Abstract: The prediction and estimation of State Of Charge（abbreviated to SOC） is an important part of the lithium battery management system. According to the advantages and disadvantages of the genetic algorithm and the ant algorithm, this paper puts forward a method that the genetic algorithm is combined with ant algorithm to be GAAA algorithm to optimize the BP neural network. In order to realize the SOC estimation, the input values should be the real-time working current, voltage, temperature, the healthy degree of lithium battery and the value of ampere-hours integration. This method is implemented by programming based on MATLAB. Results show that this method has higher accuracy of prediction of SOC and faster operating speed than the traditional BP method and the BP neural network based on the genetic algorithm.

Key words: lithium battery, battery management system, State Of Charge（SOC）, Genetic Algorithm-Ant Algorithm（GAAA）, Back Propagation（BP） neural network

摘要： SOC（荷电状态）的预测和估算是锂电池管理系统中的一个重要部分。根据GAAA算法充分利用了遗传算法和蚁群算法各自的优势，提出一种GAAA算法优化BP神经网络的SOC估算方法。使用MATLAB进行编程，将锂电池的实时工作电流、电压、温度、健康度、安时积分值作为输入，实现对SOC的估算。实验结果表明，该算法在估算精确度和运算速度上都优于传统的BP神经网络和基于遗传算法的BP神经网络。

关键词: 锂电池, 电池管理系统, 荷电状态, 遗传-蚂蚁算法（GAAA）, 反向传播（BP）神经网络