The order scheduling problem is studied under various constraint conditions. A scheduling model is established with the goal of minimizing the number of switching between two types and minimizing the sum of squared differences in the length of the car roof color blocking, and an improved genetic algorithm (IGA) is designed to solve the problem. The model is constructed through linear weights, which the weights are adjusted by combining grid search and random search to reflect the corresponding constraint priority based. During the solving process, integer encoding was applied to the data orders, and an improved sequential crossover operation was adopted, as well as four neighborhood search strategies, which improved the mutation operation in traditional genetic algorithms. By testing actual data orders of enterprises and comparing them with traditional genetic algorithms (GA) and differential evolution algorithms, the optimization effect is obtained and convergence ability of the IGA proposed are significantly improved. The IGA proposed can be solved the problems of low solution accuracy, poor convergence effect, and low constraint satisfaction rate in traditional GA.