The Research Of Fuzzy-rules Automatic Generation Method Based On Rough Sets Theory

During twenty years, fuzzy control obtains quiet great developments. But acquisition of fuzzy control rules is one of primary problems, which puzzles fuzzy control progress all the while.It is important to structure the fuzzy rules store in the fuzzy control system.However, from a lot of literatures we can see: basically , person decides the numbers of fuzzy rules. Although there are many kinds of methods to optimize compute, it is difficult to artificially set fuzzy rules, even impossible.Therefore, researching automatic generation of fuzzy rules has important values in the theory and application.Based on Rough Sets Theory, this thesis puts forward an approach to automaticly generate fuzzy rules from metrical data. Firstly, fill in vacant values of sample data in the collected fuzzy system, applied the idea of the spline insertion.Then classify the input-output sample data , which have been inserted used K- average value clustering method.Finally, disperse the data based on the projections of sample space clustering on every dimension, offer the object foundaction for the fuzzy subclass deceptions of each variable in fuzzy system, make a good base for the futher automatic establishment of fuzzy rules and the automatic adjustion of fuzzy membership function.There is a detail in the first part of the thesis.In the second part of the thesis, firstly introduce the basic conceptions of the Rough Sets Theory , then a function named occupancy is defined based on Rough Sets Theory and distill the fuzzy rules the above used ideas and methods.In the third part of the thesis, fuzzy controller is generated based on the above fuzzy rules store and is applied to the control experiments of a single closed loop direct current timing system.The result of the experiments indicates that fuzzy controller which is set by the way of automatic collecting fuzzy rules from metrical data can meet control requirements and it is robust.