【作者】 赵克刚； 罗玉涛； 裴锋；

【Author】 ZHAO Ke-gang,LUO Yu-tao,PEI Feng(Guangdong Key Laboratory of Electric Vehicle Research,South China University of Technology,Guangzhou 510640,China)

【机构】 华南理工大学广东省电动汽车研究重点实验室；

【摘要】 为有效地利用神经网络模型对电动汽车电池荷电状态(SOC)进行预测,在分析铅酸电池放电过程时变特性的基础上,提出采用电池放电过程的时变特性作为神经网络的输入,使得模型能更好地反映电池动态特性。建立基于径向基函数神经网络,用于估计电池SOC。结果表明:采用时变特性输入可以有效地提高网络模型预测SOC的能力。更多还原

【Abstract】 In order to predict state of charge(SOC)of battery in electric vehicles more effectively,by analyzing varying rule of electromotive force,residual capacity and equivalent inner resistance of lead-acid battery during discharge,input of neural network model should adopt the curve altering with time.So the model would reflect the dynamic characteristic of battery more exactly.Battery model based on radial basis function neural network was built up to estimate SOC of battery.The results show that the neural network trained using the curve altering with time has achieved improved ability to predict SOC of battery.更多还原