10
\$\begingroup\$

As electrical engineers, we often come across the terms "classic control" & "modern control"

What does classic mean here? Is it meaning "old" or of "past time"?

\$\endgroup\$
5
  • 2
    \$\begingroup\$ I think "classic" means analog, "modern" digital. \$\endgroup\$
    – tlfong01
    Dec 12, 2021 at 10:52
  • 3
    \$\begingroup\$ A little more context would help \$\endgroup\$
    – Jasen
    Dec 12, 2021 at 11:29
  • 4
    \$\begingroup\$ @tlfong01 That's not correct. Classical control systems can be implemented digitally, and I'm guessing modern control systems can be implemented analogically, too. \$\endgroup\$ Dec 12, 2021 at 13:50
  • 1
    \$\begingroup\$ @Tanner Swett, Many thanks for pointing out my silly wild guessing mistake. My apologies to everybody. I skimmed the excellent answer below and got a rough idea of what I do not know and very likely never able to appreciate even the essence of the stuff. I only know that Control theory, even linear, uses advanced mathematics much beyond my capabilities. I remember I once tried to learn things like eigenvalues, state space equations, Bode plot and Nyquist plots etc but gave up after some hard effort. \$\endgroup\$
    – tlfong01
    Dec 12, 2021 at 14:51
  • \$\begingroup\$ For an example of an analog modern MIMO see the original Librascope LC-# weight balancers at librascopememories.com/Photos/Pages/Products.html \$\endgroup\$ Dec 13, 2021 at 20:10

1 Answer 1

19
\$\begingroup\$

This is from Linear Systems Control by Hendricks, Jannerup and Sørensen pp. 5-8: -

Classical Control

"The Classical Control period was characterized by a concentration on single loop, Single Input, Single Output feedback (SISO) systems designed with the theoretical tools developed during and just after the Second World War.

For the most part these could be applied only to linear time invariant systems. The main underlying concept is that closed loop characteristics of a system can be determined uniquely given the open loop properties of the system. This includes the important disturbance rejection and steady state error properties of the feedback system

The theoretical tools were those developed by Nyquist, Bode, Evans and Nichols earlier and the connection between these methods was clarified and extended. Performance was assessed in terms of bandwidth, gain and phase margin or rise time, percentage overshoot, steady state error, resonances and damping."

Modern Control

"The large number of states in MIMO state variable systems and the possibly large number of feedback loops which might exist in a closed loop system make it necessary to consider how decisions might be made about the feedback levels in the different loops. It has turned out that this is not a simple question and it is difficult to impossible in fact to make any reasonable, balanced statement about what might be required. (...)

In 1960 and 1961 a significant set of breakthroughs were made generally available with the publication of four papers by Rudolf Kalman and co-workers. These papers dealt with (1) the optimal control of discrete time systems, (2) the design equations for the Linear Quadratic Regulator (LQR), (3) optimal filtering and estimation theory for discrete time systems and (4) the continuous time Kalman filter."

Summary

Classical control is generally about designing controllers for single input - single output systems by pole/zero placements. You look at Bode Plots and Nyquist Plots, and investigate phase/gain margin, bandwidth, etc. and assess if you are "on the right track".

Modern control is generally used for multiple input - multiple output systems by eigenvalue placements. You create a linear model of your system, set up a state space equation and define controller gains (for a specific control architecture, optimal control for example) to make sure you are meeting certain requirements (rise time, overshoot etc.)

\$\endgroup\$
4
  • 1
    \$\begingroup\$ 'Tis a shame you didn't mention Richard Bellman and dynamic programming. :-( \$\endgroup\$ Dec 13, 2021 at 4:07
  • 1
    \$\begingroup\$ B.J. Lurie and P.J. Enright: "Classical feedback control" (Marcel Dekker Inc., 2000). In this book I find Chapter 2 "Feedforward, Multiloop and MIMO Systems". So I think, there is not a strict and commonly agreed distinction between "classical" and "modern". \$\endgroup\$
    – LvW
    Dec 13, 2021 at 8:49
  • \$\begingroup\$ In about 1971 I had a control systems lecturer who used on occasion to use examples of stabilised platforms - specifically gun control on Royal Navy warships in WW2 - and would often apologise for using such non-peaceful examples but note that that is where he first learned and applied his art. I doubt that anyone had any concerns. \$\endgroup\$
    – Russell McMahon
    Dec 13, 2021 at 9:49
  • 2
    \$\begingroup\$ Good answer, but modern control isn't necessarily MIMO. Modern control theory is certainly the only robust way to solve a MIMO situation. But it's also key to solving SISO problems where the system you're controlling is not perfectly modelled, where its characteristics change over time, or where the feedback signal is buried under noise. That's why the Apollo lander needed a Kalman filter. \$\endgroup\$
    – Graham
    Dec 13, 2021 at 11:28

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.