Genetic Programming Network (GNP), a graph-based evolutionary method, has been proposed several years ago as an extension of Genetic Algorithm (GA) and Genetic Programming (GP). The behavior of GNP is characterized by a balance between exploitation and exploration. To improve the evolving speed and efficiency of GNP, we developed a hybrid algorithm that combines GNP with Ant Colony Optimization (ACO). Pheromone information in the algorithm is updated not only by the fitness but also the frequency of the transitions as dynamic updating. We applied the hybrid algorithm to Elevator Group Supervisory Control Systems (EGSCS), a complex real-world problem. Finally, the simulations verified the efficacy of our proposed method.