基于免疫遗传算法的动力电池SOC估计研究

Optimal Estimation of State of Charge for Power Battery Based on IGA-BP Algorithm

摘要:

锂离子动力电池SOC（电池荷电状态）难以直接测量且由于高度非线性所导致估计误差较大。为了减少动力电池SOC估计误差，提高估算精度。在分析了锂离子动力电池电压、温度、电流和放电电量对电池SOC影响后，提出一种新颖的免疫遗传算法（Immune Genetic Algorithm，IGA）和BP神经网络相结合的锂离子动力电池SOC值联合估计方法，该方法首次使用在锂离子动力电池SOC值估计中，采用新颖的免疫遗传算法通过对BP神经网络进行参数寻优，优化网络结构模型，增强神经网络自适应学习效率。通过仿真和动力电池实际工况下实验，结果表明使用新颖的联合估计算法提高了网络的运行效率和电池SOC值估计精度，估计均方根误差控制在2%以内，验证了这一联合估计算法的可行性和有效性，解决了动力电池SOC值估计误差较大的问题。

Abstract:

It is difficult to directly measure the SOC value (battery state of charge) of a lithium ion battery, and a large estimation error is caused due to high nonlinearity. In order to reduce the battery SOC estimation error, improve the SOC estimation accuracy. After analyzing the effect of voltage, temperature, current and discharge electricity of lithium-ion battery on battery SOC, a novel Immune Genetic Algorithm (IGA) combined with BP neural network was proposed for SOC value of lithium-ion battery. This method is used for the first time in the estimation of SOC value of lithium-ion battery, using a novel immune genetic algorithm to optimize parameters of BP neural network, optimize the network model, and effectively improve the network learning efficiency and battery SOC value. Finally, through simulations and experiments under the actual conditions of the power battery, the results show that the use of a novel joint estimation algorithm improves the network operating efficiency and the battery SOC value estimation accuracy, estimates the root mean square error control within 2%, and validates this joint estimation algorithm. The feasibility and effectiveness of the solution to the problem of large error in battery SOC estimation.