判断。本项目在分析现有状态辨识方法的基础上，针对锂电池储能系统电池数量多及可靠性要求高的特点，提出了在线辨识电池内部状态的新思路，确定了分析建模、参数敏感分析及参数复合估计三个方面的技术研究路线，重点围绕基于老化机理的电池内部物理化学反应及热效应的模型建模、融合机理模型参数敏感度分析和SOH、SOC模型参数复合估计算法开展研究，实现建模-辨识-估计三个环节关键理论和技术问题的突破。本课题属于电气工程、电化学及检测技术等多学科的交叉学科。

随着电动汽车的普及和储能系统的大量建造，锂电池的应用越来越广泛，在一些重要的应用场合对锂电池工作状态监测的要求越来越高。在工况复杂的电动汽车和各种绿色储能系统中，电池管理系统（BMS）在保障系统安全运行方面起到了至关重要的作用。而电池管理系统有效控制的基础就是锂电池工作状态参数的准确识别。本项目主要聚焦于锂电池的剩余电量（SOC）、健康状态（SOH）和剩余寿命（RUL）等参数快速准确辨识问题的探讨，并取得了重要成果，从多维度、多角度系统地研究了锂电池在全生命周期中各阶段的特性，提出了多种有成效的算法，尤其是基于部分数据的锂电池多参数联合辨识算法，解决了锂电池在不同工作状态下各参数之间的非线性耦合问题，提高了电池健康状态的辨识精度，为项目成果的推广应用打下了基础。基于本项目的研究成果，项目组还完成了国网天津电力公司的多个储能系统项目，得到了较高的评价。目前项目组正和相关企业合作将项目成果推广到国防事业中。