Earticle

This article deals with solving the flexible job shop scheduling problem in dynamic environment (DFJSSP). In this problem, environment may face with many real time events such as random arrival of jobs or breakdown machine efficiency. Jobs and their operations are processing the machines according to static scheduling in which environment might face with such events. Regarding being NP – hard of the problem , a hybrid of artificial immune and virus evolutionary algorithm are offered to solve it which use the technique of stable action – reaction scheduling . In these algorithms two objective functions are minimized: Efficiency and stability. Efficiency is the objectives value in static scheduling, whereas stability is presented in dynamic scheduling because its purpose is to improve static scheduling, reduce the deviation from the first scheduling, and increase system stability.

The genetic algorithm that simulates the virus evolutionary theory has been developed applying to combinatorial optimization problems. The algorithm in this study uses only one individual and a population of viruses. The individual is attacked, inflected and improved by the viruses. The viruses are composed of flour genes (a pair of top gene and a pair of tail gene). If the individual is improved by the attacking, the inflection occurs. After the infection, the tail genes are mutated. If the same virus attacks several times and fails to inflect, the top genes of the virus are mutated. By this mutation, the individual can be improved effectively. In addition, the influence of the immunologic mechanism on evolution is simulated.

This paper addresses the systematic approach to the fuzzy modeling of the class of complex and uncertain nonlinear systems. While the conventional genetic algorithm (GA) only searches the global solution, Virus-Evolutionary Genetic Algorithm(VEGA) can search the global and local optimal solution simultaneously. In the proposed method the parameter and the structure of the fuzzy model are automatically identified at the same time by using VEGA. To show the effectiveness and the feasibility of the proposed method, a numerical example is provided. The performance of the proposed method is compared with that of conventional GA.