Fuzzy controller design by artificial DNA assisted queen bee genetic algorithm

This paper proposes an artificial DNA assisted queen bee genetic algorithm (DNA+QBGA) to learn the gains, control structures, membership functions, and rules of the fuzzy controller. The queen bee genetic algorithm (QBGA) possesses simple and fast evolution process to figure out the best parameters and the DNA computing is adopted to determine the structure of fuzzy controller. Each fuzzy control structure can be defined by a different bee hive, which contains the control structure and dimension of the gain. The presented DNA+QBGA can make the membership functions and rules communicate with one another among different control structures. Moreover, a novel three-step crossover operation is investigated such that the crossover between different odd dimensions of membership functions can be made. Step one is that the dimensions of parents (queen and drone) and the offspring (brood) are expanded to the same dimension resolved by their least common multiple. Step two is to randomly select the genes from the parents in the corresponding space. Step three is that the offspring gene is calculated by the real-code crossover between their parents. Finally, the simulation results of the fuzzy controlled cart-pole and chaotic systems demonstrate the feasibility and effectiveness of the proposed schemes.