Designing Your Own Control System
If you want to exercise your math skills and remind yourself of all the things that have to happen in a control system, this online article is a concise guide to design concepts.
Peter Galan, Ph.D., JDS Uniphase
Editors note: The article Driving Plant Optimization with Advanced Process Control (in this issue) discusses when and how companies might want to create their own process models and control strategies. This is not for the faint of heart since it involves some highly sophisticated math processes that will likely test your ability to recall your academic training. If you want to test your chops, here’s an article from Peter Galan, Ph.D., control engineer at JDS Uniphase, a manufacturer of fiber optic and other broadband communication equipment. His experience includes 35 years working with process control systems, robotics, and optical communication.
The Quick Engineer’s Guide to Control System Design gives a very condensed (about 25 pages) description of what has to go on in an industrial process control system, beginning with the most basic mathematical foundations to PID and more sophisticated elements of advanced control, including feed-forward control, adaptive control, optimization, fuzzy logic, and more.
An excerpt from the introduction:
There is no need to stress the importance of control systems in any sphere of our life. We take them as something completely natural, but the control system designers are sometimes getting headaches while they succeed in their job, even though the basic principle of the automatic control seems to be so simple.
Figure 1 shows the main components and variables of a generic closed-loop control system. It consists of the controlled system or plant and the controller. The purpose of the controller is to maintain output variable, y(t), at the level determined by reference input. This is accomplished by the subtraction of output variable, y(t) from reference input, r(t). The result is a regulation error, e(t), which is input information for the controller. The controller calculates actuating variable, u(t), which is the input to the controlled system.
Today’s control systems are almost exclusively implemented as digital systems. The graphic at left represents the block diagram of a digital control system. Two new components are D/A and A/D converters. Only two signals, u(t) and y(t) are of the continuous-time character. The remaining signals are all in the discrete form. So, the entire signal processing including the comparison of r[n] and y[n] can be accomplished by CPU/DSP.
You might remember several articles—just in this magazine—dealing with the control system (PID) tuning. And the tuning is just one phase of the control system design. We are not going to touch even such elementary features of control systems, like stability, robustness, etc. Of course, the control theory is a subject for several semesters of study at universities. However, you can get, at least, some basic ideas of the control system design.