Plant modeling and control software design of advanced manufacturing systems under hierarchical control architecture

Hierarchical control architecture (HCA) refers to a structure where distinct controllers, each with its own specific function, are arranged in a hierarchical manner in a control pyramid. In modern manufacturing systems, HCA can be found in both shop level and machine level.; In this thesis, the issues for the controller and control software design in both machine level and cell level are studied. In the machine level, we conduct the controller and control software design for the fused deposition based layered manufacturing system. To build high quality functional parts, we investigate the means to satisfy two fundamental requirements, the positioning precision and the deposition accuracy. We establish a model for the fused deposition process, based on which a feedforward controller is developed to improve the positioning precision and a coordination controller is proposed to improve the deposition accuracy. Experimental studies demonstrate that our methodologies are effective. Since these methodologies directly deal with real time control of the process, we separate them from planning based methodologies. From a{09}planning point of view, a tool path based deposition planning methodology is proposed and implemented in the management layer in order to obtain high quality functional parts. This deposition planning procedure is further extended for improving productivity of fused deposition process. A number of samples were made and tested to manifest the feasibility and efficiency of the planning approach. In addition, a trajectory planning approach is also proposed to improve the positioning precision for high speed motions.; In the cell level, a component-based plant modeling and synthesis approach is conducted. The components in a plant hierarchy are modeled by three structures, static structures using class diagram, dynamic behaviors using Petri Nets (PNs), and functional logic using rule sets. The plant controller is synthesized in two levels, the IC (integrated component) level and the agent level. In the IC level, an inference engine takes the plant information, control specification, and the functional logic to deduce the specific transition conditions for the IC PN models. Then, the complete PN plant model is created through a replacement scheme that links the disconnected PNs within the same or across the different ICs using those specified transition conditions. In the agent level, the concept of task sequence is defined and message-passing function is particularly modeled. A task planner then derives the task sequence according to the plant information, domain theory, and control specifications. The derived task sequence is further converted systematically into the plant (PN) controller through the proposed mapping templates. Theoretical results are employed and developed to detect and guarantee the desirable properties (e.g., liveness and cyclicness) of the synthesized PN model in both IC level and agent level.; Finally, the FDMM (Fused Deposition of Multiple Materials) system developed at Rutgers University is used as a test bed for our algorithms and methodologies.