Fault-tolerant model predictive control of a de-manufacturing plant

This paper describes the efficient implementation of a model predictive control (MPC) algorithm for the management of the pallets loaded on the transportation line of a de-manufacturing plant. In order to reduce the computational burden required for the solution of the online optimization problem, and make it compatible with industrial applications, different control and prediction horizons are used. In this way, the complexity of the optimization problem is reduced without significantly affecting the performance of the plant. In addition, a detailed inspection of the transportation line configuration, and the parallelization of the optimization and implementation tasks, allows one to obtain computational times fully comparable to those of simple heuristic rules but with significant improvements in terms of the plant throughput. In the second part of the paper, a fault detection procedure is developed for the identification and isolation of sensors' and actuators' faults. Then, the basic MPC algorithm is modified to obtain a control scheme tolerant to instrumentation faults. Both simulation and experimental results are reported and discussed to show the control performance and the practical applicability of the proposed approach.