LINGUISTIC AND CHOQUET INTEGRAL BASED EVALUATION IN PROCESS PLANNING

Abstract

This paper addresses fuzzy evaluation issues as applied in a decision–making procedure of process planning. General framework of fuzzy partitioning of system variables, evaluation procedure, and system variables, evaluation procedure, and system parameters aggregation have been described in detail for two classes and three distinguishing types of evaluation. Classes of evaluation include (a) linguistic, and (b) fuzzy (Choquet) integral based evaluation. Within these classes we distinguish between: (1) single criterion evaluation, (2) global multicriteria evaluation, and (3) global integral evaluation. First type relates to fuzzification of system parameters and matching procedure of a system's parameters input values to linguistic values in pre–defined dictionary. Multicriteria evaluation provides aggregation protocol of an object's attributes values and their significance yielding its suitability measure. We can use this measure to select the most appropriate linguistic value from pre–defined dictionary for further propagation through reasoning system, as well as for ranking of possible alternatives since it provides their global rank. Both types belong to linguistic evaluation class. Finally, global integral evaluation employs fuzzy integrals (here Choquet integral) for overall assessment of alternatives status. This type of evaluation belongs to fuzzy integral based class. Each type of evaluation is illustrated by an example from the real industrial practice concerning metal cutting process planning. The whole methodology should be seen as an upgrade of traditional expert systems used for this purpose. Relaxation of some system constraints provides more generality to traditional approaches, puts more "naturality" in assessment of parameters values, and reduces knowledge bases.