Synthesis of discrete-event controllers based on the signal environment

Citation
H. Marchand et al., Synthesis of discrete-event controllers based on the signal environment, DISCR EVENT, 10(4), 2000, pp. 325-346
Citations number
25
Categorie Soggetti
AI Robotics and Automatic Control
Journal title
DISCRETE EVENT DYNAMIC SYSTEMS-THEORY AND APPLICATIONS
ISSN journal
09246703 → ACNP
Volume
10
Issue
4
Year of publication
2000
Pages
325 - 346
Database
ISI
SICI code
0924-6703(200010)10:4<325:SODCBO>2.0.ZU;2-#
Abstract
In this paper, we present the integration of controller synthesis technique s in the SIGNAL environment through the description of a tool dedicated to the incremental construction of reactive controllers. The plant is specifie d in SIGNAL and the control synthesis is performed on a logical abstraction of this program, named polynomial dynamical system (PDS) over Z/3Z = {-1, 0, + 1}. The control of the plant is performed by restricting the controlla ble input values with respect to the control objectives. These restrictions are obtained by incorporating new algebraic equations into the initial sys tem. This theory sets the basis for the verification and the controller syn thesis tool, SIGNAL. Moreover, we present a tool developed around the SIGNA L environment allowing the visualization of the synthesized controller by a n interactive simulation of the controlled system. In a first stage, the us er specifies in SIGNAL both the physical model and the control objectives t o be ensured. A second stage is performed by the SIGNAL compiler which tran slates the initial SIGNAL program into a PDS, and the control objectives in terms of polynomial relations/operations. The controller is then synthesiz ed using SIGNAL. The result is a controller coded by a polynomial and then by a Ternary Decision Diagram (TDD). Finally, in a third stage, the obtaine d controller and some simulation processes are automatically included in th e initial SIGNAL program. It is then sufficient for the user to compile the resulting SIGNAL program which generates executable code ready for simulat ion. Different academic examples are used to illustrate the application of the tool.