Abstract: A new hybrid genetic algorithm is proposed for nonlinear programming problems in this paper, which combines genetic algorithm ( GA) with sequential linear programming method. During the iterative computation process, if crossover or mutation operation don't happen at the iterative points in the GA, the objective function and constraints at these points will be linearized. In order to satisfy the constraints within the neighborhood of these points, this paper considers adding soft constraints, and the linearized optimization problem can be solved with the linear programming. The new hybrid genetic algorithm is globally convergent; it does not require that the iterative points must be feasible. Simulation results show that the algorithm is effective and reasonable.