作　　者： ; ; ;

机构地区： 华南理工大学机械与汽车工程学院广东省汽车工程重点实验室

出　　处： 《电源技术》 2007年第11期914-917,共4页

摘　　要： 目前电动汽车动力电池荷电状态(SOC)的辨识误差约在8%左右,且主要集中在电池恒流放电过程的辨识,对电池交流放电状态中SOC的辨识研究不是很多。在实际应用中,尤其是在混合动力电动汽车中,电池多处于变流放电状态中,而且电流幅值变化较大。为此,提出了基于电池时变特性的径向基神经网络SOC辨识法。该方法摒弃了以电池单点时刻状态参数作为网络输入的做法,采用动力电池变流放电参数为输入,使辨识精度提高到3%。此方法尤其对动力电池处于交流放电状态时,效果更加明显。 The estimation error of the state of charge （SOC） for power battery pack is about 8% at present. The estimation methods are mainly focusing on the constant discharging current process, and less research on the variable discharging current state. In fact, the battery is mostly in the variable discharging current state and the current range varies in a much value especially in the hybrid electric vehicles. A SOC estimation method was presented, which was based on the radial radix neural network taking the variable current in account. This method spurns the disadvantage of using the single state parameter but using the variable discharging parameter of power battery as the network input and improved the estimation error to 3% .The method was effective, especially on the variable discharging current state of power battery.

关 键 词： 动力电池 荷电状态 辨识 神经网络 变流放电

领　　域： [电气工程]