In this work, we suggest a variety of optimal daily diets for people with diabetes in Morocco. To this end, we model the foods glycemic load in term of triangular fuzzy numbers. Then, we model the problem of daily diet in term of fuzzy quadratic optimization programming. To transform this model into a line model, we use a very performance ranking function. To solve the obtained model, we call particles swarm algorithm due to its performance and simplicity. Various experiments were carried out on the basis of feeds available on the Moroccan market and for different sizes of the PSO swarm. The experimental results show that our system proposes optimal diets with a reasonable glycemic load between, acceptable favorable gaps and unfavorable nutrients excesses. We show how to combine the proposed diets to obtain a balanced weekly regime.