Health-aware Model Predictive Control including Fault-tolerant Capabilities for Drinking Water Transport Networks

This paper presents a health-aware Model Predictive Control (MPC) including fault-tolerant capabilities for drinking water transport networks. When a fault has occurred, the predictive controller must be redesigned to deal with the fault. This is done by considering the system reliability that is incorporated into the MPC model. The inclusion of components and system reliability in the MPC model is done through the Linear Parameter Varying (LPV) modelling approach with the aim of maximizing the availability of the system by considering system reliability. As a result, the MPC design is modified by considering the reliability model such that additionally to achieve the best achievable performance after the fault, the controller try to preserve the remaining useful life. The solution to the optimization problem related to the MPC problem is achieved by solving a series of Quadratic Programming (QP) problems thanks to the proposed LPV formulation. The proposed approach is applied to a part of a real drinking water transport network of Barcelona for demonstrating the performance of the method.