An agent based architecture for model based control

New materials and processing techniques offer new possibilities for designers, but these are often associated with tighter processing windows. Model based control built on the integration of process know-how and simulation tools can address these ever increasing control requirements. Such a system is highly adaptive and can provide cycle-to-cycle as well as in-cycle control. The design of an architecture for model based control is described. The control model sends and receives information from a data model, which contains sensor inputs, process model variables and output parameters. The process model is composed of many sub-models, which simulate several aspects of process behaviour across different time scales. The outputs of the sub-models are integrated into an overall view of the process and then fed into the data model. The system is designed to control multiple related manufacturing processes simultaneously. An example based on the thermoforming process is given.