یک روش تقویت شبکه پتری برای بازیابی خطا در کنترل سیستم های تولید

The construction of error recovery Petri subnets and similar representations have received considerable attention in the literature. Previous work has presented a multi-agent system representing various levels of control in a reconfigurable architecture. Agents pertaining to production, mediation, and error recovery within such an architecture were considered. Our focus here is on the workstation level of a hierarchy where the workstation has the capability for recovery from physical errors. The implications of error recovery tasks from the perspective of control are also discussed. The approach is based on integrating Petri subnet models within a general Petri Net model for a manufacturing system environment. In essence, the error recovery plan consists of a trajectory (Petri subnet) having the detailed recovery steps that are then incorporated into the workstation control logic. The logic is based on a Timed Petri Net model of the total production system. The Petri subset models consist of a sequence of steps required to reinstate the system back to a normal state. Once generated, the recovery subnet is incorporated into the Petri Net model of the original expected (error-free) model. Petri Net augmentations pertaining to various issues are discussed in detail throughout the paper. Issues include the implication of generated error recovery trajectories in the production activities, linking of production activity Net and the error recovery subnet, potential deadlocks, the role of resources, and part handling.

In this paper, we focus on the workstation level of a hierarchical manufacturing system. A workstation is typically a set of parallel machines linked by material handling devices that perform one or more manufacturing and assembly operations. The workstation controller is the entity responsible for the coordination, execution and regulation of the activities at the physical workstation. The workstation controller receives a higher level command, generally form a higher level controller that issues a set of operations to be performed by the workstation with desired start and finish times. The workstation controller decomposes such a command into a lower level set of coordinated activities. In addition to executing activities, the workstation controller should also provide a reactive and adaptive response to errors and other disturbances [1]. Petri Nets have been successfully used for modeling and control the dynamics of flexible manufacturing systems. Several modeling approaches based on Petri Nets that include those of [1], [2] and [3] have been proposed. Generally, the operations required on a part are modeled with combinations of places and transitions. The movement of tokens throughout the Net models the execution of the required operations. In this paper, we follow the modeling approach previously presented by Odrey and Ma [1] and Mejia and Odrey [4]. The Petri Net formalism can handle the complexities of highly detailed activities of a manufacturing workstation such as parallel machines, buffers of finite capacity, dual resources (multiple resources required simultaneously on one operation), alternative routings, and material handling devices to name a few.

This paper has discussed the issues of incorporating recovery trajectories into the control logic of a workstation control agent. A contribution is the real-time error treatment which involves the addition and deletion of recovery paths from the control logic. In terms of Petri Nets, the recovery activities are modeled with a Petri subnet that is attached to the existing activities Net. In this research, three types of recovery actions, namely input conditioning, backward recovery and forward error recovery were investigated from the perspective of the workstation level in a hierarchical intelligence-based architecture. Since the recovery actions were previously developed for very low levels of control (i.e. equipment level), modifications have been proposed to the three types of recovery action features that characterize the workstation level. Such features are the preservation of the level of detail workstation commands and the handling of resource allocation during the execution of recovery actions. The incorporation of recovery subnets at the workstation level brings two undesirable situations: