An Adaptive Annealing Genetic Algorithm for job-shop scheduling

Most production planning and scheduling applications are complex combination optimization in nature. Genetic Algorithm (GA), Simulated Annealing Algorithm (SAA) and Optimum Individual Protecting Algorithm (OIPA) have application limitations due to their performance in global convergence, population precocity and convergence speed, which make them not suitable for workshop daily operation planning applications. The Adaptive Annealing Genetic Algorithm (AAGA) studied in the paper has unique advantages to deal with the above limitations through 1) adaptively changing mutation probability to shorten the optimizing process and avoid the local optimization; and 2) integrating the Boltzmann probability selection mechanism from simulated annealing algorithm to select the crossover parents to avoid the population precocity and local convergence. The detail of AAGA is introduced and a typical application example for daily workshop operation scheduling is studied using GA, SAA, OIPA, and the proposed AAGA, respectively. As seen from the simulation results, the proposed AAGA shows an improved performance.