Yves Piguet

Interactive learning modules for PID control [Lecture Notes]

This article describes three interactive learning modules that are designed to develop intuition as well as a working knowledge of proportional-integral-derivative (PID) control. These three modules comprise a package called interactive learning modules for PID (ILM-PID). By illustrating concepts such as tuning, robustness, loop shaping, and antiwindup, ILM-PID can be used for demonstrations, exercises, and self-study. In this article we use interactive graphics to learn PID control and develop intuition.

On compensation of nonminimum-phase zeros

Abstract Nonminimum-phase (NMP) zeros may cause severe transients, like undershoots and oscillations, in a step response. The discrete-time case is studied in which a design of a set-point feedforward by placement of additional zeros is proposed. Optimal placements of zeros are given that minimize the maximal undershoot when the system has one real NMP zero. A system having two complex NMP zeros can be compensated such that a monotonie response is guaranteed, even in presence of model uncertainty.

A minimax approach for multi-objective controller design using multiple models

A minimax approach for multi-objective controller design is proposed, in which structured uncertainty is characterized by multiple discrete-time SISO models. Typical engineering objectives are optimized for all models, such as bounds on different sensitivity functions and time-domain responses. The approach is illustrated by improving the best performing controller of a flexible arm benchmark example.

Understanding PID design through interactive tools

Hundreds of PID design methods are available in literature. Many of them are very similar and sometimes it is not straightforward to understand their purposes. This paper presents an interactive software tool to help in the study and understanding of several wellknown PID tuning rules. Frequency- and time-domain responses are analyzed in order to show the robustness and performance properties of each method. Furthermore, a free tuning option is available to provide comparisons with user-defined rules or other existing tuning methods. A wide range of stable process models are included in the tool, where model-reduction techniques are applied for high-order processes to obtain simple models for design purposes.

MOOLs for MOOCs: A first edX scalable implementation

Deploying remote laboratories at a large scale is the next challenge in remote experimentation. It is referred as Massive Open Online Labs (MOOLs). Being able to sustain a massive number of users accessing a specific resource concurrently is a complex task. This paper presents technical solutions, currently deployed, to implement such a remote lab in a MOOC. The control system lab deployed in edX, where more that 200 students access concurrently a farm of 20 electrical drives, is used as an example. Gamification mechanisms which accommodate even more students are proposed.

INTERACTIVE LEARNING MODULES FOR PID CONTROL

Abstract This paper describes a collection of interactive learning modules for PID control based on the graphical spread sheet metaphor. The modules are designed to speed-up learning and to enhance understanding of the behavior of loops with PID controllers. The modules are implemented in Sysquake, a Matlab dialect with strong support of interaction. Executable versions of the modules for Mac and PC are available on the web.

Multi-Model Weighted Pole-Placement Design

A CAD method based on approximate pole placement for a set of models is developed. Dominant poles for all models are simultaneously placed approximately in the least-squares sense. Weighting of the poles allows the constraint of the pole-placement error of dominant poles vis-a-vis non-dominant poles. It is demonstrated that a large number of specifications can be taken into account while keeping the complexity of the controller low. The controller obtained gives robust performance in the sense that all specifications are met for all models in the set. A benchmark example is investigated to illustrate the use and the effectiveness of the proposed method.

Interactive Learning Module: Basic Modelling and Identification Concepts

This paper describes an interactive tool focused on teaching and learning basic concepts about modelling and identification. Modelling and identification play fundamental roles in the field of automatic control, since it is not possible to perform a control system design without having a model describing the dynamics of the process to be controlled. The tool presented in this work is dedicated to describe the most important aspects of basic modelling centralized on first and second order systems with dead-time and zero effects.

INTERACTIVE APPLICATIONS IN A MANDATORY CONTROL COURSE

Abstract This paper describes how a mandatory control course is enhanced with interactive applications. Basic concepts of automatic control, different ways to assess performance and robustness of a controlled system, and limits of closed-loop systems are illustrated with graphics which can be manipulated directly with the mouse, letting the student acquire a deeper and broader understanding of theory. The applications, implemented with Sysquake as stand-alone programs, offer a simple user interface while supporting a wide range of problems.

Feedforward Control Concepts through Interactive Tools

This paper presents an interactive software tool for basic and advanced concepts of feedforward control. The ideal compensator is analyzed to show how the effects of the load disturbances are eliminated from the process output. Furthermore, it is also shown that the ideal compensator is seldom realizable, since the compensator may be non-causal, it may be unstable, it may have infinite high-frequency gain because of derivative action, and it may require a more complicated structure than what is available. The drawback of not taking the feedback controller into account during the typical open-loop feedforward design is also studied. The proposed software tool helps to evaluate all these situations in an interactive way, enabling users to find possible design solutions to these problems. (Less)