Fuzzy feedback controller and method

A fuzzy feedback control system which determines a control output by fuzzy inference so that a process value coincides with a set value, and causes an operating device to control an object to be controlled, comprising:

• means for generating an actual process value from the object;
• an arithmetic memory device for obtaining a deviation between a process value and a set value, and a variation of said deviation;
• a first normalizing device for generating a normalized deviation from said deviation;
• a second normalizing device for generating a normalized variation of said deviation from said variation of said deviation;
• a fuzzy inference operating device for performing a fuzzy inference by using fuzzy control rules, membership functions, said normalized deviation and said normalized variation of said deviation, to obtain a control output variation;
• a third normalizing device for generating a normalized control output variation from said control output variation;
• an addition device for determining a present control output by adding said normalized control output variation to a preceding control output stored in said addition device, and for applying said present control output to said operating device;
• a parameter adapting device for determining and applying a first normalizing parameter for said first normalizing device, a second normalizing parameter for said second normalizing device and a third normalizing parameter for said third normalizing device, respectively, said second and third normalizing parameters being adjusted based on a response waveform of said process value;
• and
• means for applying the present control output to the operating device to control the object.

A fuzzy feedback controlling method for operating a fuzzy inference control system to regulate a process value of an object with improved stability and accuracy, the steps of said method comprising operating a fuzzy inference control system as follows:

• determining a fuzzy control rule including the substeps of:
• representing a reference straight line in a phase plane with a deviation and a variation of said deviation as coordinates axes, a speed type control output variation of a proportional/integral control being zero;
• generating a deviation between an actual process value from the object and a set value;
• generating a variation of the deviation;
• dividing said phase plane into rectangles equal in size in such a manner that vertices of said rectangles are on said reference straight line and a line parallel with said reference straight line;
• and
• regulating a variation of a control output variation corresponding to a distance between said reference straight line and said parallel line which is so defined as to be proportional to absolute values of said deviation and said variation of said deviation;
• regulating antecedent membership functions concerning said deviation and said variation of said deviation as fuzzy variables, with centers of isosceles triangles of said antecedent membership functions being deviation components and deviation variation components of coordinates of centers of said rectangles;
• regulating consequent membership functions concerning a control output variation as fuzzy variables, with centers of isosceles triangles of said consequent membership functions being defined by a distance between said reference straight line and said parallel straight line;
• normalizing said deviation, and said variation of said deviation at least with respective reference variation widths;
• determining by fuzzy inference a variation of said control output using said fuzzy control rule, said antecedent membership functions, said consequent membership functions, said normalized deviation, and said normalized variation of said deviation;
• normalizing said variation of said control output at least with a reference variation width;
• and
• applying the control output, as varied by fuzzy inference, to control the object.