The liquid metal battery is a new type of large capacity energy storage battery meeting require-ments of large scale energy storage of power grid and the accurate on line estimation of its charge state is animportant foundation for its large scale application. However

there is presently no published literature foron-line estimating the state of the battery. For this purpose

the on-line estimation method of the chargestate of the battery is established based on extended Kalman filter algorithm. The equivalent physical modelconsidering the self-discharge factor of the battery is established and

on this basis

the on-line estimationmethod of charge state based on extended Kalman filter is proposed. It is shown by the experimental resultthat under different battery working conditions and different initial value of the state of charge

the estima-tion of the state of charge of the battery has good accuracy. The standard error between the calculated andmeasured value is about 0.018 3 and the average error between them is about 0.014 5. This algorithm can ef-fectively solve the problem caused by inaccurate estimation of initial battery parameters.