A recursive identification algorithm for estimating abrupt parameter changes

A simple and effective algorithm is derived for estimating abrupt parameter changes by using a combination of detection theory and a Kalman filter algorithm. Simulation results show its excellent ability for tracking sudden parameter changes, and show that the estimates of unchanged parameters are not influenced by the tracking of the changed parameter. The disadvantages of the method, shared with other algorithms for the same purpose, may be a delay of several steps in detecting several parameter changes, and the possibility of false alarm or failure in detecting a change. The estimation may be beneficial when the associated delay is acceptable