Interval Type2Fuzzy Neural Network System and Application Research 

Author  HanChunYu 
Tutor  WangTao 
School  Liaoning University of Technology 
Course  Applied Mathematics 
Keywords  type2fuzzy set fuzzy logic system neural network interval type2fuzzyneural network system BP algorithm 
CLC  TP183 
Type  Master's thesis 
Year  2014 
Downloads  24 
Quotes  0 
In recent years, the fuzzy neural network system has been a hot topic in many fields ofacademic and practical application. The existence of the knowledges in the fuzzy logic systemis explicit, the algorithm is directly, the system is simple, the data processing is small, but thesystem flexibility is poor and no selflearning ability. The neural network is very difficult tounderstand, and the structure is more complex, and it will deal with large amount ofinformations. But the system has selflearning ability, a high degree of parallelism, and thesystem has good robustness. Visible, they have the advantage of complementary; the systemcan overcome the shortages of the single system. The thesis designs two kinds of intervaltype2fuzzy neural network system which combines type2fuzzy logic system and neuralnetwork. Its structure as neural network and its performance like type2fuzzy logic system.Then using the software of MTALAB and the Error Back Proragation algorithm (BPalgorithm) adjusts the system parameters and weights. Finally, using the two kinds of intervaltype2fuzzy neural network system to solve actually problem which was to forecast the priceof West Texas Intermediate (WTI) crude oil and the price of Consumer Price Index (CPI).Main works are summarized as follows:(1) Introduce the development and application of the fuzzy logic system and neuralnetwork, and their relevant knowledge.(2) Introduce the structure of type1fuzzy neural network system. And introduce thestructure, the development and the application of interval type2fuzzy neural network system.(3) Research type1fuzzy neural network system based on the type1fuzzy logic systemand neural network. Using BP algorithm adjust the parameters and weights. Given theapplication forecast the price of WTI, and simulation with MATLAB. The simulation resultsshow that the design of the type1fuzzy neural network system is effective and feasible.(4) Research interval type2fuzzy neural network based on type2fuzzy logic systemand neural network. Designed two kinds of interval type2fuzzy neural network system,using BP algorithm adjust parameters and weights.(5) The MATLAB simulation results can be seen the design of the two kinds of intervaltype2fuzzy neural network system is feasibility and effectiveness which was applied toforecast the price of WTI crude oil and the price of CPI. Though the tracking diagrams andthe RMSE that can be seen the interval type2fuzzy neural network system has better controlperformance than type1while the system has uncertainty; And compare the two kinds ofinterval type2fuzzy neural network system, we can see the more adjustment quantity ofweights the more accurate.