Hierarchical Control and Restart of Flexible Manufacturing Systems

Licentiate thesis, 2006

Product life-cycles are constantly shortening, and the amount of new product types increases, for example in the automotive industry. Growing demands on flexibility and ability to decrease time to market has made it increasingly important to find methods for fast and accurate development of control programs for flexible manufacturing cells. There are several means to achieve this goal. Information present in early steps of the development process, as well as already developed software, should be reused when possible. In addition, formal methods should be used to automatically generate the control functions. To enable information reuse and usage of formal methods, a suitable structure has to be defined for the control program. Also, the methods have to be adapted to industrial work procedures. This thesis presents a methodology for generation of control functions for flexible manufacturing cells. The control program is structured hierarchically and partitioned into two parts. The first part corresponds to hardware and basic cell functions that are essentially the same for different manufacturing cells, and the second part is the actual control function, i.e. the part of the control program that is specific to each cell. By structuring the control information similarly in a hierarchical way, models of the system can be produced and supervisor synthesis can be used to generate the high-level control function. Because of the hierarchical approach, the problem of state space explosion is alleviated. A method for presenting the control function in a simple and easy-to-read way has also been developed. This facilitates the use of the control function by the maintenance personnel handling the manufacturing cell. Furthermore, this thesis contains an approach to solve the restart problem in flexible manufacturing systems. When an error has been detected and solved, it is often a difficult and time-consuming task to resynchronize the controller and the cell, and restart production. Most often the entire restart must be handled manually. The method presented in this thesis analyzes the control function off-line, and synthesizes certain restart states in it, from which a restart is guaranteed to be safe. In contrast to earlier methods, the issues of not increasing the complexity of the cell more than necessary, as well as how to physically move the machines in the cell to the restart states in a safe way, are addressed.