Fault Detection and Adaptive Parameter Estimation with Quantum Inspired Techniques and Multiple-Model Filters

Fault detection and quick and adaptive estimation of parameters to be used in controllers are crucial to successful operation of the plants. In fact, when a fault occurs, parameters of a system can change drastically; in order to cope with such changes, a quantum-boost scheme for multiple-model Kalman filter is presented in this study. In fault detection, it is of vital significance that the posterior probability corresponding to the best model rises and converges quickly. This can be achieved with the extended Grover's algorithm, originally developed for quantum information processing. Efficacy of the quantum-boosted multiple-model Kalman filter is tested in this study with two examples. In both examples, the quantum boost scheme is seen to accelerate the rise and convergence of the probability corresponding to best model for the unknown initial parameter and also the changed parameter. With quantum-boosted multiple-model Kalman filter, fast fault detection can be achieved by monitoring the probabilities of the assumed values of the parameters.