Restarting Manufacturing Systems; Restart States and Restartability

A method for restart after an error in a manufacturing system is introduced. The method is able to restart systems even after nonforeseen errors that cannot be planned for, and the online part of the restart method does not require use of more powerful computers than a standard Programmable Logic Controller. It is shown what properties the control function must have to ensure that there is at least one restart state for each controller state. Sufficient conditions to guarantee the possibility to restart a system regardless of where an error occurs are given, along with indications on how the system could otherwise be rebuilt to be restartable. Note to Practitioners-The motivation for this work has been the difficulties of resuming normal production in a manufacturing system after the (often manual) correction of an error. These difficulties are especially pronounced for systems where errors cannot be foreseen, a common situation, e. g., in the automotive industry. Existing restart methods often assume extensive knowledge of the possible errors and/or require more computational resources than are typically available on the shop floor. Therefore, the impact on industrial practice has to date been limited. We propose a restart method where the above mentioned problems are avoided. The control program is analyzed offline to find the states from where a restart is possible. If paths to these states are offline synthesized and stored in the controller, a simple online method executable by any industrial control system can then restart the system. This facilitates the work of the maintenance personnel who no longer need to check online the safety of each path used in the restart. The offline analysis also tells if the system can be restarted by the proposed method from all possible states or if certain situations require special treatment. If all situations cannot be handled by our method, indications of suitable modifications of the system are given. In future research, we will address the task of restarting a number of machines that have simultaneously broken down. We will also study how to model the system and structure the work to facilitate restart.