Probabilistic duration of power estimation for Nickel- metal- hydride (NiMH) battery under constant load using Kalman filter on chip

For a battery powered safety critical system the safe duration of power for executing a specific task is extremely important. It is necessary to avoid unacceptable consequences due to unwanted battery power failure. An early stage estimation of this duration reduces the overall risk through optimization of current consumption by switching off noncritical load ahead of delivery of power to a critical load. In order to address this issue, an online battery state of charge estimator on chip is conceived and implemented using Kalman filter. The Kalman filter estimates the true values of measurements by predicting a value, considering the estimated uncertainty of the predicted value, and then computing a weighted average of the predicted value and the measured value. The basic idea is more accurate state prediction is possible when the state predicted value is fused with sensor prediction under any uncertain disturbance. The state estimator is developed in the form of an algorithm and stored into a single chip microcontroller. It is finally used to generate an early stage warning signal against battery failure. The paper presents a methodology for creating energy aware system that would avoid sudden system failure due to power outage. The authors used a generalized state space model of the battery to estimate the effect of unobserved battery parameters for duration estimation. An experiment was conducted in this regard through discharging the battery under constant load. Subsequently the internal parameters of battery were calculated. The model was simulated through MATLAB/simulink R2008a software and efficiency was tested. The program for prediction was finally emulated in a microcontroller and found satisfactory result.