基于Adam优化算法和长短期记忆神经网络的锂离子电池荷电状态估计方法

State of charge estimation of lithium-ion battery based on Adam optimization algorithm and long short-term memory neural network

锂离子电池是电动汽车、无人机及电力电子设备的储能系统组件,对其进行准确的荷电状态(SOC)估计对于正确决策、安全控制和维护具有重要意义。针对锂离子电池SOC估计问题,本文采用长短期记忆(LSTM)神经网络搭建锂离子电池SOC估计模型,将电池电压、电流、温度作为输入,建立多层LSTM预测模型,采用Adam优化算法与Dropout正则化方法完成LSTM模型的训练。测试结果表明,在模型训练过程中加入Adam优化算法与Dropout正则化方法,使模型对实验数据集的非线性和初始荷电状态的不确定性具有鲁棒性。

Lithium-ion batteries are used as energy storage system components for electric vehicles,unmanned aerial vehicles and power electronic equipment.Accurate state of charge(SOC)estimation plays an important role in correct decision-making,safe control and maintenance.Aiming at the problem of lithium-ion battery SOC estimation,this paper uses long short-term memory(LSTM)neural network to build a lithium-ion battery SOC prediction model,taking battery voltage,current and temperature as inputs,building a multi-layer LSTM prediction model and adopting Adam optimization algorithm.The Adam optimization algorithm complete the training of the LSTM model.The verification of training and test results shows that adding Adam algorithm and Dropout regularization method to the model training process is robust to the nonlinearity of the experimental data set and the uncertainty of the initial SOC.