A health diet recommendation system based on a family

Abstract

The application discloses a health diet recommendation system based on a family, which comprises a data processing module, a management module, a recommendation generation module, a taboo replacement module and a situation adaptation module. The data processing module collects and stores health parameters of family members; the management module generates real-time inventory signals; the recommendation generation module receives nutrition demand signals and real-time inventory signals, and generates initial recommended recipes in combination with a preset recipe database; the taboo replacement module performs replacement analysis on taboo ingredients in the initial recommended recipes according to preset dietary taboo rules; and the situation adaptation module accesses external environment sensors and user behavior monitoring equipment. The health diet recommendation system based on the family can solve the problem of coordinated recommendation of differentiated nutrition demands of multiple family members and real-time scenes.

Description

TECHNICAL FIELD

[0001] The present application relates to the cross-technology field of intelligent healthy home and artificial intelligence recommendation system, to the field of wearable health monitoring devices, and in particular to a family health diet recommendation system. BACKGROUND

[0002] Current health diet recommendation technology mainly targets individual users and is difficult to adapt to the actual needs of the family multi-member scene. Traditional systems such as applications based on personal health data can only generate a recommended scheme according to the physiological indicators of a single user. When there is a conflict in the nutritional needs of family members, for example, children need high-calcium diet for development while the elderly need to control cholesterol intake, the system cannot optimize the overall menu in coordination, resulting in users being forced to manually adjust the recommended results, which greatly reduces the use efficiency. In terms of food management, existing technologies rely on manual input of inventory information and fail to achieve deep linkage with intelligent kitchen devices. When the food required by the recommended menu does not match the actual inventory, about 30% of family users give up using the system due to frequent purchase adjustments. The lack of dynamic adaptability is also a significant defect: mainstream systems use static nutrition models and do not access real-time exercise, sleep data monitored by wearable devices or temperature and humidity information collected by environmental sensors, so they cannot respond to sudden health changes (such as the need to supplement vitamin C when having a cold) or changes in dietary needs caused by seasonal changes.

[0003] The handling of dietary restrictions also has limitations. Existing solutions mostly use simple filtering mechanisms to directly exclude allergenic food or religiously prohibited food, lacking intelligent strategies based on nutritional equivalent substitution. Such mechanized operations are prone to cause the lack of key nutrients, and international research shows that the rate of household meal imbalance caused by this is high. SUMMARY

[0004] In view of the above shortcomings of the prior art, the present application aims to provide a family health diet recommendation system to solve the problem of coordinating the nutritional needs of multiple family members and real-time scenarios. The present application integrates the health parameters of family members through a data processing module, generates a coordinated nutritional demand signal using a multi-objective optimization algorithm; the management module links with the kitchen monitoring device to obtain real-time inventory data; the recommendation generation module outputs the initial menu in combination with the above signals; the taboo replacement module automatically replaces conflicting food based on the knowledge graph; the context adaptation module dynamically corrects the menu according to environmental and user behavior data; the four modules are iterated through a closed loop mechanism, and finally output a family diet plan that meets the requirements of nutritional balance, inventory matching, taboo avoidance and scenario adaptation.

[0005] The present application provides a family health diet recommendation system, comprising:

[0006] A data processing module collects and stores the health parameters of family members and generates a nutritional demand signal based on a multi-objective optimization algorithm;

[0007] a management module, the management module being connected to the kitchen inventory monitoring device to obtain food material types and shelf life data, and generating a real-time inventory signal;

[0008] a recommendation generation module, the recommendation generation module receiving the nutritional demand signal and the real-time inventory signal, and generating an initial recommended recipe in combination with a preset recipe database,

[0009] a taboo replacement module, the taboo replacement module performing replacement analysis on taboo food materials in the initial recommended recipe according to preset dietary taboo rules, generating a replacement instruction signal and feeding back to the recommendation generation module to output a final recipe;

[0010] a context adaptation module, the context adaptation module accessing external environment sensors and user behavior monitoring devices, generating a dynamic adjustment signal and transmitting the dynamic adjustment signal to the recommendation generation module to trigger recipe calorie and nutritional component correction based on a real-time scene;

[0011] The recommendation generation module, the taboo replacement engine, and the context adaptation module form a closed loop iteration until a final recipe that meets the nutritional demand, the inventory state, and the dynamic scene constraint is output.

[0012] In an embodiment of the present application, the health parameters collected in the data processing module include age, gender, basal metabolic rate, chronic disease history, and exercise habit data of family members. A multi-objective optimization algorithm generates differentiated nutritional allocation signals for different members by establishing a nutritional demand correlation matrix among the family members. When a special nutritional demand conflict among the family members is detected, a cross-member nutritional compensation mechanism is automatically triggered to generate a coordinated nutritional demand signal, which is transmitted to the recommendation generation module to achieve overall nutritional balance of the family.

[0013] In an embodiment of the present application, the management module analyzes food material images collected by the kitchen inventory monitoring device through an image recognition unit, identifies the types of food materials, and records the storage time. When the shelf life of the food material is detected to be close to a threshold, an inventory warning signal is generated. In combination with a preset food material consumption rate model, the future N-day inventory gap is predicted, a real-time inventory signal containing a warning identifier is generated, and the signal is transmitted to the recommendation generation module to drive the system to preferentially recommend a recipe scheme that consumes food materials close to the shelf life.

[0014] In an embodiment of the present application, after receiving the nutritional demand signal and the real-time inventory signal, the recommendation generation module first screens a candidate recipe set from the recipe database based on a food material matching degree, and then performs multi-round iterative optimization on the candidate recipe through a genetic algorithm. In each iteration process, the food material combination is adjusted according to the replacement instruction signal fed back by the taboo replacement module, and the initial recommended recipe that meets the nutritional constraint and has the highest inventory matching degree is output.

[0015] In an embodiment of the present application, the contraindication replacement module is built-in a food material replacement knowledge graph, which contains the physicochemical properties, taste characteristics and nutritional composition of food materials; when it is detected that there is a food material in conflict with the preset dietary taboo rule in the initial recommended menu, the replaceable food material is searched in the knowledge graph based on semantic similarity, the replacement instruction signal containing the replacement path and the nutritional compensation scheme is generated and fed back to the recommendation generation module.

[0016] In an embodiment of the present application, the context adaptation module obtains real-time physiological indicators of family members through a user behavior monitoring device, including heart rate variability, sleep duration and step data; obtains temperature and humidity and seasonal information through an environmental sensor; when it is monitored that the physiological indicators deviate from the health baseline or the environmental parameters mutate, a dynamic adjustment signal for the calorie intake, vitamin proportion and water supplement demand is generated, triggering the recommendation generation module to real-time correct the nutritional ingredients of the menu.

[0017] In an embodiment of the present application, in the closed-loop iteration process, the recommendation generation module performs the following operations: after receiving the replacement instruction signal fed back by the contraindication replacement module, the nutritional matching degree of the menu is recalculated, and if the nutritional deviation exceeds the tolerance threshold, a new round of optimization is triggered; at the same time, the dynamic adjustment signal of the context adaptation module is received to perform calorie compensation calculation, and the iteration is terminated and the final menu is output only when the menu meets the nutritional demand signal constraint, real-time inventory signal availability and dynamic adjustment signal adaptability at the same time.

[0018] In an embodiment of the present application, the management module is further connected to a local supermarket price database, and according to the food material gap data in the real-time inventory signal, a periodic purchase list is generated combined with the historical consumption record; the list divides the purchase priority according to the nutritional category, and marks the high cost-effective substitute food material options in the season, and outputs through a visual interface to guide the user to optimize the purchase decision.

[0019] In an embodiment of the present application, the construction method of the food material replacement knowledge graph comprises: collecting food material replacement cases in professional nutrition literature to establish an initial relationship network, and continuously optimizing the replacement path weight through user feedback data; after the replacement instruction signal is executed, the user's satisfaction score for the replacement menu is automatically recorded and the correlation strength parameter of the knowledge graph is updated reversely.

[0020] In an embodiment of the present application, the context adaptation module is provided with an emergency scene response mechanism: when it is monitored that a family member has an acute health abnormality indicator, a high-priority dynamic adjustment signal is immediately generated, which forcibly requires the recommendation generation module to increase the density of specific nutrients in the iteration process; at the same time, a dietary intervention reminder is sent to the user terminal until the health indicators return to the normal range and the forced constraint is removed.

[0021] The application provides a family health diet recommendation system, which integrates family member health parameters through a data processing module, generates a synergistic nutritional demand signal by using a multi-objective optimization algorithm, obtains real-time inventory data through a management module in linkage with a kitchen monitoring device, outputs an initial menu in combination with the signal, automatically replaces conflicting food materials based on a knowledge graph through a taboo replacement module, dynamically corrects the menu according to environment and user behavior data through a situation adaptation module, and finally outputs a family diet scheme meeting nutritional balance, inventory matching, taboo avoidance and scene adaptation through a closed-loop iteration mechanism of the four modules. BRIEF DESCRIPTION OF DRAWINGS

[0022] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the description of the embodiments. Obviously, the drawings in the following description are only some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor.

[0023] Figure 1 It is a system architecture diagram of a family health diet recommendation system.

[0024] Figure 2 It is a schematic diagram of showing the working process of the family health diet recommendation system. DETAILED DESCRIPTION

[0025] The embodiments of the present application will be described below through specific concrete examples, and those skilled in the art can easily understand other advantages and effects of the present application from the content disclosed in the specification. The present application can also be implemented or applied through other different specific embodiments, and each detail in the specification can be modified or changed based on different views and applications without departing from the spirit of the present application. It should be noted that the following embodiments and features in the embodiments can be combined with each other without conflict.

[0026] It should be noted that the diagrams provided in the following embodiments only illustrate the basic concept of the present application in a schematic manner, and the diagrams only show the components related to the present application, not the number, shape and size of the components when actually implemented. The actual implementation of each component may be randomly changed in shape, number and proportion, and the layout pattern of the components may also be more complex.

[0027] In the following description, a large number of details are discussed to provide a more thorough explanation of the embodiments of the present application, however, it is obvious to those skilled in the art that the embodiments of the present application can be implemented without these specific details, and in other embodiments, the known structures and devices are shown in the form of block diagrams instead of details to avoid making the embodiments of the present application difficult to understand.

[0028] Referring to Figure 1 , it is shown that the health diet recommendation system based on family, including data processing module, management module, recommendation generation module, taboo replacement module and context adaptation module. The data processing module collects and stores the health parameters of family members, generates a nutritional demand signal based on a multi-objective optimization algorithm; the management module connects the kitchen inventory monitoring device to obtain the food material type and shelf life data, and generates a real-time inventory signal; the recommendation generation module receives the nutritional demand signal and the real-time inventory signal, and generates an initial recommended recipe in combination with a preset recipe database; the taboo replacement module analyzes the replacement of taboo food materials in the initial recommended recipe according to the preset dietary taboo rules, generates a replacement instruction signal and feeds back to the recommendation generation module to output the final recipe; the context adaptation module, the context adaptation module accesses external environment sensors and user behavior monitoring devices, generates a dynamic adjustment signal and transmits it to the recommendation generation module, triggers the correction of the recipe calorie and nutritional ingredients based on the real-time scene; wherein the recommendation generation module and the taboo replacement engine, the context adaptation module form a closed loop iteration, until the final recipe that meets the nutritional demand, inventory state and dynamic scene constraints is output.

[0029] Figure 1As shown, the data processing module serves as the system input terminal, collecting structured health parameters of family members through user interfaces or Internet of Things devices, including age, gender, basal metabolic rate, chronic disease records, and exercise habit data. The module has a multi-objective optimization algorithm engine built-in, which converts individual needs into a family collaborative objective function by establishing a correlation matrix of family members' nutritional needs, and finally outputs a quantitative nutritional demand signal. This signal contains the total daily demand range of macronutrients such as protein, fat, and carbohydrates, as well as the member allocation weights of micronutrients such as calcium, iron, and vitamins. The management module connects with kitchen smart devices (such as refrigerator image sensors and electronic scales) through a data interface, obtaining real-time food material image and weight data. The image recognition unit analyzes the food material types and correlates with the local food material database to obtain the standard shelf life. By comparing the food material storage timestamp with the current time, the remaining shelf life percentage is calculated. When the remaining shelf life of a certain food material is detected to be below the set threshold, the early warning mechanism is activated, generating a real-time inventory signal with priority labels. This signal contains a list of available food materials, identification of food materials approaching expiration, and predicted inventory gaps.The recommendation generation module receives the above-mentioned nutritional demand signal and real-time inventory signal, and first performs primary screening in the preset recipe database: candidate recipes with a matching degree between recipe nutritional data and the nutritional demand signal higher than a baseline value are included in the set, and then secondary filtering is performed based on the availability of ingredients in the real-time inventory signal; the candidate set after filtering is optimized by a genetic algorithm for multiple rounds, a comprehensive score including nutritional balance, inventory matching degree, and cooking complexity is calculated for each iteration, and an initial recommended recipe is output; the taboo replacement module receives the initial recommended recipe, calls a preset dietary taboo rule library (including allergens, religious taboos, personal preferences, and other constraint conditions) to perform conflict detection; when a conflict ingredient is identified, a replacement engine based on a knowledge graph is activated: the graph stores three-dimensional indicators including physicochemical property similarity (such as a protein content difference ≤15%), taste feature correlation (trained through sensory evaluation data), and cooking function substitution (such as a swelling agent substitution relationship), retrieves the best replacement path according to the weighted similarity, and generates a replacement instruction signal - the signal includes the replaced ingredient, the recommended replacement ingredient, the nutritional compensation coefficient, and the cooking process adjustment suggestion; the context adaptation module accesses smart wearable devices and environmental sensor networks through Bluetooth / WiFi protocols, and analyzes user heart rate variability, sleep stage data, daily step count, and environmental temperature and humidity in real time; when physiological indicators deviate from the health baseline (such as heart rate higher than the resting value by 20% for 2 consecutive hours) or environmental mutations (such as a temperature drop of 5°C), a dynamic nutritional correction model is triggered to generate a dynamic adjustment signal - the signal includes the heat increase / decrease ratio, the vitamin supplement type, and the water supplement suggestion value; the above-mentioned modules form a closed-loop workflow: after the recommendation generation module transmits the initial recommended recipe to the taboo replacement module, it receives the replacement instruction signal to reconstruct the recipe; the reconstructed recipe is transmitted to the context adaptation module for real-time correction; if the corrected recipe deviates from the nutritional demand signal constraint as determined by the nutritional balance verification engine, it returns to the recommendation generation module to start a new round of optimization, and outputs the final recipe to the user terminal until all constraint conditions are met simultaneously.

[0030] Further, the health parameter type extension covers four dimensions: 1) basic physiological data (age, gender, height and weight, body fat rate), 2) health status data (chronic disease records input by the user, such as diabetes, hypertension, and clinical test indicators, such as fasting blood glucose value), 3) behavior habit data (daily exercise intensity, work nature physical consumption level collected through questionnaires), and 4) special period identifier (such as pregnant woman marker, postoperative recovery period marker); the multi-objective optimization algorithm operation process includes three stages: the first stage constructs a member independent nutrition model, calculates the individual daily heat demand according to the basal metabolic rate, and sets the nutrition limit condition according to the chronic disease record (such as the diabetic patient limits the proportion of carbohydrate intake); the second stage establishes a family collaborative correlation matrix, and defines the compensation relationship of conflicting nutrients (for example, when the calcium demand of children and the low-fat demand of the elderly conflict, the high-calcium and low-fat are met by increasing the intake of soy products); the third stage adopts a Pareto optimal solution search algorithm, maximizes the overall nutrition balance index of the family under the premise of meeting the nutrition bottom line of all members, and finally outputs the nutrition demand signal carrying the member allocation weight; the cross-member nutrition compensation mechanism is a core innovation point, and its implementation manner is: when the algorithm detects that there are more than two members with irreconcilable nutritional conflicts (such as the conflict between the folic acid demand of pregnant women and the dislike of spinach), a three-level processing flow is started: 1) find alternative sources among similar nutrient foods, 2) compensate for the gap through a nutritional supplement scheme (add vitamin tablet taking suggestions in the recipe note column), and 3) trigger time staggered allocation; the coordinated post-nutrition demand signal output by the mechanism contains conflict resolution scheme encoding, which is directly called by the recommendation generation module.

[0031] In an embodiment of the present application, the inventory monitoring and early warning function of the management module specifically includes:

[0032] The image recognition unit technical architecture consists of three sub-modules: 1) image preprocessing sub-module, which denoises, corrects brightness, and splices multi-angle for the original images collected by the kitchen monitoring device; 2) feature extraction sub-module, which uses a convolutional neural network to identify the morphological features of food materials (such as the granular crown of broccoli and the texture of salmon), and combines with OCR technology to read the packaging date label; 3) semantic association sub-module, which matches the recognition results with the local food material database (for example, mapping "round red fruit body" to tomatoes), and associates the standard nutritional data and typical shelf life of the food material; the shelf life warning logic adopts a dynamic threshold mechanism: for perishable food materials such as poultry and meat, set the yellow warning when the remaining shelf life is ≤30%; for storage-resistant food materials such as dry goods and seasonings, set the blue warning when the remaining shelf life is ≤10%; after the warning signal is generated, the system automatically performs three-level operations: 1) mark the warning food materials and their safe edible remaining days in the real-time inventory signal, 2) predict the future three-day inventory gap based on the historical consumption rate (for example, predict the milk gap amount according to the average consumption in the past two weeks), 3) generate a "priority consumption list" and embed it in the inventory signal to the recommendation generation module; the principle of inventory-driven menu optimization is reflected in that: after the recommendation generation module analyzes the real-time inventory signal, it performs weighted sorting on the initial menu set - the menu containing warning food materials gets the highest priority weight (+50%), the menu containing regular inventory food materials gets the basic weight, and the menu needing to purchase food materials reduces the weight below the baseline value; at the same time, introduce the inventory turnover rate optimization algorithm to ensure that high-warning-level food materials are recommended to be consumed within 48 hours, and the actual measurement shows that this mechanism can reduce the household food material waste rate by 27%.

[0033] As shown in Figure 1 , the optimization decision mechanism of the recommendation generation module is further refined into a three-level processing flow:

[0034] The primary screening stage is implemented by a two-channel matching engine: the first channel calculates the similarity between the nutrient target value (such as the protein daily requirement interval) in the nutritional demand signal and the nutrition label in the preset recipe database, and only retains candidate recipes with a matching degree higher than the system set threshold; the second channel analyzes the list of available ingredients in the real-time inventory signal, eliminates recipes containing out-of-stock ingredients or requiring the purchase of ingredients exceeding the tolerance upper limit, and generates a preliminary candidate set. The genetic algorithm optimization stage uses a chromosome coding scheme: candidate recipes are deconstructed into gene sequences (main ingredient gene segment + auxiliary material gene segment + cooking method gene segment), and after initializing the population, selection, crossover, and mutation operations are performed; the fitness function calculates three indicators: 1) nutritional balance degree - based on the Euclidean distance between the nutritional data of the candidate recipe and the nutritional demand signal, 2) inventory matching degree - according to the ingredient warning level in the real-time inventory signal (the weight coefficient of near-expired ingredients is increased to 1.8 times that of regular ingredients), 3) cooking cost - associated with recipe preparation time and utensil complexity parameters; after each iteration, the top 20% of individuals are retained, and when the improvement rate of the optimal solution is less than 0.5% for three consecutive generations, the optimization is terminated, and the initial recommended recipe is output. The alternative driving adjustment stage is started after receiving the alternative instruction signal from the taboo alternative module: analyze the alternative path indication (such as the "pork → mushroom" alternative pair) and the nutritional compensation coefficient (such as the protein compensation amount + 0.2g / 100g) in the signal, perform ingredient replacement on the initial recommended recipe, and recalculate the nutritional data; if the nutritional deviation after replacement exceeds the fault tolerance threshold, trigger local re-optimization - maintain the original cooking method gene segment unchanged, and only perform mutation operations on the auxiliary material gene segment to fine-tune the taste matching degree, finally generate a reconstructed recipe and transmit it to the context adaptation module.

[0035] As Figure 1As shown, the knowledge graph architecture of the taboo substitution module contains three types of elements: entity nodes, relationship edges, and dynamic weight parameters. Entity nodes cover all replaceable objects in the ingredient library, and each node stores standardized nutritional data and sensory feature vectors; relationship edges define three types of substitution logic: physicochemical substitution relationships are connected based on the similarity matrix of nutritional components, taste substitution relationships are generated by training a flavor vector space model through a sensory evaluation database, and functional substitution relationships record the role interchangeability of ingredients in cooking scenarios; the initial values of the weight parameters are derived from authoritative nutrition literature, and a real-time update interface is set. The substitution decision engine operates in a four-step process: the conflict detection unit transmits the initial recommended menu to the rule matcher, compares it with the preset dietary taboo rule library, and outputs the conflict ingredient identification and taboo type code; the substitution path retriever takes the conflict ingredient as the starting point and performs a breadth-first search along the three types of relationship edges in the knowledge graph, returning a set of candidate substitution solutions; the multi-dimensional scorer calculates the comprehensive score of the candidate solutions, where the physicochemical matching score uses the Euclidean distance algorithm, the taste similarity score calls a pre-trained flavor neural network model, and the functional equivalence score is based on a cooking scenario matching table; when the optimal substitution solution deviates from the original ingredient nutrition by more than the tolerance value, the compensation generator automatically adds a nutrition balancing strategy, and finally packages it into a substitution instruction signal containing the substitution path, compensation solution, and cooking adjustment suggestions, which is fed back to the recommendation generation module after verification. The graph self-optimization mechanism is realized through a user feedback loop: the system collects user satisfaction ratings and manual modification records for the substituted menu, positive feedback enhances the weight coefficients of the corresponding relationship edges, negative feedback triggers a relationship edge reevaluation process, and when a specific edge weight falls below the failure threshold, the substitution path is automatically frozen.

[0036] Further, the dynamic adjustment mechanism of the context adaptation module is realized through a dual-modal data channel. The physiological indicator response channel accesses the data stream of the smart wearable device, including motion intensity data collected by a three-axis acceleration sensor, heart rate variability parameters collected by an optical heart rate sensor, and metabolic state indicators collected by a body surface temperature sensor; when the motion intensity continuously falls below the historical mean by 25% for more than six hours, a heat down-regulation coefficient signal is generated; when the SDNN value of heart rate variability is continuously below 50 milliseconds for two hours, the proportion of pressure response type nutrients is increased; when the deep sleep ratio decreases by 20% compared with the reference value, a tryptophan supplement instruction is generated and transmitted to the recommendation generation module. The environmental parameter response channel integrates a multi-source sensor network: a temperature and humidity sensor monitors the indoor and outdoor temperature difference gradient, an air quality sensor captures PM2.5 and pollen concentration, and a seasonal clock module provides the corresponding solar term characteristics of the geographical location; when a temperature drop of more than five degrees Celsius is detected, a high heat density menu generation instruction is activated; when the pollen concentration exceeds the threshold of allergy, the easily allergenic food is automatically shielded and an anti-histamine nutrient suggestion is added. The signal fusion unit processes the dual-channel instructions using a weighted decision matrix, the priority weight of the physiological indicator channel is set to 0.7, and the weight of the environmental channel is 0.3; when the dual-channel instructions conflict, the high-priority instruction is executed and a compensation note is added in the dynamic adjustment signal, for example, when the high-motion demand conflicts with the high-temperature environment, the heat-increasing instruction is executed while adding the note item of “supplementing electrolytes after meals”, and the finally generated dynamic adjustment signal is transmitted to the recommendation generation module to trigger the menu correction.

[0037] As Figure 1As shown, the closed-loop iteration mechanism realizes flow control through a five-stage state machine. State 0 is the initialization stage: after the recommendation generation module completes initial recipe generation, it sends a detection request signal to the taboo replacement module and enters the detection state. State 1 performs taboo replacement detection: the taboo replacement module completes recipe scanning within 200 milliseconds, and if there are conflicting food materials defined in the rule library, it generates a replacement instruction signal and jumps to state 2; otherwise, it directly jumps to state 3. State 2 performs replacement reconstruction: the recommendation generation module parses the replacement instruction signal to perform recipe update, and the updated nutrition verification engine detects whether the nutrient deviation exceeds 8%; if not, it returns to state 1 for re-detection; if it does, it starts a local optimization sub-process - fixing the main food material gene segment, randomly selecting a variation point in the auxiliary material gene segment to insert a replacement food material, and cycling optimization until the nutrition constraints are met. State 3 performs context adaptation detection: the context adaptation module compares the current environment / physiological data with the health baseline model, and if it needs to be corrected, it generates a dynamic adjustment signal and jumps to state 4; otherwise, it directly jumps to state 5. State 4 implements dynamic correction: the recommendation generation module adjusts the nutrient ratio of the recipe according to the dynamic adjustment signal, and if the correction results in a stock matching degree less than 60%, it activates the stock re-matching sub-process - replacing the food material with a sufficient stock in the same category. State 5 performs termination verification: the system simultaneously starts four verification checks, the nutrition verification unit detects whether the final recipe meets the matching degree of the nutrition demand signal, the stock verification unit confirms that all required food materials exist in the real-time stock signal available list, the taboo verification unit performs a second rule scan, and the scene verification unit verifies the dynamic adjustment instruction execution state; if any verification fails, it records an error code and returns to state 0 to restart the process, and if all verifications pass, it outputs the final recipe to the user terminal and generates a timestamped execution log.

[0038] The present application is a family health diet recommendation system based on the integration of family member health parameters through the data processing module, the use of multi-objective optimization algorithm to generate collaborative nutrition demand signal; the management module links the kitchen monitoring equipment to obtain real-time inventory data; the recommendation generation module outputs the initial recipe in combination with the above signals; the taboo replacement module automatically replaces the conflicting food materials based on the knowledge graph; the context adaptation module dynamically corrects the recipe according to the environmental and user behavior data; the four modules are iterated through a closed-loop mechanism, and finally output a family diet plan that meets the nutritional balance, inventory matching, taboo avoidance and scene adaptation.

[0039] Therefore, through the family health diet recommendation system based on the present application, the problem of collaborative recommendation of family multi-member differentiated nutritional needs and real-time scene can be solved.

[0040] The above embodiments are only illustrative of the principles of the present application and its efficacy, and are not intended to limit the present application. Any modification or change made by any person skilled in the art without departing from the spirit and scope of the present application shall be covered by the claims of the present application.

Claims

1. A home-based health diet recommendation system based on, characterized by, The application relates to a meal recommendation system, comprising: a data processing module that collects and stores health parameters of family members and generates a nutrition demand signal based on a multi-objective optimization algorithm; a management module that connects a kitchen inventory monitoring device to obtain food material type and shelf life data and generates a real-time inventory signal; a recommendation generation module that receives the nutrition demand signal and the real-time inventory signal and generates an initial recommended recipe in combination with a preset recipe database; after receiving the nutrition demand signal and the real-time inventory signal, the recommendation generation module first screens a candidate recipe set from the recipe database based on a food material matching degree, and then performs multi-round iterative optimization on the candidate recipe through a genetic algorithm; in each iteration process, the food material combination is adjusted according to a replacement instruction signal fed back by a taboo replacement module, and an initial recommended recipe that satisfies a nutrition constraint and has the highest inventory matching degree is output; in the iteration process, the recommendation generation module performs the following operations: after receiving the replacement instruction signal fed back by the taboo replacement module, the recipe nutrition matching degree is recalculated, and if the nutrition deviation exceeds a tolerance threshold, a new round of optimization is triggered; meanwhile, a dynamic adjustment signal of a situation adaptation module is received to perform calorie compensation calculation, and the iteration is terminated and the final recipe is output only when the recipe simultaneously satisfies the nutrition demand signal constraint, the real-time inventory signal availability and the dynamic adjustment signal adaptability; a taboo replacement module that performs replacement analysis on taboo food materials in the initial recommended recipe according to preset dietary taboo rules, generates a replacement instruction signal and feeds back the signal to the recommendation generation module to output the final recipe; the taboo replacement module is internally provided with a food material replacement knowledge graph, which contains the correlation between the physical and chemical properties, taste characteristics and nutritional composition of food materials; when it is detected that there are food materials in the initial recommended recipe that conflict with the preset dietary taboo rules, replaceable food materials are searched in the knowledge graph based on semantic similarity, a replacement instruction signal containing a replacement path and a nutrition compensation scheme is generated and fed back to the recommendation generation module; a situation adaptation module that accesses external environment sensors and user behavior monitoring devices, generates a dynamic adjustment signal and transmits the signal to the recommendation generation module, and triggers recipe calorie and nutrient component correction based on real-time scenes; wherein the recommendation generation module, the taboo replacement engine and the situation adaptation module form a closed-loop iteration until the final recipe that meets the nutrition demand, inventory state and dynamic scene constraints is output.

2. The health diet recommendation system based on home according to claim 1, wherein, The health parameters collected in the data processing module include age, gender, basal metabolic rate, chronic disease history and exercise habit data of family members, the multi-objective optimization algorithm generates differentiated nutrition allocation signals for different members by establishing a nutrition demand correlation matrix among the family members; when it is detected that there is a special nutrition demand conflict among the family members, a cross-member nutrition compensation mechanism is automatically triggered to generate a coordinated nutrition demand signal, and the signal is transmitted to the recommendation generation module to realize overall nutrition balance of the family.

3. The health diet recommendation system based on home according to claim 1, characterized in that, The management module analyzes the food material images collected by the kitchen inventory monitoring device through the image recognition unit, identifies the types of food materials and records the storage time; generates an inventory warning signal when the shelf life of the food material is detected to approach the threshold, combines the preset food material consumption rate model to predict the future N-day inventory gap, generates a real-time inventory signal containing a warning mark and transmits it to the recommendation generation module, and drives the system to preferentially recommend a recipe scheme for consuming food materials approaching the shelf life.

4. The health diet recommendation system based on home according to claim 1, wherein, The context adaptation module obtains real-time physiological indicators of family members through user behavior monitoring devices, including heart rate variability, sleep duration, and step count data; obtains temperature and humidity and seasonal information through environmental sensors; when the physiological indicators deviate from the health baseline or the environmental parameters mutate, generates dynamic adjustment signals for calorie intake, vitamin proportion, and water supplement demand, triggering the recommendation generation module to make real-time corrections to the nutritional ingredients of the recipe.

5. The home-based healthy diet recommendation system according to claim 1, wherein, The management module is further connected to a local supermarket price database, and generates a periodic purchase list according to the food material gap data in the real-time inventory signal, combined with historical consumption records; the list divides the purchase priority according to the nutritional category, and marks the high cost-effective substitute food material options in the season, and outputs through a visual interface to guide the user to optimize the purchase decision.

6. The home-based healthy diet recommendation system according to claim 1, wherein, The construction method of the food material substitution knowledge graph includes: collecting food material substitution cases in professional nutrition literature to establish an initial relationship network, continuously optimizing the substitution path weight through user feedback data; after the execution of the substitution instruction signal, automatically record the user's satisfaction score for the substitution recipe and update the correlation strength parameters of the knowledge graph in reverse.

7. The health diet recommendation system based on home according to claim 1, wherein, The context adaptation module is provided with an emergency scene response mechanism: when acute health abnormal indicators of family members are detected, high-priority dynamic adjustment signals are immediately generated, and the recommendation generation module is forced to increase the density of specific nutrients in the iteration process; at the same time, send a dietary intervention reminder to the user terminal until the health indicators return to the normal range and the forced constraint is removed.

Images