Robust multivariable predictive control: an application to an industrial test stand

The air conditioning system of an aircraft is used to regulate the cockpit temperature and pressure during flight and usually generates its airflow from the compressor turbine of the jet engine. Testing an air conditioning system requires simulation of the running conditions at ground level. The article concerns the application to such a simulation of α-MPC, a robust extension of the initial multivariable predictive control (MPC) law that improves the disturbance-rejection properties of the closed-loop system, reducing the H^∞-norm of the multivariable sensitivity function with an extra parameter. This augmented algorithm has been chosen to carry out the new tests on the industrial process. Experimental recordings reported here have confirmed significant performance improvement with this new approach relative to the former PID regulation. The article is organized as follows. First we introduce the original MPC. Next we describe the extended α-MPC algorithm and analyze the robustness of the closed-loop system through the H^∞ approach. Then we discuss the methodology of the control design task and describe the experimental test stand, focusing on the software and hardware implementation. Finally, we report the results of the α-MPC control law on the actual test stand. Special attention is given to the comparison with the former control system