Complex system modeling by using artificial neural network based intelligent methods

ABSTRACTPhD ThesisCOMPLEX SYSTEM MODELING BY USING ARTIFICIAL NEURAL NETWORK BASEDINTELLIGENT METHODSDavut HANBAYFirat UniversityGraduate School of Natural and Applied Sciences, Department of Electrical - Electronics Engineering2007, Pages: 100Modeling complex systems are being studied by scientists for a long time. In most of themmathematical relationships are used. Modeling complex systems are become harder or sometimesimpossible when the numbers of input parameters are increased or environmental conditions areconsidered. The offered solutions in various science fields by intelligent methods take care of systemsannalists. Nonlinear system modeling is made easy by intelligent system modeling studies. But someproblems such as training time, increased sample data, memorization, necessitates to use delayed inputand output arise with intelligent methods when system input parameters or the level of complexity ofthe relationships between input-output pairs are increased. In this thesis, some solutions are suggestedfor these problems with developed intelligent modeling structures and offered a new modelingapproach based on artificial intelligence. The offered modeling approach is based on getting linearparts of complex system and modeling with artificial intelligence techniques.The developed intelligent structures for complex system modeling are consisting of twoimportant parts. The first of them is obtaining linear key model features which can represent thecomplex systems. The second is learning of these model features by artificial intelligence techniques.So contrary to traditional complex system modeling studies, model feature extracting process is addedto modeling stages. In this thesis, the studied complex systems are analyzed by using wavelettransform and entropy methods. Adaptive network based fuzzy inference systems, fuzzy logic andartificial neural networks are used as artificial intelligence techniques. In another developed modelingstructure, model feature extracting process is adapted related to the performance of artificialintelligence techniques. Thus model feature extracting process and artificial intelligence stage areintegrated.Keywords: Complex Systems, Intelligent Modeling, Feature Extracting, ArtificialIntelligence Techniques.I