基于改进滑模观测器的锂离子电池荷电状态估计方法

Improved sliding mode observer for state of charge estimation of lithium-ion battery

提出一种基于改进滑模观测器的SOC估计算法,采用sigmoid函数代替传统滑模观测器中的符号函数,并对观测器中的模型参数进行在线更新,进一步消除参数变化带来的模型误差,良好的解决了传统滑模观测器存在的抖振问题,且对参数扰动和模型不确定性有很强的鲁棒性。此外,根据电池状态空间方程构建ARMAX模型,并将预报误差法应用于参数辨识中,可充分体现电池的动态特性。为了验证所提方法的估计效果,对锂离子电池在UDDS工况和混合充放电工况下进行测试,结果证明了整套方案的有效性和鲁棒性。

This paper presents a state of charge(SOC)estimation method for Lithium-ion battery based on an improved sliding mode observer.A sigmoid function is chosen to substitute the sign function of the conventional sliding mode observer.The improved observer has a capability of compensating modeling errors caused by the parameters variation and minimizing chattering level in SOC estimation using conventional sliding mode observer.Further,an ARMAX model is established according to the state space functions and the model parameters are identified through the prediction error method.This model can describe the dynamic characteristics of the battery.Finally,the robustness and effectiveness of the proposed algorithm is validated by the urban dynamometer driving schedule drive cycle(UDDS)and the UDDS test mixed with constant current charging condition.