A contribution to the development of a HMS simulation tool and proposition of a meta-model for holonic control

Abstract

The present context and tendencies in modern production system, as mass customization, requires improvements with respect to the agility of the production organizations. In this sense, agile approaches have been proposed, such as the holonic approach. In Holonic Manufacturing System (HMS) the production entities, as resources and products, are envisaged with a type of intelligence. These smart-entities are called holons (HLs) whose intelligence is related to their autonomy and collaboration skills. The HMS also comprises a Holonic Control (HC) that must properly organize holon collaborations in order to become agile. Actually, HMS development requires engineering tools for design and testing. In this doctoral thesis, a meta-model for HC is proposed, whose instances are simulated within a particular tool called ANALYTICE II. This tool presents a clear separation between high-level control and emulated resources. Firstly, before the proposition of the HC meta-model, the resource holonification is proposed in this environment. Each Resource-HL is obtained by means of a virtual resource that provides data and services of an emulated-resource at a high level of control. Subsequently, the meta-model for HC over Resource-HLs following a process-driven production approach is proposed. The essence of the solution is based on Rule Base System (RBS) concepts being the causal relations of control dealt with by entities called Rules. The inference process in this RBS is realized through collaborations based upon notifications. The Resource-HLs notify the Rules about factual knowledge with respect to their states. Each Rule that is notified deliberates about the proper moment to execute some control action, as the coordination of a set of Resource-HLs, using causal knowledge. The inference occurs within a notification chain enabled by a group of Resource-HL agents and Rule agents. This kind of inference can be expected to provide advantages for the HC, such as high reactivity and entity decoupling. Furthermore, it allows for the creation of co-operative mechanisms for dealing with determinism and conflict issues. Moreover, this approach of rule-oriented control allows for coherent control implementation and expression. The control mechanisms emerge based on causal control knowledge expressed by experts in the Rules. Experts are exclusively concerned with the proper control knowledge needed for exploiting system flexibilities in order to increase system agility. Furthermore, some experts could even be artificial agents automatically dealing with knowledge of the Rules. Briefly, this process-driven HC solution concomitantly treats a set of control issues while also being a self-contained and open solution. Indeed, the solution openness allows its interpretation as a product-driven solution. The product-driven control is a tendency to reach agility by the decoupling of production demands and execution via entities like Smart-Product-HLs. Each Smart-Product-HL is concerned with a specific customized production order. The Smart-Product-HLs, with certain autonomy, use Resource-HLs to reach their production goals. In the meta-model interpretation, their interactions are organized by Rules for Resource-HL cooperation that avoids inappropriate system behavior. In this context, the execution of Rules depends upon the explicit Smart-Product-HL interest in their utilization. In some manner, each Smart-Product- HL deals with Rules as a kind of expert agent. The solution has been applied in a set of examples in ANALYTICE II presenting some simulation independence because each control instance is not aware that Resource-HLs and Smart-Product-HLs are simulated.