Today world food system is facing major challenges, including how to feed a growing world population, how to reduce rural poverty in the world, how to achieve nutrition balance, and how to manage ecosystem goods and services in light of global environmental change. In order to plan sustainable, equitable, and healthy food systems for the future, we establish RFS model. We first set up two models. One is the current Food System (FS) evaluation model, which uses entropy weight method and gray correlation analysis to evaluate food systems. The other is Re-optimizing Food System (RFS) model. Specifically, RFS operation threshold is introduced into RFS model, which is used to decide which model should be run, through comparison operation. Then, we compare the first model and the second model in task 1 to give the effect of the optimized food systems, and find the differences between the two food systems. In addition, after deciding whether to use the RFS model by the RFS operation threshold, BP neural network algorithm is used to predict that the implementation time of re-optimizing food system in Pakistan will be 2029.