Johan Richardsson

Coordination of Operations by Relation Extraction for Manufacturing Cell Controllers

A method for generation of the control function for flexible manufacturing cells is presented in this paper. The control function is separated from the rest of the programmable logic control (PLC) program, and partitioned into a high-level part handling the operation sequence and a low-level part defining the execution process of the operations. The program structure enables high-level supervisor synthesis, which alleviates the state-space explosion problem. Information present in earlier steps of the development process is reused and processed, to automatically generate the automata needed for supervisor synthesis. An algorithm for generating automata, from the low-level safety requirements that restrict the high-level behavior, is presented. Algorithms are also presented for extracting the relations between the operations defining the work in the cell, from the synthesized supervisor. These relations give an easy-to-read representation of the control function that makes it interpretable by machine operators and maintenance personnel, an important feature to make the method applicable in an industrial setting. Hence, the control function generated by our method combines the benefits of a traditional supervisor, namely optimality and flexibility, with simplicity and clarity.

Automatic generation of PLC programs for control of flexible manufacturing cells

Shortened product life-cycles decrease the output rate of manufacturing systems. Offline verification of the control systems promises to increase the output. However, to make offline verification possible some major improvements of the current development process of manufacturing systems are needed. Information handling and development of control programs based on information reuse are the two most important improvement areas. This paper deals with one problem up the many connected to enabling offline verification. A general control program structure, adapted to information reuse and formal mathematical verification methods, is needed. This paper proposes a program structure that makes it possible to generate the PLC-program out of information that already exists in the development process of a manufacturing system. In order to further increase the ability of offline verification the proposed program structure is adapted to import information processed by formal mathematical methods.

Implementing a Control System Framework for Automatic Generation of Manufacturing Cell Controllers

Quickly adapting the manufacturing system to the production of new or modified products is critical for manufacturers in order to stay competitive. For flexible manufacturing systems this typically implies modifications of the control programs. In previous work a framework for automatic generation of cell controllers has been developed. In this paper an implementation of the framework is presented. Important properties of the presented implementation are: the information from earlier design phases is reused; automatic code generation for faster development and reduced implementation errors; the supervisory control theory is used to generate control functions that are correct by construction; object oriented principles are used in order to allow the reuse of existing library functions. The implementation is generic in the sense that it may generate control programs for a number of target platforms, but in this paper the focus is on generating a control program for the Java platform. An industrial example of a reconfigurable manufacturing cell has been used in the implementation process and shows that the framework is feasible for large manufacturing systems.

Reuse of information as a base for development and verification of control programs for flexible manufacturing cells

Shortened product life-cycles decreases the output rate of manufacturing systems as the introduction of new products into the manufacturing system becomes more frequent. Improvements of the development process of manufacturing systems are needed to increase the output. Information handling and development of control programs based on information reuse are two of the most important improvement areas. These areas, among other things, can be a support for offline verification, which promises to directly increase the output rate due to shortening product introduction times. This paper deals with two problems of the many connected to enabling offline verification. First, a general control program structure, adapted to information reuse, is needed and secondly, the information necessary to generate the control programs needs to be defined. A method is proposed where information from the mechanical design of a cell, from the product, and from manual simulation are reused and automatically converted into control programs that schedule the work in a collision-free, deadlock-free and time-optimized way. The correctness of the generated programs is guaranteed by use of formal methods, simulation and an uncorrupted conversion of specifications into control programs.

Design of control programs for efficient handling of errors in flexible manufacturing cells

Insufficient indication of errors is a problem in many manufacturing systems. Lack of support for resynchronization of the cell and its control system is another, less obvious problem. A third problem connected to errors in manufacturing cells is lack of support for manual control. In order to resolve an error situation manual control of the cell is often required. The problem is that some of the manual operations may be blocked due to machine protection. When an operator is to execute a blocked operation the only response is that nothing happens. This paper proposes a method where control programs with integrated functions for error detection, resynchronization, and support for manual control are generated out of information that already exists in the development process of a manufacturing system.

Reliable Control of Complex Manufacturing Cells

Lack of tools for analyzing the control of complex manufacturing cells is a problem which consequences are functional errors and in some cases a lower throughput than expected. A typical property of complex cells is processing of several products at the same time. This paper presents a model for automatic generation of control programs based on the output of a previously presented algorithm for calculation of the control of multi product cells.

Minimization of Expected Cycle Time in Manufacturing Cells with Uncontrollable Behavior

Finding optimal working sequences for flexible manufacturing systems (FMS) is a challenging task, especially if uncontrollable events, such as machine breakdowns, can occur in these systems. In this paper, we continue our previous work within the area of FMS scheduling by extending our methods to uncontrollable systems. For this sake, the ideas of probabilistic finite automata are combined with the modeling formalism of extended finite automata. Next, an A*-based scheduling algorithm, minimizing the expectation value of the FMS cycle time is designed.