Fuzzy logic controller

A fuzzy logic controller for receiving input signals from a process, utilizing data in a knowledge-base and, based on said process input signals and said data, providing output control signals to said process in order to provide knowledge-based control of said process, said knowledge-base of the type containing

• fuzzy rules comprising representations of fuzzy patterns of consequents (THENs) and representations of fuzzy patterns of antecedents (IFs);
• clearness distributions comprising collections of representations of clearness degrees, said clearness degrees defining cognitive measures which assess the fuzziness of fuzzy patterns, and
• representations of thresholds of transitions, said thresholds of transitions being the intersection points of every two adjacent clearness distributions,
• said fuzzy logic controller comprising:
• a fuzzifier comprising first means for mapping said process input signals into representations of input fuzzy patterns and second means for mapping said process input signals into representations of input clearness degrees of said input fuzzy patterns, said input clearness degrees being valued in the interval of zero to one, where zero represents "absolutely unclear" ("absolutely false") and one represents "absolutely clear" ("absolutely true");
• pattern matching and rule selection means for receiving said representations of input fuzzy patterns from said first means of said fuzzifier and, responsive thereto, for selecting one rule from said knowledge-base, and for generating representations of control action fuzzy patterns based on the consequent (THEN) portion of said rule;
• approximate reasoning means for receiving said representations of input clearness degrees from said second means of said fuzzifier and for mapping said representations of input clearness degrees into a representation of a control action clearness degree, said approximate reasoning means operating in parallel with said pattern matching and rule selection means;
• and
• defuzzifier means for receiving said representations of control action fuzzy patterns from said pattern matching and rule selection means and for receiving said representation of a control action clearness degree from said approximate reasoning means and for mapping said representations of control action fuzzy patterns and said representation of a control action clearness degree into said output signals.

A method for controlling a process having process signals comprising the following steps:

• mapping said process signals into representations of input fuzzy patterns;
• mapping said process signals into representations of input clearness degrees of said input fuzzy patterns, said input clearness degrees defining cognitive measures which assess the fuzziness of input fuzzy patterns, said input clearness degrees being valued in the interval of zero to one, where zero represents "absolutely unclear" ("absolutely false") and one represents absolutely clear" ("absolutely true");
• selecting a fuzzy rule from a collection of fuzzy rules of the type comprising representations of fuzzy patterns of consequents (THENs) and representations of fuzzy patterns of antecedents (IFs) having an antecedent matching said representations of input fuzzy patterns;
• generating representations of control action fuzzy patterns based on the consequent (THEN) portion of said selected rule;
• while selecting a fuzzy rule, mapping said representations of input clearness degrees into a representation of a control action clearness degree, each said control action clearness degree defining cognitive measures which assess the fuzziness of control action fuzzy patterns;
• and
• mapping said representations of control action fuzzy patterns and said representation of a control action clearness degree into output signals and inputting said process with said output signals in order to control said process.