Petri Net Model Based Implementation of Hierarchical and Distributed Control for Discrete Event Robotic Manufacturing Cells

Abstract

Manufacturing systems, where the materials which are handled are mainly composed of discrete entities, for example parts that are machined and/or assembled, are called discrete manufacturing systems. The rapid development of industrial techniques makes the manufacturing systems larger and more complex, in which system control is divided into a hierarchy of control levels: planning, scheduling, coordination and local control (Silva, 1990). At the upper level, the global system is considered and its representation is strongly aggregated. Each low level is a disaggregation of the upper one; the portion of the system considered is smaller, but more details are taken into account. The time horizon of the low level is shorter, and real-time constraints are progressively hard. At the coordination level (shop, cell), the function is to update the state representation of the workshop in real-time, to supervise it, and make real-time decisions. The decision making system has to schedule and synchronize the machine utilizations to decide at each instant what has to be done and on which machine. Frequently, a flexible manufacturing system is structured into manufacturing cells. A cell is an elementary manufacturing system consisting of flexible machine tools (or assembly devices), some local storage facilities for tools and parts and some handling devices such as robots in order to transfer parts and tools between the cell and the global transport system. Elementary manufacturing cells are called workstations. In a manufacturing cell, some devices operate concurrently and cooperatively or interfere to each other. Conventional representation methods such as time chart and state transition graph can not cope with the system like this. So a powerful method which can represent a discrete event system including nondeterministic parallel and concurrent action is desired. As computer technology has been continuing to be upgraded to higher level year by year, the control system architecture has changed from centralized processing to distributed processing in order to reduce the development cost and to improve reliability. In developing distributed processing systems, major difficulties are that adequate expression methods and analyzing techniques for control mechanisms have not sufficiently been established. In the field of manufacturing systems that are typical examples of the event driven system, demands for the automatic control has diversified and the control logic has become extremely complicated. To deal with the complexity, a new methodology on control system design based on the discrete event driven system is necessary.