Design And Applications Of Fuzzy Logic Systems Based On Ant Colony Intelligent Optimization

Fuzzy logic systems have been successfully used in automatic control,image recognition,decision analysis and other fields,because they can combine with some known knowledge and expert experience and have unparalleled advantages.However,there are still some difficult problems: two main problems are obtained parameters and the “rule explosion”.The so-called “rule explosion” is that with the input increase the number of rule is increase sharply,thus,both of the complexity of fuzzy logic system and the calculation are increased.It has been a research hot spot that adopt intelligent algorithms,such as ant colony optimization,Genetic Algorithm and particle swarm optimization to design fuzzy logic system,due to the strong robustness and global convergence of intelligent algorithms.In this thesis,Ant Colony Optimization(ACO)is used to design fuzzy logic system,which includes the structural identification and parameters identification.Meanwhile an Improved Ant Colony Optimization(IACO)with the update of global pheromones is proposed based on ACO,and also used to design fuzzy logic systems.The intelligent fuzzy logic systems are used to predict the data of Mackey-Glass chaos time series and the price of international petroleum.The simulation results indicate the feasibility and high-performance of ACO and IACO.In addition,the compared results with Back Propagation(BP)show that the superiority of the intelligent algorithms.The specific work is as follows:(1)Introduce the basic concepts and relevant knowledge of the Type-1 TSK fuzzy logic systems,A1-C1,A2-C0,A2-C1 interval Type-2 TSK fuzzy logic system and Ant Colony Optimization(here,A represents antecedent,and C represents consequent).The concrete method and advantage of the Improved Ant Colony Optimization are given in the thesis.(2)The design problem of Type-1 TSK fuzzy logic system based on ACO and IACO are researched respectively,including the parameter update and rule selection.The concrete methods and steps have been given.The fuzzy logic systems are integrated into the five-layer neural network to design the fuzzy neural network systems.The systems are used to predict the data of Mackey-Glass chaos time series and the price of the international petroleum,and the simulations show that the feasible and high-performance of ACO and IACO,the compared results with BP show that the superiority of intelligent algorithms,and the compared results between ACO and IACO show that the perform of update the global pheromones trail can accelerate the convergence speed of the algorithm and increase the performance.(3)The design problem of several classes interval Type-2 TSK fuzzy logic systems based on ACO and IACO are researched,including the parameters adjustment and rule selection,the interval Type-2 TSK fuzzy logic systems include A1-C1,A2-C0 and A2-C1 interval Type-2 TSK fuzzy logic system.Design five-layer fuzzy neural network by integrated fuzzy logic system into neural networks for A2-C0 interval Type-2 TSK fuzzy logic system,and design of six-layer fuzzy neural networks by integrated fuzzy logic systems into neural networks for A1-C1 and A2-C1 interval Type-2 TSK fuzzy logic systems.The intelligent fuzzy logic systems are also used to predict the data of Mackey-Glass chaos time series and the price of the international petroleum.The simulation results show that they are feasible and high-performance;meanwhile,IACO with the global pheromones trail update has a better performance than ACO.Both the intelligent fuzzy logic systems based on ACO and IACO are superior to BP.(4)The compared results between above intelligent fuzzy logic systems show that interval Type-2 TSK fuzzy logic systems have better tracking performance than Type-1 TSK fuzzy logic system,for the Type-2 fuzzy logic systems,more complex fuzzy degree,and more output precise.