7847

7312

كارشناسي ارشد

كنترل

اصفهان: دانشگاه صنعتي اصفهان، دانشكده برق و كامپيوتر

1391

نه،90ص.: مصور،جدول،نمودار

ص.ع.به فارسي و انگليسي

جواد عسگري

مريم ذكري

22/4/92

فريد شيخ الاسلام، مرضيه كمالي

مهندسي برق و كامپيوتر

ID7312

به فارسي و انگليسي: قابل رويت در نسخه ديجيتالي

1 Controller Design Using Type 2 Fuzzy Neural Networks Hossein Moradi Farahani h moradifarahani@ec iut ac ir Date of Submission 2013 01 23 Department of Electrical and Computer Engineering Isfahan University of Technology Isfahan 84156 83111 Iran Degree M Sc Language PersianSupervisor Javad Askari j askari@cc iut ac irAbstractFuzzy logic is a subset of the soft computing that gives the ability to make decisions inuncertain conditions to computer systems Today the fuzzy expert systems are successfulin some states such as making decisions in the conditions of uncertainty and control of thecomplex systems In a fuzzy system it is very difficult to determine the exactmembership degree especially in unknown systems or highly nonlinear systems Thisproblem is solved by using type 2 fuzzy logic and type 2 fuzzy systems In the type 2fuzzy logic membership degree is a fuzzy number In recent years type 2 fuzzy systemshave been more attention because of more flexibility and capability in systems modelingin the high uncertainty condition Fuzzy neural networks are a kind of hybrid intelligent systems obtained from fuzzysystems and neural networks These structures have the learning ability of neuralnetworks and inference ability of fuzzy systems So they can be used for variousapplications In recent years type 1 fuzzy logic generalizes to type 2 fuzzy logic andtype 1 fuzzy neural networks have been developed to type 2 fuzzy neural networks In this thesis type 2 fuzzy logic and type 2 fuzzy systems is briefly introduced andvarious structures of type 2 fuzzy neural networks and their learning algorithm to controlof nonlinear dynamic systems are reviewed Due to the complex nature of type 2 fuzzyneural models than the polynomial models to expand the use of type 2 fuzzy neuralmodels these models can be further simplified In this thesis the proposed method tosimplify type 2 fuzzy neural network is to reduce the number of fuzzy rules By reducingthe number of fuzzy rules using manual and automated method number of modelparameters will be very low and the network training time will be reduced Reduction inthe number of rules when using type 2 fuzzy neural network in online identification andcontrol will help greatly Also two controller design methods including adaptive inversecontrol and indirect adaptive control using type 2 fuzzy neural networks are expressed These controllers respectively for water temperature control system and twin tank systemare designed and their simulation results are investigated Keywords Adaptive Inverse Control Indirect Adaptive Control Type 2 Fuzzy Logic Type 2 Fuzzy Systems Type 2 Fuzzy Neural Network Temperature Control System Twin Tank System

جواد عسگري

مريم ذكري

فريد شيخ الاسلام، مرضيه كمالي