A case based reasoning approach to mesh specification for adaptive finite element analysis

In this paper the application of an Artificial Intelligence paradigm called Case Based Reasoning (CBR) to the issue of mesh specification for adaptive finite element analysis is discussed. It is argued that CBR can provide greater flexibility than the traditional rule-based approach, permitting solutions of problems to be proposed even when domain knowledge is lacking. Moreover, CBR allows the exploitation the numerical knowledge extracted by a posteriori analysis of previously solved problems, encompassing a learning mechanism in a natural manner. We believe that the system proposed here can provide a powerful problem solving strategy, as well as providing a means of knowledge acquisition. Ultimately, coupling this approach with a rule-base can provide a problem solving environment for tackling complex real-world problems.