Abstract: To solve the flexible job-shop scheduling problem more effectively, an improved invasive weed algorithm was proposed. A random key encoding scheme based on transformed sequences was proposed and an adaptive Gauss mutation operator was introduced to diversity the population in the process of weed breeding. In spatial diffusion stage, the standard deviation of normal distribution based on tangent function was used as seed’s new step size search method. In competition of invasive weed stage, by using the guided search strategy in the bee colony algorithm, the weed was guided to improve its ability to jump out of the local optimum. A random key encoding scheme based on transformed sequences was proposed. The proposed algorithm was compared with other different algorithms, the statistical results show that proposed algorithm has better convergence than other algorithms for solving the scheduling problem.

Key words: flexible job shop scheduling, weed optimization, adaptive Gauss mutation, guided search, random key coding