Invasive Weed Optimization Algorithm with Discrete Multi-population for the Flexible Job-shop Scheduling Problem

After analyzing the characteristics of the flexible job-shop scheduling problem (FJSP), we propose an invasive weed optimization algorithm with discrete multi-population based on the basic invasive weed optimization to solve the FJSP. The proposed algorithm uses the multi-population without population exchange at the early stage, while the weeds use the crossover operator to communicate with each other within their own population. Self-adaptive mutation and local search are used in the space diffusion to improve the global search ability at the early stage of the algorithm and the local search ability at the later stage of the algorithm. At the later stage of the algorithm, the convergence rate and the optimization accuracy of the algorithm are improved through population exchange. This algorithm is used in FJSP and a matrix-decoding method is proposed when decoding. The effectiveness and the advantage of the algorithm are demonstrated using some examples.