Qualitative/quantitative simulation of process temporal behavior using clustered fuzzy digraphs

A methodology is presented for qualitative modeling and simulation of the t emporal behavior of individual variables and the composite process. The tem poral behavior of a variable in a rt windowed time scale is captured catego rically using principal-component analysis, while the process is modeled as a causal diagraph with intel acting and recycle nodes. Rigorous, rather th an nd hoc, reasoning rules can be devised for the causal diagraph using a l earning mechanism. Quantitative information can also be incorporated into t he method with the introduction of fuzzy c-means clustering without comprom ising the cognitive level of information embedded in the nodes and the digr aph.