The modified state prediction algorithm based on KF

The state prediction based on Kalman filter (KF) for linear stochastic discrete-time system is investigated. Predicting future states by using the information of available measurements is an effective method to solve time delay problems; it not only helps the system operator to perform security analysis but also allows more time for operator to take better decision in case of emergency. In addition it can make the system real time monitoring, control, and robust. KF is useful not only for state estimation but also for state prediction. However, the accuracy of prediction degrades notably once a filter uses a much longer future prediction. In this paper, a confidence interval (CI) is proposed to overcome the problem. The advantages of CI are that it provides information about states coverage, which is useful for treatment-plan evaluation, and it can be directly used to specify the margin to accommodate prediction errors. Meanwhile, the CI of prediction errors can be used to correct the predictive state, and thereby it improves the prediction accuracy. Simulations are provided to demonstrate the effectiveness of the proposed method.