Stochastic vendor selection problem

We study the vendor selection problem in which capacity, quality level, service level, and lead time associated with each vendor are considered to be stochastic. The problem is modeled as a stochastic dependent-chance programming model. As stochastic programming models are difficult to solve by traditional methods, a hybrid adaptive genetic algorithm, which embeds the neural network and stochastic simulation, was designed and implemented. To further improve the performance of the algorithm, the adaptive genetic algorithm was adjusted by varying the crossover probability and mutation rate according to the stage of evolution and fitness of the population. The solution procedure was tested on several randomly generated problems with varying parameters. Our extensive computational experience on these problems indicates that the hybrid adaptive genetic algorithm has strong adaptability on the tested problems as the algorithm converged more rapidly than the simple genetic algorithm.