Research On Personalized Nutrition Catering System

With the development of the times,people's awareness of healthy eating has gradually improved.However,due to objective restrictions,most people currently cannot customize a scientific diet plan for themselves.Therefore,how to design a nutritional catering system that meets the user's interest according to the user's nutritional needs and dietary preferences has become a realistic subject.This article mainly studies methods based on personalized recommended diet and optimized dietary balance.The main contributions of this article are as follows:1)Propose a diet recommendation algorithm based on attention attribute heterogeneous network embedding(Attention Attributed Heterogeneous Network Embedding,AAHNE).In this paper,we first use the attribute heterogeneous network to model the interaction between the user and the diet,and then introduce the feature attention and self-attention mechanism in the embedding layer to learn the influence of the interaction between different nodes and different interactions in the network on the embedding.Finally,a random walk algorithm based on meta-path is used to optimize the node embedding.Using the kitchen interactive data set,this pape r discusses the effect of parameter settings on the performance of the AAHNE model,and compares it with the current more classic recommendation algorithm based on network embedding.The experimental results show that the AAHNE algorithm has better performance.2)Propose an improved fast non-dominated genetic algorithm(Food Nondominated Sorting Genetic Algorithm,FNSGA)and apply it to the nutritional catering system.This article first redesigned the genetic algorithm coding for the nutritional catering scenario and added restrictions.Secondly,for the high complexity of the algorithm and easy to fall into the local optimal problem in the multi-objective optimization,three improved operators were designed.They are to use the selection operator based on micronutrient adjustment to ensure that the algorithm reduces the number of primary optimization goals while taking into account secondary optimization goals and reduce the complexity of the algorithm;use the crossover operator based on recipe similarity to reduce the frequency of invalid crossovers and improve population exploration Efficiency: The use of mutation operators based on survival rounds enables the algorithm to dynamically adjust the mutation probability to avoid premature convergence or slow convergence.This paper compares FNSGA with NSGA-II.The experimental results show that the FNSGA algorithm has faster population diversity expansion and optimization speed,and is suitable for multi-objective optimization requirements in the nutritional catering system.3)Constructed a personalized nutritional catering system.This article uses We Chat applet as the front-end display on the Android system mobile phone,the Flask framework is used in the back-end,and the database uses My SQL.According to the user's information and historical diet records,the user's theoretical recommended nutrient value is calculated on the day,and a weighted inheritance algorithm is proposed to ensure that the user's long-term nutrient intake is balanced.Subsequently,the AAHNE algorithm is used to generate a diet recommendation list for the user,and the FNSGA algorithm is used to calculate a suitable meal list for the user to choose from the recommendation list during meal preparation.Finally,record user feedback as a basis for recommendation and optimization.