Search

Behavior of Radio Frequency (RF) cavity as a dynamical system has been well explained by its state space model especially in case of normal conducting cavity. In case of superconducting cavity, one needs to also model Time Varying Detuning (TVD) of cavity by a additional non-linear equations. The TVD forms one of parameter of system matrix of cavity state space model and is function of number of mechanical modes of cavities. Often one misses out or fails to model one or more of these modes. Given, that system designer has modeled system to maximum possible extent, there is a need to detect if any dynamics are unmodeled. Kalman filter (KF) and its many variants have been at forefront as estimators for states and parameters of system governed by dynamical equation in presence of uncertainty. It estimates states of system by running in parallel as a system with model similar to that of plant and with its inputs and outputs fetched from plant/system. Under such a situation, KF, while its operation, outputs an innovation sequence. If all dynamics of process are well captured by KF model, the innovation sequence is white noise and zero mean. Any model mismatch due to unmodeled or wrongly modeled dynamics are indicated by innovation sequence. In this work, authors attempt to detect such dynamics using superconducting RF cavity model as a plant in a case study. This paper must be viewed as a novel usage of KF. The proposed method may find potential use in field of dynamics detection or system identification. This paper describes RF cavity model and a discrete KF followed by a mechanism to detect unmodelled dynamics by modeling and simulations studies.