SD Builder®: A production rules-based tool for on-line simulation, decision making and discrete process control

https://doi.org/10.1016/j.engappai.2007.05.005Get rights and content

Abstract

The paper presents and discusses a structure and components of a distributed decision-making system for complex discrete systems (CDSs) and processes control. The decision-making nucleus of the system—system dynamics builder (SD Builder)—is described. The SD Builder is based on a hierarchical object-based representation of a CDS and on a production rules-based representation of discrete processes. It allows the using of on-line simulations, state-graph search, expert systems and other methods for decision making. A conceptual prototype of a sawmill in a timber factory illustrates the concepts described in the paper. In a case of a manufacturing system, the decision-making system makes a part of a manufacturing execution system (MES).

Introduction

The on-line decision making while controlling a complex discrete system (CDS) or process is a very important and difficult task. Speaking about manufacturing systems, the on-line decision making is one of the functions of manufacturing execution system (MES). The MES is found on the intermediate level between computer numerically controlled (CNC) machine tools and programmable logic controllers (PLCs) on the lower level and the enterprise requirement planning (ERP) system, customer relation management (CRM) system, warehouse management system (WMS), supply chain management (SCM) system and other enterprise management systems on the higher side. To be efficient, a CDS control system should be based on a model of controlled system and should implement a wide range of decision-making techniques such as simulation, optimization, expert systems (ESs), etc.

This paper talks about one possible approach to the development of complex software systems monitoring, controlling or managing the CDSs. In the approach proposed, the control system has a modular structure; different modules are linked by means of messages or via SQL data bases, or via PLCs, or via XML files. Decision-making modules are developed in SD (system dynamics) Builder—a kind of hybrid of discrete-event simulator, ES, state-graph search block and a data-driven programming tool. The SD Builder uses modified production rules to describe a CDS dynamics and decision-making logic, to define data processing and message generation rules, etc. Modular structure allows an easy integration of third-party modules. A connection to other software systems is assured by dedicated modules transforming the information and control flows to a common internal format—the messages.

The paper is structured as follows. After a brief review of the state of art, we present the general structure of the system proposed. Then, the SD Builder is presented. We highlight the CDS elements and subsystems description, process and activities description, state graph search and the structure of the software. Some details on the application development environment for SD Builder are given. After that, we introduce a demonstrative example—a sawmill in a timber factory. The decision-making problem in this example is to saw logs so to minimize waste taking into account current order portfolio. The structure of the distributed prototype is presented and different modules are discussed.

Section snippets

Literature review

To be proactive in a rapidly changing and highly competitive business environment the control system should have ready access to critical real-time production information. A MES is fitting the gap between the business planning systems and the real-time process control systems (Young, 1995). The on-line decision support systems (DSS) must be adapted and integrated to the requirements of MES functions. The on-line DSS would combine different solving tools to elaborate solutions in reasonable time

The architecture

Our system has a distributed architecture shown in Fig. 1. The different software modules of the system are spread over several computers linked by local area networks and/or global networks. The modules communicate by the way of messages exchange. The message level assures the messages transport and addressing according to the system configuration.

The main feature of the architecture is what it is really destined to on-line decision-making support and articulated around the decision-making

The SD Builder

The SD Builder is essentially a hybrid of a discrete-event simulator, a production rules-based ES and a state-graph search based optimization block. The ideas lying in the basis of the SD Builder are partly inspired from the RAO-method (Emelyanov, and Iassinovski, 1992) and simulator (Artiba et al., 1998). The system dynamics description in SD Builder is based on an object oriented representation of the elements of the CDS and on a description of the state changes using production rules (Artiba

The conceptual prototype of the control system

Let us consider an implementation of the ideas presented above taking as an example a sawmill in a timber factory. At the shop, each log is sawn into blocks to be further processed to planks (boards) or beams. Thus, batches of logs come to the shop inlet and sawn blocks (log pieces) differing by length and diameter are accumulated at its outlet. Initial information for the shop is the portfolio of orders. The orders are of rather small size so the portfolio is changing quickly. An order

Conclusion and perspectives

The fact we use the same modelling language and software tool (SD Builder) for both the factory model and the cutting out application development, allows one to facilitate the testing of different decision-making methods and control system configurations and the evaluation of the impact of different decision-making algorithms to the global performances of the factory without perturbing the real factory functioning.

The conceptual prototype described above runs stably and can operate with the

Acknowledgements

The research and development work presented here is financed by the Simpartners SA Society and by the Walloon Region of Belgium under a project of a refundable financing.

References (12)

There are more references available in the full text version of this article.
View full text
Copyright © 2007 Elsevier Ltd. All rights reserved.