Guillermo Bejarano

Design and Application of Suboptimal Mixed

This brief tackles the problem of designing suboptimal H2/H∞ controllers for linear networked control systems (NCS) subject to time-varying delays and packet dropouts. The formulation provides state feedback NCS controllers allowing to tradeoff performance and disturbance rejection. The control objective consists in designing an H2 suboptimal control minimizing a quadratic performance index, with a disturbance rejection constraint (H∞ constraint). To characterize the network, only the lower and upper bounds for the delay, as well as the maximum number of consecutive dropouts are required. The approach relies on the formulation of the problem in terms of the minimization of a single scalar parameter, that can be cast as a standard linear matrix inequality (LMI) problem, yielding a suboptimal cost-guaranteed solution. As a difference from previous works, the solution provided is independent of initial conditions. Stability and robustness properties of the proposed controller are theoretically demonstrated and tested on an experimental testbed consisting in the stabilization of a robot arm in the proximities of the unstable upright position. The application shows good performance and disturbance rejection capabilities even for stringent network conditions.

H_{2}/H_{\infty}

This paper describes the design, implementation and automation of a two-stage, two-load-demand experimental refrigeration plant. The facility is fully configurable, since cycles with one or two compression stages, and one or two load demands can be set up. Detailed description of the experimental facility concerning its physical components, actuators and sensors is addressed, as well as low-level control and communication bus issues. The high-level control environment and its communication with the low-level controller are also approached. Furthermore, a well-known in industry decentralised control strategy is applied to a one-stage, two-load-demand refrigeration cycle, analysing some preliminary control results from an energy efficiency point of view.

Controllers for Networked Control Systems

Abstract This paper describes the controllability analysis and the design of a multivariable robust controller for a one-stage refrigeration cycle, which is the most used system for cooling issues and whose energy management and control are key factors. The controlled variables are the superheating degree of refrigerant at the evaporator outlet and the outlet temperature of evaporator secondary flux, whereas the control variables are the electronic expansion valve opening and the compressor speed. The system is considered therefore as a MIMO one. Up to eight different operating points are identified in order to design the multivariable H ∞ controller based on the S / KS / T Mixed Sensitivity Problem. Simulation results comparing this controller to a decentralised PID and to a multivariable predictive one are shown, proving a better performance of the robust controller as well as accomplishment with the controllability analysis conclusions.

Design, automation and control of a two-stage, two-load-demand experimental refrigeration plant

This paper describes the steady-state parameter identification process of the components which constitute a vapour-compression refrigeration cycle. Particularly relevant is the heat exchanger steady-state parameter identification due to the refrigerant phase change which occurs during a typical cycle operation. Experimental results that prove the validity of the considered assumptions are shown and discussed.

Controllability analysis and robust control of a one-stage refrigeration system ☆

This paper proposes a suboptimal control solution for the L2-gain disturbance rejection problem for networked control systems. The problem, usually referred as mixed H2/H∞, aims at designing a linear state feedback stabilizing controller minimizing a quadratic cost functional subject to a L2-gain disturbance rejection constraint. The formulation deals with time-varying delays and dropouts in both, sensor-to-controller and controller-to-actuator paths, with the only assumption of known minimum and maximum bounds of the round trip delay, and the maximum number of consecutive dropouts. The solution is based on a Lyapunov-Krasovskii approach and is formulated as a Nonlinear Matrix Inequality (NLMI) problem. This problem is cast into more a treatable LMI-based minimization problem, for which a well-known optimization algorithm is provided. Experimental results on a networked controlled direct-drive 2 degree-of-freedom (dof) robot manipulator are provided to verify the performance of the methodology.

Steady-state parameter estimation of an experimental vapour compression refrigeration plant

This paper describes the analysis and robust control of a one-stage refrigeration cycle. Using the compressor speed and the expansion valve opening as control variables, the outlet temperature of evaporator secondary flux and the superheating degree of refrigerant at evaporator outlet are controlled. The main dynamics of the system are identified at various operating points and a controllability analysis is carried out in order to highlight existing constraints in its closed-loop performance. A multivariable robust H∞ controller is designed and applied, using the S/KS/T Mixed Sensitivity Problem approach. Simulation results prove that the robust controller has better performance than a decentralized PID controller, agreeing with the controllability analysis conclusions and the control design specifications.