【作者】 刘斌；

【导师】 胡志坤； 林勇；

【作者基本信息】 中南大学 ， 电子科学与技术， 2014， 硕士

【摘要】 摘要：本文以镍氢电池为实验对象,设计了一套基于单片机C8051F020的电池恒流放电实验装置。为研究动力电池放电特性,将10节额定电压为1.2V,额定容量为30Ah的镍氢电池进行了常温条件下的放电实验,并在此基础上对反映动力电池性能及工作状态的荷电状态(State of Charge, SOC)估计方法进行了研究。针对卡尔曼滤波算法在估算SOC时过于依赖模型提出了利用子空间辨识算法先确定模型阶数,以便选取更合适的电池模型,根据辨识的结果,本文选的实验对象的模型是一阶的,以此为据,选取了经验公式模型。另外利用最小二乘法结合开路电压法辨识出了模型的未知参数,得到的等价模型的最大误差为0.12V。在估算SOC时,针对平方根无极卡尔曼滤波算法中假设噪声协方差为定值,不能实时的更新而带来估计误差的缺陷,根据反馈原理,本文做了改进,即将每个时刻的模型输出量的残差作为新息来估算相应时刻的噪声协方差,使其实时跟新,具有自适应性,从而降低了估计误差。本文利用了一种非线性的普通模型对改进后的算法进行了仿真,结果表明改进是有效的。最后将改进后的算法应用到了电池SOC估计中,以安时法估算得到的SOC作为标准值,与利用改进后的算法估算得到的SOC值比较,误差降到1.5%以内。因此得到了一种的新的、估计精度更高的自适应平方根UKF算法。更多还原

【Abstract】 Abstract:This paper put the Ni-MH batteries as experimental object and designed a battery constant exile electric circuit based on MCU C8051f020. For the study of power battery discharge characteristics,10sections Ni MH battery that rated voltage is1.2V and rated capacity is30Ah were used to make discharge experiments at room temperature, and studied estimation methods of State of Charge(SOC) that reflect the performance and working State of power battery on this basis. View of the Kalman filtering algorithm rely too much on the model in estimating SOC,this study used the subspace identification algorithm to determine the order number, in order to choose more appropriate battery model. According to the result of identification, the selected subjects is one order model, based on this, selected the model of the empirical formula. And least squares method combined with open circuit voltage method is used to identify the unknown parameters of the model, the equivalent model of the maximum error is0.12V. In estimating SOC, To the square root UKF algorithm assume the noise covariance is a constant that can produce error when use it to estimate SOC. According to the principle of feedback, this article made improvements, put model output residuals of each moment as the new rate to estimate the noise covariance of the moment corresponding, make it new with time, adaptability, which reduces the estimation error. And the improved algorithm were simulated with a model that is a general and nonlinear model. The result showed that the improvement is effective. Last the improved algorithm was applied to the battery SOC estimation. The error is less than1.5%compared with the standard values that geted with AH method. So get the adaptive square root UKF that is a new, higher estimation precision algorithm.更多还原