A Knowledge-Based Approach for Microwire Casting Plant Control

Background

The aim of modeling is capturing the essence of phenomena behavior. When complete knowledge of a process eludes us, we build models in order to obtain some measure of control over that process. A model is never entirely correct but it is useful if it explains and predicts the behavior of the process within the limits of precision required for the task. It is essential that the users of the model understand well the conditions over which the model has been developed and consequently the regimes of its validity. If the phenomenon to be modeled is understood well enough to construct a model and if its mathematical formulation is suitable to be analytically or numerically solved, then the resulting model is a powerful tool as it enables us to explain and predict system behavior within the bounds of the validity of model.

The process of microwire casting is qualified with a highly elevated level of complexity. It represents a joining of interactions, such as mechanical, thermal, electrodynamical, physical, and chemical. More than that, those interactions are not just multiple but are overlapped during the time of casting. It can be supposed that a model of casting might be derived from the underlying properties of the process. However, even under various approximations, the final model will be too difficult, of high order, nonlinear and so on.

Having a high degree of complexity, the creation of a mathematical complete and proper model for the automation and optimization of the process of microwire casting represents a highly difficult problem (Berman, 1972). We believe that pure physical and mathematical evaluation of the discussed process doesn't represent a practical solution.