Online Monitoring of State of Health for AGM Lead Acid Batteries
Type of DegreeMaster's Thesis
MetadataShow full item record
Prediction of degradation states like power and capacity fade is the most challenging issue for rechargeable batteries, which is called State of Health (SOH). Power and capacity fade is evaluated by predicting Available Power and estimating Maximum Capacity in a battery at any instant. This thesis proposes online estimation methods for SOH of an Absorbed Glass Mat (AGM) lead acid battery based on a second order Randles’ equivalent circuit model (ECM). Since the Maximum Capacity can be simply predicted based on estimated state-of-charge (SOC), this thesis has been mainly focused on prediction of Available Power. The Available Power is calculated based on a maximum allowable current and the terminal voltage using a second order Randles’ ECM. However, the parameters of the model vary continuously because of effects of amplitude of current, temperature, SOC in addition to aging process. After review of different methods of parameter estimations, I reformulated the continuous equation of the model into a difference equation of the Autoregressive Model with Exogenous input (ARX) and applied Linear Kalman Filter (LKF) to estimate the parameters. The performance of this technique has been better than recursive least square (RLS) methods, particularly at rapidly varying parameters optimized by selection of appropriate covariance matrices. In addition, the pre-calculated open circuit voltage (OCV) can reduce the number of the parameters that allows for stability of the estimation. For capacity fade, the Maximum Capacity is estimated using RLS under assumption that SOC is known. Experimental validation of both Available Power prediction and Maximum Capacity estimation are conducted under aging condition as batteries are cycled. At the end, this thesis shows evaluation of SOH using the Available Power and Maximum Capacity.