Wavelet-based Model Predictive Control of PWR Nuclear Reactor using Multi-Scale Subspace Identification

This work presents multi-scale model predictive control design scheme employing wavelet basis function. The proposed scheme is established upon multi-scale subspace identification technique. It is aimed to utilize the proficiency of wavelets in multi-scale data projection and the robustness of subspace identification during estimation in a model predictive control setup. The multi-scale state-space models estimated at different scales are used for output prediction and for designing predictive control strategy. The competence of the proposed approach is established for constrained load-following problem of a pressurized water-type nuclear reactor. In addition, the fault-tolerant capability of the control algorithm is also tested.

https://eventi.unibo.it/acd2019