A hardware implement Adaptive Genetic Algorithm (AGA) is proposed in this paper. The adaptive algorithm uses three parameters, i. e. fmax , fmin and fave to determine the fc and fm of the whole generation adaptively. The selection , crossover and mutation operators which are suitable for hardware implement are selected and they are designed in a pipelining architecture . The parallelism of the selection operator and the computation of the fitness of the individual enhance the efficiency of the algorithm greatly. The hardware GA processor has been implemented in XILINX FPGA(Field Programmable Gate Arrays) XC2V1000. The VHDL language is used to describe the whole algorithm. Experimental results indicate that the adaptive genetic algorithm improves the global convergence and search performance of the algorithm greatly. The hardware implementation of the algorithm reduces the running time efficiently and makes it possible to apply in time-critical systems.
潘正君，康立山，陈毓屏. 演化计算[M]. 北京：清华大学出版社，1998: 3-4.[2]Petridis V, Kazarlis S, Bakirtzis A. Varying fitness function in genetic algorithm constrained optimization: The cutting stock and unit commitment problems[J]. IEEE Trans. on SMC Part B: Cybernetics , 1998, 28 (5): 629-639.[3]Shaunna M, Tom L, Abdulla H. A gene- tic algorithm environme- mnt for star pattern recognition[J]. Journal of Intelligent and Fuzzy Systems, 1998, 6(1): 3-16.[4]Bhandarkar S M , Zhang H.Image seg- ment using evolutionary computation[J]. IEEE Trans. on Evolutionary Computation,1999, 3 (1): 1-21.[5]唐加福，汪定伟，高振，等. 面向非线性规划问题的混合式遗传算法[J]. 自动化学报，2000，26(3): 401-404.[6]骆晨钟，邵惠鹤. 采用混沌变异的进化算法[J]. 控制与决策，2000，15(5): 557-560.[7]Scrinvas M, Patnaik L M. Adaptive Probabilities of Crossover and Mutation in Genetic Algorithms . IEEE Trans. on SMC, 1994, 24(4): 656-667.[8]王蕾，沈庭芝，招扬. 一种改进的自适应遗传算法[J]. 系统工程与电子技术, 2002, 24(5): 75-78.
DownLoad: