root/library/doc/tutorial/030userguide3.dox @ 704

Revision 704, 3.1 kB (checked in by smidl, 15 years ago)

Controlloop

Line 
1/*!
2\page userguide3 BDM Use - Control and Design of Decision Making Strategy
3
4Bayesian estimation as introduced in Section \ref userguide2 is a special case of Decision Strategy, see. ....
5A more demanding scenario is to design a control strategy that will influence the system operating under uncertainty, this is implemented as scenario \c closedloop.
6
7The function of the \c closedloop is graphically illustrated:
8\dot
9digraph estimation{
10        node [shape=box];
11        "Data Source" -> "Controller 1" [label="data"];
12        "Data Source" -> "Controller 2" [label="data"];
13        "Controller 1" -> "Data Source" [label="actions"];
14        "Controller 2" -> "Data Source" [label="actions"];
15       
16        "Data Source" -> "Result Logger" [label="Simulated\n data"];
17        "Controller 1" -> "Result Logger" [label="Internal \n variables"];
18        "Controller 2" -> "Result Logger" [label="Internal \n variables"];
19}
20\enddot
21The numer of controllers in not limited, however, they can not influence the same variable. Thus, it is limited by the number of available actions.
22
23Here,
24\li Data Source is an object (class DS) providing sequential data, \f$ [d_1, d_2, \ldots d_t] \f$ and accepting input actions \f$[u_{1,t},\ldots u_{m,t}]\f$
25\li Controller is an object (class Controller) performing adaptive control. Typically is has some internal Bayesian Estimator.
26\li Result Logger is an object (class logger) dedicated to storing important data from the experiment.
27
28Since objects  datasource and the logger has already been introduced in section \ref userguide, it remains to introduce
29object \c Controller (bdm::Controller).
30
31\section ug3_theory Design of DM strategy
32The object bdm::Contoller has two principal methods:
33 * redesign, which (re)designs the control strategy - this may be computationally costly operation,
34 * ctrlaction, which evaluates the current decision strategy and returns a numerical value of the action variable.
35 
36At the most general class, we do not prescribe any technique how to design the strategy in \c redesign(). Most of the implemented techniques are, however,
37variants and approximations of dynamic programming (or optimal control), see [].
38
39TODO...
40
41\section ug3_arx_basic Control of ARX models
42
43Autoregressive models has already been introduced in \ref ug_arx_sim and \ref ug2_arx_basic where their simulator and estimator has been presented.
44Control of an ARX model under the qudratic loss function is known as LQG control. This is implemented in object bdm::LQG_ARX.
45
46The following code is from bdmtoolbox/tutorial/userguide/lqg_arx_example.m
47\code
48\endcode
49This is the minimal configuration of an LQG_ARX estimator. Optional elements of bdm::ARX::from_setting() were set using their default values:
50
51\section ug3_arx_distributed Naive distributed control
52
53In the example above, only one controller was influencing the system. We extend this approach to a scenario where more controllers is acting independently.
54
55A trivial example how this can be done is presented in file bdmtoolbox/tutorial/userguide/lqg_arx_distrib_example.m. The code extends previous example as follows:
56\code
57\endcode
58
59*/
Note: See TracBrowser for help on using the browser.