Early prediction of relaxation voltage & detection of state of battery for functional safety

During parking condition of vehicle, state of the battery is uncertain as it goes through the relaxation process. In such scenarios the battery voltage may exceed the functional safety limits. If we cross the functional safety limits it is hazardous to the driver as well as the occupant. In this case relaxed voltage plays a crucial role in identifying the safe state of the battery. To estimate the relaxed cell voltage there are methods such as RC filter time constat modeling and relaxation voltage error method. The problem with these solutions is the waiting time and accuracy to determine the relaxation voltage. In this manuscript a solution is proposed which ensures the above problem gets reduced. To achieve the reduction of relaxation voltage estimation time, a python sparse identification of nonlinear dynamics (PySindy) is used which identifies and fits an equation model based on observing the battery characteristics at different SOC and temperatures. The implementation is done and compared with the existing algorithms at different temperature and SOC levels. It is being validated that manuscript predicts relaxation voltage within 1s having Mean Squared Error (MSE) of 0.04mV. In the existing method, it takes minimum 30seconds of data to estimate relaxation voltage having a mean absolute error of 2.99mV. As a conclusion, manuscript being efficient and accurate to predict the relaxed voltage (OCV) which enhances the estimation of state of battery for functional safety aspects.