Parameter optimization of a fuzzy inference system using the FisPro open source software

This paper proposes a flexible optimization sequence that can be applied to any parameter of a fuzzy inference system. Interrelated parameters can be optimized together, and criteria include system accuracy and coverage. The fuzzy inference system structure is preserved and constraints are imposed to respect the fuzzy partition semantics. The procedure described here uses a Solis & Wets based algorithm, but the approach remains valid for other optimization techniques, provided that they accept semantic constraints. The optimization sequence is implemented in an open source software, FisPro, made for fuzzy inference system design and tuning.