Table of Contents
Previous Chapter
Table of Contents
Previous Chapter
An advanced manufacturing system usually contains several manufacturing cells and their associated material handling systems. A manufacturing cell, in general, contains a cell controller and several devices, such as computer numerical controlled (CNC) machines and robots. Each device usually has a device controller which can control the motion of the device and communicate with other devices and the cell controller. A device controller normally has a central processing unit (CPU), memory and a user interface. A device controller is often also equipped with some communication interface, such as RS-232C and discrete inputs/outputs (DIO). All the devices in a manufacturing cell need to be coordinated and controlled by the cell controller. Due to their local processing capabilities, these devices are referred to as intelligent manufacturing devices (IMDs) in this thesis.
The material handling systems associated with manufacturing cells can include conveyors, automated guided vehicles (AGVs), and automatic storage and retrieval systems (AS/RSs), etc. It is common for these material handling systems to be equipped with controllers, such as PLC (Programmable Logic Controller), Computers. These controllers also have a CPU, memory, user interface, and communication interface. Thus, these material handling systems are also referred to as IMDs in this thesis.
To a certain extent, most device controllers can be programmed via user programs, be controlled either from a host computer or by an operator, or be coordinated with other devices to perform a variety of tasks. However, an IMD's capability to be controlled by a host computer or by another IMD is often limited. These limitations can arise either due to limitations imposed by the CPU(s), memory, communication interface or by vendor software in the device controller. Limitations may also arise because the device controller is so busy in performing motion control that it doesn't have capability to communicate with a host computer or other devices in real-time.
In general, the manufacturing cells and their associated material handling systems can communicate with each other through networks. In a distributed manufacturing system, it is possible that the control decisions may be made at one location and be sent to other locations for execution. In order to achieve system control objectives, control commands may be delivered from the controller where the decisions are made to the device where the decisions are carried out via communication media. For example, a cell controller makes decisions and transfers commands to the controller of the IMDs via a communication media.
The process of sending a command from a host computer to the cell controller, processing the command in the cell controller, sending the command to the IMD through the interface in the cell controller, reading the command by the hardware and software in the IMD, and processing and executing the command by the IMD, is called "message passing" in this research. Cell control is attained when a set of messages can be transmitted through the complex collection of software and hardware components from where they originate to their destinations. The capabilities of the individual controllers to receive messages determine the level of cell control that can be realized.
In some cases, transferring a command from a cell controller to an IMD may depend upon the state of the IMD. For example, a new robot program may not be downloadable to a robot controller while the robot is executing a command. The state of an IMD changes in real-time. Thus, time is also an important issue in message passing.
In general, devices are selected from commercial vendors based mainly on mechanical and process specifications. For a system integrator, the limited message passing capabilities of IMD device controllers often emerge as an unexpected problem during implementation. To rectify these message passing problems may necessitate the creation of additional low-level software modules and/or the modification of hardware. Both require special expertise, and may require long development times and significant expense, causing costly delays in start up.
Due to the aggregated nature of the system requirements, the non-standardized interfaces, and the complex software structures involved in complex manufacturing systems, it is often difficult to identify the communication/control pitfalls inherent in a design configuration before implementation. Therefore, it is desirable to have a systematic method for analyzing the message passing capabilities of a manufacturing system during the design stage.
In order for a manufacturing system to perform its tasks, the software and hardware components in the system must be able to transmit control commands (referred to as messages in this thesis) to their destinations. Message reachability is defined as the ability for a specific type of message to reach its destination(s) in an acceptable time interval. The origin is where (and when) the message is created in the manufacturing system, and the destination is where the message must go to trigger the desired action. The time it takes for a given message to transmit from origin to destination(s) is referred to as the transmission delay.
In general, system integrators need to select IMDs from commercial vendors in order to configure a manufacturing system to achieve required process and control objectives. However, commercial IMDs often have a very limited ability to transmit/receive messages. It is crucial to identify the message reachability of a manufacturing system early in the system design. The system integrators usually need to reference commercially supplied equipment manuals in order to determine the capability of an IMD to pass messages. In general, these manuals are not documented from a message passing perspective and different terminologies are used by different vendors. Therefore, it is a challenging task for a system integrator to a) identify the message passing capabilities of a device, b) to analyze the message passing capabilities of the manufacturing system, and c) to determine message reachability of the system.
Currently, there are no systematic procedures, representation methods, or analysis tools that can be used to determine message passing capability and identify message reachability of a manufacturing system.
The objectives of this dissertation are to systematically address this problem by:
The problem domain for this dissertation is the distributed manufacturing control system. A distributed manufacturing control system is a control system in which decisions are made in various locations of the system, such as machine tool controllers, robot controllers, the material handling system, and cell and system computers. Randomness within each component is not considered. Therefore, the message reachability analysis in a system containing local area network may require some extensions.
This thesis proposes a technique to model message passing in manufacturing systems. By using this modelling technique, a system integrator can represent and analyze the message passing capability of a system.
The result of this modelling technique is termed an object-oriented connection model. A set of templates were developed to build an object oriented connection model and will be called reference models. The references models capture both the commonalities of the real world components and the interactions of these components. The properties of the reference models regarding the message passing and the delay times are identified.
An object oriented connection model can be represented mathematically and graphically. The graphic representation of this object-oriented connection model is called a graphic model. Guidelines for constructing this graphic model are provided along with an example.
The characteristics of message passing path and delay times of the object oriented connection model are identified. Algorithms are provided to analyze model and to determine the existence of message paths and the upper and lower bounds on the transmission delay for a given message.
The chapters that follow present the modelling technique, how to use the technique to generate the models, and analysis tools for determining message reachability.
Chapter 2 provides an introduction to message passing in manufacturing systems and a survey of related modelling techniques. Since the message passing capability of an IMD has not been explicitly and systematically identified, several examples are used to illustrate the various types of message passing and their embedded characteristics in manufacturing systems. What is learned about the characteristics of message passing provides insights for developing the modelling technique. Literature related to the modelling approach proposed in this thesis are surveyed. Reasons for selecting the modelling technique proposed in this thesis are explained.
Chapter 3 presents a modelling technique to represent message passing in manufacturing systems. This modelling technique integrates IDEF0 and Petri net techniques into an object-oriented approach in which objects are the basic components. Since many varieties of object-oriented definitions appear in the literature, the object-oriented terminologies used in this thesis are defined. A color timed Petri net is used to represent the behavior of objects. The static structure and dynamic behavior of the Petri net is presented. The token propagation on the Petri net is also defined. Graphic symbols for representing the Petri nets and objects are illustrated. The integration scheme for these modelling technique is provided. Chapter 4 presents reference models for constructing models for message passing. The reference models consist of a set of templates that capture the commonalities of the real world components and the interactions among these components. The templates provide mathematical and graphic representations. The properties of the reference model regarding to the message passing are identified. The application of the modelling technique and reference models to a manufacturing system will result in what will be termed as an object-oriented connection model.
A set of basic Petri nets representing typical message passing are provided to assist the system integrator in constructing the model. In addition, the time characteristics of these typical Petri nets are identified to assist the analysis presented in a later chapter.
Some properties, such as object message liveness, will be identified and proved. These properties may provide some insights toward a good design of a manufacturing system.
Chapter 5 presents applications of the modelling technique and reference models. Guidelines for constructing an object oriented connection model will also be presented. The graphic representation of this object-oriented connection model is called a graphic model. For a given system, this graphic model can be used to represent the system and to facilitate analyzing message passing. This graphic model consists of three types of diagrams: the hierarchy diagram, the message diagram, and the behavior diagram. An example of the use of the guidelines to construct a graphic model is presented.
Chapter 6 presents a message reachability analysis to allow the system integrator to determine the lower and upper bounds of time delay for the transmission of a message. Algorithms are developed to determine the lower and upper bounds on the transmission delay of a given message. In order to for system integrators to understand the algorithms, time delay properties of the object oriented connection model are identified and explained with examples.
Chapter 7 presents the results of this research, conclusions, and areas of future research.
Next Chapter