Fuzzy inference subsystem for soft computing optimizer of knowledge bases

Software architecture of fuzzy inference subsystem for soft computing optimizer is considered. Proposed architecture is based on segregation of interface of algorithms, interested in active rules, to separate interface. Fuzzy inference algorithm is implemented as a template method; its primitive operations are implemented in concrete realizations of fuzzy inference models and fuzzy rule bases. Algorithm is also configurable by the object which will receive a list of active rules. Three operational modes of subsystem are described: fuzzy inference, LBRW rule database creation and rule analysis. Performance of soft computing optimizer is demonstrated on the task of creating control system for instable dynamic object. This control system exhibited increasing robustness comparing to systems created with other state-of-the-art tools.