Grid task scheduling is a non-deterministic problem with polynomial complexity. All uncertain decision problems of polynomial time can constitute NP class problem. So how to quickly find the global optimal solution is the difficult point of grid task scheduling. The genetic algorithm, which has the characteristics of automatic acquisition and fast search, is the best solution to solve the nonlinear problem. This paper mainly analyzes the grid task scheduling method based on genetic algorithm, and completes the simulation experiment research of genetic algorithm through the grid task scheduling model construction, resource allocation and other operations.