The method of fuzzy inverse inference for desicion-support systems

The role of inverse problems in decision-support systems is considered.

Currently, decision support system (DSS) has been extensively developed and applied in various fields of human activity, particularly in those in which the solvable problems are poorly structured or hardly formalised. There is increasing of need to use methods, working in conditions of uncertainty, inaccuracy, incompleteness and inconsistency of the source data. To meet the challenges of decision support in these conditions are widely used methods of "Soft computing", in particular, methods based on fuzzy sets theory, fuzzy logic and fuzzy arithmetic.

There are two classes of problems arising in the process of generation and decision-making: direct problems and inverse problems.

For fuzzy systems, inverse problems are formulated as the problem of fuzzy inverse inference. It should be noted that currently there is no universal method for solving inverse problems in the fuzzy formulation for arbitrary fuzzy systems.

Existing inverse problem definitions for fuzzy systems are described.

Classical inverse problem definition for fuzzy systems was proposed by E. Sanchez. Its main limitation consists in the roughness of specified output fuzzy membership function. Later works in this field didn-t improve situation.

So inverse problem definition for fuzzy systems that is free of exposed limitations is considered.

Method of fuzzy inverse inference is proposed.

Example illustrating method-s operation is presented.