基于SABO-GRU-Attention的锂电池SOC估计  

Prediction of lithium battery SOC based on SABO-GRU-Attention

作　　者：薛家祥[1] 王凌云 XUE Jiaxiang;WANG Lingyun(School of Mechanical and Automotive Engineering,South China University of Technology,Guangzhou Guangdong 510640,China)

机构地区：[1]华南理工大学机械与汽车工程学院,广东广州510640

出　　处：《电源技术》2024年第11期2169-2173,共5页Chinese Journal of Power Sources

基　　金：广东省自然科学基金项目(2214050007061)。

摘　　要：提出一种基于SABO-GRU-Attention(subtraction average based optimizer-gate recurrent unitattention)的锂电池SOC(state of charge)估计方法。采用基于平均减法优化算法自适应更新GRU神经网络的超参数,融合SE(squeeze and excitation)注意力机制自适应分配各通道权重,提高学习效率。对马里兰大学电池数据集进行预处理,输入电压、电流参数,进行锂电池充放电仿真实验,并搭建锂电池荷电状态实验平台进行储能锂电池充放电实验。结果表明,提出的SOC神经网络估计模型明显优于LSTM、GRU以及PSO-GRU等模型,具有较高的估计精度与应用价值。A state of charge(SOC)estimation method of lithium battery based on SABO-GRU-Attention(subtraction average based optimizer-gate recurrent unit-attention)was proposed.The subtraction average based optimization algorithm was used to adaptively update the hyper-parameters of GRU neural network,and SE(sequence and exception)attention mechanism was used to adaptively allocate the weights of each channel to improve learning efficiency.The University of Maryland battery data set was preprocessed to input voltage and current parameters,and lithium battery charge and discharge simulation experiments were carried out,and the lithium battery state of charge experimental platform was built for the charge and discharge experiments of energy storage lithium batteries.The lithium battery state of charge experimental platform was built for charging and discharging experiments of energy storage lithium battery.The results show that the proposed SOC neural network estimation model is obviously superior to LSTM,GRU and PSO-GRU models,and has high estimation accuracy and application value.

关 键 词：SOC估计 SABO算法 GRU神经网络 Attention机制 

分 类 号：TM912[电气工程—电力电子与电力传动]