APPLICATION OF KALMAN FILTER IN SEQUENTIAL UPDATE OF RECOVERY PROJECTION OF ELECTRIC POWER SUPPLY DURING DISASTER

Abstract

 In case where lifeline functions are degraded due to disaster, rapid and accurate dissemination of recovery projection is essentially important. In this study, aiming at improving the reliability of such information, we applied Kalman filter technique for sequential update of recovery projection of electric power outage. First, decreasing process of power outage is sequentially fitted by an exponential function. On the basis of state space model, the sequential updating process is formulated for the predicted distribution of the model parameter representing recovery pace. Then, Kalman filtering and Kalman prediction are performed, and three expected restoration curves are obtained based on the mean and the 95% confidence interval of the predicted distribution. In addition to two extreme curves assuming the worst and best pace are also shown. It is suggested that the possible range of recovery projection should be addressed by considering the uncertainty in recovery estimates.