Condition monitoring of nonlinear systems via multiple Kalman filters

PhD project. Supervisor – Professor T X Mei

Model-based methods are widely studied for applications in FDI and general condition monitoring, but are more suited to tackle linear time invariant systems and can be problematic for systems with very complex and non-linear properties. This project studies a multiple Kalman filters approach that divide a non-linear problem into a number of linearised models that represent an approximation of different conditions. A selected number of Kalman filters are then used to assess the likelihood of a ‘best’ match of the condition of the system is operating at any particular instance to achieve the desired condition monitoring.

This project is supported financially by the Government of Pakistan through a postgraduate scholarship programme.