طراحي شبكه هاي موجك فازي چند متغيره و ارائه الگوريتم آموزشي تركيبي بهبوديافته در شناسايي سيستم هاي غير خطي

چكيده انگليسي :

Design of MIMO fuzzy wavelet neural networks and propose of improved hybrid learning algorithm in nonlinear system identification Mehrnoush Davanipour m davanipoor@ yahoo com March Department of Electrical and Computer Engineering Isfahan University of Technology Isfahan Iran Degree M Sc Language Persian Farid Shekholeslam sheikh@cc iut ac ir Maryam Zekri mzekri@cc iut ac ir AbstractIn recent years combination of soft computing with wavelet theory has eventuated new guidelines Fuzzywavelet neural networks which are a combination of fuzzy logic neural network and wavelet theory havebeen used in many researches The ability of fuzzy wavelet network and extended usage of multi variablesystems have made the motivation of designing multi input multi output fuzzy wavelet network in thisresearch Also a new algorithm which is called improved hybrid learning algorithm has been proposed inorder to increase training speed of fuzzy wavelet neural network The fuzzy wavelet neural network isconstructed on the basis of fuzzy rules that incorporate wavelet functions in their consequent parts Theability of fuzzy wavelet neural network in function approximation and system identification has been thesubject of many researches In addition to the inherent ability and property of fuzzy wavelet network that slearning process is one of the most important effective items Many different learning algorithms have beenproposed for fuzzy wavelet neural network but the back propagation method is probably the most frequentlyused technique in order to train a fuzzy wavelet neural network Although this algorithm has high ability infinding optimum points it has some shortcomings One of the most important of them is its slowconvergence to a minimum that is the main topic focused in this study In this research an improved hybridlearning algorithm which is a combination of clustering method recursive least square and accelerated backpropagation algorithm is applied in order to train a fuzzy wavelet neural network In this method fuzzywavelet neural network has been learned in three steps These steps include initialization optimization oflinear parameters and optimization of nonlinear parameters This proposed method gives the initialparameters by clustering algorithm then updates them with a combination of back propagation and recursiveleast square methods The parameters are updated in the direction of steepest descent but with a localadaptive learning rate which is different for each epoch and only depends on the sign of gradient errorfunction In order to accelerate the convergence speed a new idea is applied to determine the learning ratewithout trial and error That is inspired from the halving method for finding function roots The convergencecondition of the algorithm has been obtained by expressing a theory Even though the results are muchsatisfactory the algorithm is much simpler than other reported Also it does not include any excessive term inadapting formulation unlike most of researches in this area Simulation results indicate a superiorconvergence speed in comparison to other researches in fuzzy wavelet neural networks Furthermore thisalgorithm somewhat increases accuracy while using much fewer parameters Simulation results indicate asuperior convergence speed in comparison to other training methods in FWNN After proposing the hybridalgorithm a multi input multi output fuzzy wavelet network has been designed This multi variable fuzzywavelet neural network has been used for identification of multi variable nonlinear systems Simulationresults indicate the ability of designed multi variable network Then the proposed improved hybrid learningalgorithm has been developed for applying to designed multi variable fuzzy wavelet network Key WordsFuzzy wavelet neural network Improved hybrid learning algorithm Multi variable fuzzywavelet neural network System identification

دواني پور، مهرنوش

طراحي شبكه هاي موجك فازي چند متغيره و ارائه الگوريتم آموزشي تركيبي بهبوديافته در شناسايي سيستم هاي غير خطي

شناسايي سيستم