Research

year 1997 
author Sangkyun Kim 
Keyword Computer Integrated Manufacturing, Cell Control, Discrete Event Dynamic System Control, Maximally Permissive Adaptive Control 
Abstract For an automated manufacturing system to operate successfully, there should exist software support to integrate and operate the system effectively in addition to hardware support including highly automated machines and reliable communication systems. But, in reality, while the technological achievements in electronics and communications made great strides in the physical integration of manufacturing facilities under the principle of standardization and modularization, software technology necessary to control and manage the system on such physical backgrounds is not so much matured that some people even regards it as one of the main causes of the failure of automatization. Therefore, it is highly important to establish methodologies and software technologies to design and implement control softwares effectively. Such an effort is in line with the standardization and modularization trends of the manufacturing sector.

In this thesis, we discuss on a methodology to support synthesis of a control logic and implementation of a control software based on the characteristics of an automated manufacturing cell as a discrete-event dynamic system. The basic strategy for generating control logic we adopt is that of systhesis, which ease the automatic reconstruction of the control logic against changes in system structure. For these, we specifically discussed on methods for modeling target manufacturing systems, tasks, and control requirements, for synthesizing maximally permissive control logic satisfying control requirements, and for implementing cell control software which incorporates the synthesized control logic.

Control model consists of plant model, task model, and control requirements model. First, we define resource models to describe the behavior of component resources for a manufacturing cell. Essentially the overall behavior of the cell which is characterized by the interactions among the components are modeled by the command-response sequences among them. Similarly the behavior of a task which must be processed within the cell is also modeled by the command-response sequences which can occur while processing the task within the cell. We use finite state automata called CoreNet to model such interactions. A CoreNet describes the evolutionary pattern of the state transition of a system according to the exchanges of command-response messages. Control requirements involve rules which should be imposed on the system to restrict the overall behavior of it to a certain bound. Insisting mutual exclusion principle on sharable resources, preventing conflicts and deadlocks, and keeping priorities are examples of control requirements. We summarize the possible requirements and discusses on appropriate modeling methods such as state-constraints or events-constraints.

We adopt the supervisory control approach under the framework of Ramadge and Wonham to synthesize control logic for a manufacturing cell. Specifically, we propose and use an improved version of the variant lookahead policy to extract the maximally permissive control logic guaranteeing satisfaction of the control requirements from the control models. The proposed policy is shown to be equivalent to that of Ramadge and Wonham. Moreover, we propose a maximally permissive adaptive control policy which takes into account changes in the set of concurrent tasks within the cell due to newly input tasks. Under the policy we can show that the successive control rules are consistent through the analysis of a dynamic equation describing the evolution of control models and control rules.

Lastly, we propose an architecture for the cell control software which embeds the proposed control policies and develop an object-oriented class library to support the rapid implementation of the event-driven controller. The operation of the library is demonstrated using an example FMC. 
c PhD 

Download :