Intelligent refrigerator article storage management method and system

Abstract

The invention provides an intelligent refrigerator article storage management method and system. The method comprises the steps that accurate basic data is provided for dynamic storage optimization and healthy diet recommendation; the partition and environmental parameters are dynamically adjusted according to the storage requirements of the food materials, and the preservation time is prolonged; the fresh-keeping state and the storage environment of the food materials are evaluated in real time, and the partition optimization effect is ensured; recording food material taking behaviors of the user, and generating a healthy intake model; and generating personalized diet recommendation and reminding in combination with the health demand and the storage state of the user. According to the intelligent refrigerator article storage management method and system, the food material storage efficiency is optimized through the intelligent technology, the food material utilization rate is increased, and meanwhile the personalized healthy diet requirement of a user is met.

Description

Technical Field

[0001] The present invention belongs to the field of intelligent refrigerator article storage management, and in particular relates to an intelligent refrigerator article storage management method and system. Background Art

[0002] In modern families, refrigerators, as core equipment for food storage, have become an indispensable part of life. However, with the continuous changes in family dietary structure and user needs, the functional limitations of traditional refrigerators have become increasingly prominent. On the one hand, the trend of diversified food types has led to significant differences in the storage conditions required by different ingredients. For example, vegetables and fruits usually require higher humidity to stay fresh, while meat, dairy products, etc. require lower temperatures to prevent corruption. However, traditional refrigerators mostly use a fixed temperature and humidity zoning design, which cannot be dynamically adjusted according to the specific needs of the stored items, resulting in some ingredients spoiling prematurely due to poor storage conditions. In addition, traditional refrigerators also generally have the problem of insufficient optimization in terms of space utilization efficiency. Users often face the situation where ingredients are stacked in a messy manner and cannot be effectively accessed, which further aggravates the waste of ingredients.

[0003] On the other hand, with the improvement of users' health awareness, people have put forward higher requirements for the scientificity and rationality of their diet. However, most refrigerators on the market can only provide simple storage functions and do not have the ability to guide users on healthy eating. For example, it is usually difficult for users to clearly understand the current status of ingredients in the refrigerator (such as shelf life, remaining quantity) and how to match healthy meals based on existing ingredients. In addition, traditional refrigerators lack the ability to dynamically analyze users' eating habits, cannot make dietary recommendations based on personalized needs, and cannot effectively avoid waste caused by forgetting to use ingredients. Although some smart refrigerators have introduced basic item recognition and management functions in recent years, these functions are mainly focused on static information recording and display, and lack the ability to dynamically adjust storage conditions and deeply optimize health management.

[0004] Therefore, the existing technology still has significant deficiencies in the scalability of refrigerator functions, the intelligence of food management and the support of user health needs. Summary of the invention

[0005] The purpose of the present invention is to provide a smart refrigerator item storage management method and system, which overcomes the defects of traditional technology through a number of innovative designs.

[0006] In order to achieve the above object, a first aspect of the present invention provides a method for managing storage of items in a smart refrigerator, the method comprising:

[0007] S1. Collect the internal data of the refrigerator and the food nutrition label, wherein the internal data includes the item image, the item RFID tag information, the item weight and the refrigerator partition environment parameters, and identify the item image according to the preset multi-label classification model, and generate the intelligent storage label data set according to the identification result combined with the item RFID tag information and the item weight;

[0008] Among them, the recognition result includes the name of the food, the category of the food and the recognition confidence; the RFID tag information of the item includes the shelf life and production date marked on the package, and the remaining shelf life calculated in combination with the current time, and the recommended storage temperature range and recommended storage humidity range for food storage are obtained from the domain food database; the intelligent storage tag data set includes the name of the food, the category of the food, the recommended storage temperature range, the recommended storage humidity range, the remaining shelf life, the item weight and the storage priority; the refrigerator partition environmental parameters include the current temperature and humidity;

[0009] S2. Initialize refrigerator partitions and design a dynamic allocation model for refrigerator partitions. The dynamic allocation model is used to maximize the storage and preservation effect of food and minimize the overall energy consumption of the refrigerator. The intelligent storage label data set and the initialized refrigerator partitions are used as inputs of the dynamic allocation model to perform partition optimization calculations to obtain optimized partition results. According to the optimized partition results, a storage mapping table for each food is generated to obtain the partition number to which the food is assigned, the actual adjusted temperature and humidity parameters of the partition, and the storage priority;

[0010] S3, regularly obtain the real-time environmental parameters of each partition through the partition sensor and calculate the deviation between the actual environment of the front partition and the target storage condition, and update the remaining shelf life of each food. Calculate the food freshness index based on the deviation and the remaining shelf life to evaluate the status of each food in real time;

[0011] If the partition sensor detects an abnormal deviation, and the abnormal deviation exceeds a preset threshold, an abnormal adjustment process is triggered to generate a new food storage mapping table and partition environment parameters;

[0012] S4. When the user takes food, the real-time weight of each time the food is taken is recorded and the change in the amount of food taken each time is calculated. Each taking behavior is recorded according to the change in the amount of food taken, and the taking timestamp and the name of the food are associated to form an intake log. The nutritional value of each intake is calculated according to the change in the amount of food taken and the nutritional components of the corresponding food, and the nutritional value of each intake is input into a predefined user health model to analyze the user's dietary status in real time;

[0013] S5. Generate personalized dietary recommendations and reminders based on user health needs and storage status.

[0014] Furthermore, the loss function of the preset multi-label classification model is It is expressed as:

[0015]

[0016] in, is the cross entropy loss; S j For the confidence of each category, low confidence categories contribute more loss; λ is the regularization weight, which controls the regularization effect; ∈ prevents the denominator from being zero; n is the total number of categories;

[0017] The storage priority P i Calculated based on the remaining shelf life and category weight, expressed as:

[0018]

[0019] in, is the class weight, α controls the impact of the class weight, D exp For the remaining shelf life.

[0020] Furthermore, the objective function of the dynamic allocation model is as follows:

[0021]

[0022] Among them, Φ is the objective function, the overall optimization value; is an indicator function, indicating whether the i-th ingredient is assigned to partition Z k ; ΔT k =|T k -T opt (Z k )|, ΔT k is the current partition temperature T k and demand T opt (Z k ) deviation; ΔH k =|H k -H opt (Z k )|, ΔH k is the current zone humidity H k With demand H opt (Z k ) deviation; β,γ respectively represent the influence weights of temperature and humidity deviation on the optimization target;

[0023] Each partition Z k The target temperature and humidity are calculated using the weighted average of ingredients:

[0024]

[0025] Among them, T opt ,H optRecommended storage temperature and humidity, which come from the smart storage label dataset for each ingredient i is the current ingredient i; W i is the weight of the food;

[0026] At the same time, the freshness protection regularization term is introduced into the dynamic allocation model Prioritize the allocation of ingredients with the shortest shelf life to the best environment:

[0027]

[0028] The regularization term increases the remaining shelf life D exp The inverse of the square of , so that the remaining ingredients with short shelf life are allocated to the most suitable partition first to avoid waste.

[0029] Furthermore, the partition optimization calculation includes:

[0030] Store all ingredients according to the priority P i Arrange in descending order, giving priority to ingredients with higher priority;

[0031] Use the greedy strategy combined with the objective function to gradually allocate ingredients to ensure the partition capacity V k Not overloaded;

[0032] Calculate the target temperature and humidity T for each zone opt (Z k ) and H opt (Z k ), adjust the current refrigerator partition environmental parameters T k ,H k , so that it gradually approaches the target value;

[0033] If the devices in some partitions cannot reach the target value, re-evaluate the low storage priority P i Food distribution ensures that the freshness requirements of high-priority ingredients are met.

[0034] Furthermore, the method of calculating the food freshness index according to the deviation and the remaining shelf life to evaluate the state of each food in real time includes:

[0035] A nonlinear freshness index model combining environmental deviation and shelf life is used to evaluate the status of each ingredient in real time:

[0036]

[0037] Among them, α, β represent the weight of the deviation on the index, which is used to control the influence of temperature and humidity; ∈ represents a small value to prevent the denominator from being zero; It indicates the nonlinear effect of temperature deviation on freshness preservation. The greater the deviation, the greater the effect on B i The more significant the reduction; Used to reflect the direct impact of the remaining shelf life on the freshness status.

[0038] Introducing regularization terms Consider the environmental conflict effects between multiple ingredients and give priority to protecting high-priority ingredients:

[0039]

[0040] The regular term Assign higher environmental resource adjustment weights to ingredients with greater storage priority conflicts in the same partition to avoid negative impacts on key ingredients.

[0041] Furthermore, the food nutrition label includes the basic nutritional components of each food, specifically including the gram content of calories, protein, fat, and carbohydrates;

[0042] After the timestamp and the name of the food are combined to form the intake log, the change in the amount of food consumed ΔW is calculated. i Update remaining weight of ingredients W i :

[0043] W i (t) = W i (t-Δt)-ΔW i

[0044] At the same time, according to the current time T cur and storage start time T start Update remaining shelf life D rem :

[0045] D rem =D exp -(T cur -T start )

[0046] Simultaneously monitor D rem The critical value of D rem ≤δ exp , marking the ingredient as about to expire, triggering a priority recommendation strategy.

[0047] Furthermore, the nutritional value of each intake is calculated as follows:

[0048] The change in the amount of food consumed ΔW i And the nutritional content of the corresponding ingredients η i , calculate the nutritional value of each intake:

[0049]

[0050] in, Include total calories and total macronutrients;

[0051] Introducing freshness adjustment factor γ i , used to simulate the effect of food freshness on nutrient absorption:

[0052] γ i =exp(-λ·B i )

[0053] Among them, λ is the adjustment parameter, which controls the weight of the freshness index on the nutritional correction; γ i The smaller it is, the lower the freshness is and the greater the impact on the nutritional value is;

[0054] Based on gamma i Correct the nutritional value to get the actual nutritional intake

[0055]

[0056] The user health model The update rules are as follows:

[0057]

[0058] Where f is the update function and Δt is the change time.

[0059] Furthermore, the S5 specifically includes:

[0060] Define optimization goals for dietary recommendations;

[0061] Combining user health needs, food freshness and inventory status, the objective function Φ of dietary recommendation is designed:

[0062]

[0063] Among them, η cal,i is the heat of food i; ΔW i is the recommended amount of ingredients; α, β are weight coefficients, which control the influence of freshness and shelf life respectively, B i It is the food freshness index;

[0064] The Lagrange multiplier method is used to integrate constraints such as dietary balance and waste reduction into the objective function, and the recommended solution is obtained through optimization:

[0065]

[0066] in As an indicator function, ensure that ingredients with sufficient inventory are recommended first;

[0067] Generate a meal plan with specific recipes and usage amounts based on the optimization results It also generates smart reminders to remind users through the refrigerator screen or mobile app, providing users with personalized operation suggestions.

[0068] Furthermore, the constraints of the optimization objective are:

[0069] Ensure nutritional balance of recommended recipes and user health model The target value in Closest to:

[0070]

[0071] Among them, ∈ cal ,∈ pro is the acceptable deviation range;

[0072] Give priority to ingredients that are about to expire:

[0073]

[0074] Among them, ∈ is a small value to prevent the denominator from being zero;

[0075] To avoid insufficient stock of ingredients, add the following recommendations:

[0076]

[0077] Among them, W min The minimum stock required for the recommended dish.

[0078] In a second aspect of the present invention, a smart refrigerator article storage management system is provided, the system comprising:

[0079] A refrigerator data collection unit is used to collect the internal data of the refrigerator and food nutrition labels, wherein the internal data includes item images, item RFID tag information, item weight, and refrigerator partition environment parameters, and recognize the item images according to a preset multi-label classification model, and generate an intelligent storage label data set based on the recognition results combined with the item RFID tag information and item weight;

[0080] Among them, the recognition result includes the name of the food, the category of the food and the recognition confidence; the RFID tag information of the item includes the shelf life and production date marked on the package, and the remaining shelf life calculated in combination with the current time, and the recommended storage temperature range and recommended storage humidity range for food storage are obtained from the domain food database; the intelligent storage tag data set includes the name of the food, the category of the food, the recommended storage temperature range, the recommended storage humidity range, the remaining shelf life, the item weight and the storage priority; the refrigerator partition environmental parameters include the current temperature and humidity;

[0081] A refrigerator partition storage unit is used to initialize refrigerator partitions and design a dynamic allocation model for refrigerator partitions. The dynamic allocation model is used to maximize the storage and preservation effect of food ingredients and minimize the overall energy consumption of the refrigerator. The intelligent storage label data set and the initialized refrigerator partitions are used as inputs of the dynamic allocation model to perform partition optimization calculations to obtain optimized partition results. According to the optimized partition results, a storage mapping table for each food ingredient is generated to obtain the partition number to which the food ingredient is assigned, the actual adjusted temperature and humidity parameters of the partition, and the storage priority.

[0082] The refrigerator partition storage optimization unit is used to regularly obtain the real-time environmental parameters of each partition through the partition sensor and calculate the deviation between the actual environment of the front partition and the target storage conditions, and update the remaining shelf life of each food. The food preservation index is calculated based on the deviation and the remaining shelf life to evaluate the status of each food in real time;

[0083] If the partition sensor detects an abnormal deviation, and the abnormal deviation exceeds a preset threshold, an abnormal adjustment process is triggered to generate a new food storage mapping table and partition environment parameters;

[0084] The refrigerator partition access change adaptation unit is used to record the real-time weight of each time the food is taken and calculate the change in the amount of food taken each time when the user takes food, record each access behavior according to the change in the amount of food taken, associate the access timestamp and the name of the food, form an intake log, calculate the nutritional value of each intake according to the change in the amount of food taken and the nutritional components of the corresponding food, and input the nutritional value of each intake into a predefined user health model to analyze the user's dietary status in real time;

[0085] The personalized setting unit is used to generate personalized dietary recommendations and reminders based on the user's health needs and storage status.

[0086] The beneficial technical effects of the present invention are at least as follows:

[0087] (1) The present invention introduces food attribute analysis technology to accurately classify and label the food in the refrigerator. The system generates intelligent storage labels according to the storage requirements of different food ingredients (such as temperature, humidity, and shelf life), and realizes partition storage optimization by dynamically adjusting the internal environmental parameters of the refrigerator (such as temperature and humidity). Compared with the traditional fixed partition design, dynamic storage optimization technology can greatly improve the preservation effect of food ingredients, and improve the space utilization rate of the refrigerator through reasonable layout, fundamentally solving the problem of food waste caused by mismatched storage conditions.

[0088] (2) The present invention records the user's food use behavior and combines it with the existing food data in the refrigerator to establish a user's eating habit model, and provides scientific dietary recommendations based on nutritional principles and personalized health goals (such as fat loss and muscle gain). At the same time, the system gives priority to recommending ingredients that are about to expire, helping users to reasonably match existing ingredients and reduce waste. This design not only makes up for the lack of healthy diet guidance in traditional refrigerators, but also enhances users' willingness to actively use the refrigerator management system.

[0089] (3) The present invention allows users to view food status (such as shelf life, remaining quantity, storage location) and dietary recommendations in real time through the refrigerator display or mobile device, and triggers intelligent notification functions based on user behavior, such as reminders for the use of ingredients that are about to expire and abnormal temperature and humidity in the refrigerated area. This multi-mode interactive design greatly improves the user experience and upgrades the refrigerator from a passive storage tool to an active service terminal. BRIEF DESCRIPTION OF THE DRAWINGS

[0090] The present invention is further described using the accompanying drawings, but the embodiments in the accompanying drawings do not constitute any limitation to the present invention. A person skilled in the art can obtain other drawings based on the following drawings without creative work.

[0091] Figure 1 This is a flow chart of a method for managing storage of items in a smart refrigerator according to the present invention. DETAILED DESCRIPTION

[0092] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be understood as limiting the present invention.

[0093] like Figure 1 As shown, an embodiment of the present invention provides a method for managing storage of items in a smart refrigerator, comprising the following steps:

[0094] S1. Collect the internal data of the refrigerator and the food nutrition label, wherein the internal data includes the item image, the item RFID tag information, the item weight and the refrigerator partition environment parameters, and identify the item image according to the preset multi-label classification model, and generate an intelligent storage label data set based on the identification result combined with the item RFID tag information and the item weight.

[0095] Specifically, the multimodal sensor system inside the refrigerator collects the following input data:

[0096] V i : Item images, captured by the camera, are used to extract the appearance features of the food;

[0097] R i : Item RFID tag information, used to obtain preset data in the package (such as shelf life, production date);

[0098] W i : The weight of the item, measured by the weight sensor, serves as the basis for calculating the remaining quantity;

[0099] E env : Refrigerator partition environmental parameters (current temperature, humidity).

[0100] Data {V i ,R i ,W i ,E env} is a direct reflection of the refrigerator environment and food characteristics, forming the basic data set for subsequent intelligent storage label generation.

[0101] Furthermore, the input image V i Through a customized multi-label classification model The model combines visual and label information (R i ), output the recognition result:

[0102]

[0103] Among them, N i Indicates the name of the food and extracts the food type information; C i Indicates the category of food (such as vegetables, fruits, meat, etc.), which is used for classification and partition optimization; S i Indicates recognition confidence. Results with low confidence are left to the user for manual correction.

[0104] Furthermore, in order to improve the model's adaptability to food diversity and uneven distribution, a regularization loss term is designed:

[0105]

[0106] in, is the cross entropy loss; S j For the confidence of each category, the low confidence category contributes more loss; λ is the regularization weight, which controls the regularization effect; ∈ prevents the denominator from being zero. The output of the model {N i ,C i ,S i Combined with R i Data provides support for subsequent extraction of storage characteristics and remaining shelf life.

[0107] Furthermore, from R i Extract T exp (shelf life marked on the package) and Tprod (production date), combined with the current time T cur Calculate remaining shelf life:

[0108] D exp =T exp -T cur

[0109] If R i If the data is missing or abnormal, please complete it through the user interface.

[0110] Further, query the food database based on N i and C i Determine the storage characteristics of your ingredients:

[0111] Among them, T opt Indicates the recommended storage temperature range, used for dynamic adjustment; H opt Indicates the recommended storage humidity range for zoning allocation. This step combines the classification results with environmental requirements to ensure that the storage attributes of each ingredient are complete and accurate, providing information support for generating smart labels.

[0112] Furthermore, the extracted data {N i ,C i ,D exp ,W i ,T opt ,H opt}, generate smart storage labels for each ingredient Among them, P i Indicates storage priority, calculated based on remaining shelf life and category weight:

[0113]

[0114] in, is the category weight, such as perishable food has a higher weight; α controls the impact of category weight. Smart Storage Tags It is the direct input to the dynamic storage optimization module and contains comprehensive food information and storage requirements.

[0115] Furthermore, the generated tags Stored in the refrigerator central database as input for subsequent dynamic storage partition optimization. Complete label dataset Directly drive the next operation, so that the dynamic management of the refrigerator is based on accurate basic information.

[0116] S2. Initialize refrigerator partitions and design a dynamic allocation model for refrigerator partitions. The dynamic allocation model is used to maximize the storage and preservation effect of food ingredients and minimize the overall energy consumption of the refrigerator. The intelligent storage label data set and the initialized refrigerator partitions are used as inputs of the dynamic allocation model to perform partition optimization calculations to obtain optimized partition results. According to the optimized partition results, a storage mapping table for each food ingredient is generated to obtain the partition number to which the food ingredient is assigned, the temperature and humidity parameters of the partition after actual adjustment, and the storage priority.

[0117] Specifically, the refrigerator partition is initialized to Z = {Z1, Z2, ..., Z k}, where Z k Represents the kth partition, the initial parameters of each partition: E k ={T k ,H k} indicates the current partition temperature and humidity; V k Indicates the maximum storage capacity of the partition. This initialization phase provides basic input data for subsequent dynamic partition optimization and establishes the initial state of the partition environment.

[0118] Furthermore, a dynamic partition allocation model is constructed:

[0119] The dynamic allocation model aims to maximize the storage and preservation effect of food ingredients and minimize the overall energy consumption of the refrigerator. The objective function is as follows:

[0120]

[0121] Among them, Φ represents the objective function and the overall optimization value; Represents an indicator function, indicating whether the i-th ingredient is assigned to partition Z k ; ΔT k =|T k -T opt (Z k )| represents the deviation between the current partition temperature and the demand; ΔH k =|H k -H opt (Z k )| represents the deviation between the current partition humidity and the demand; β and γ represent the influence weights of the temperature and humidity deviation on the optimization target.

[0122] Furthermore, each partition Z k The target temperature and humidity are calculated using the weighted average of ingredients:

[0123]

[0124] Where T opt ,H opt Label data from each ingredient

[0125] Furthermore, the freshness protection regularization term is introduced Prioritize the allocation of ingredients with shorter shelf life to better environments:

[0126]

[0127] The regularization term is obtained by increasing D exp The inverse of the square of , so that the remaining ingredients with short shelf life are allocated to the most suitable partition first to avoid waste.

[0128] Further, the partition optimization algorithm:

[0129] Input: Food label dataset and the initial partition state {Z,E k ,V k};

[0130] Output: Optimized partition map and updated partition environment parameters {T k ,H k}.

[0131] Specific steps:

[0132] Put all the ingredients in according to P i Arrange in descending order, giving priority to ingredients with higher priority;

[0133] Use the greedy strategy combined with the objective function to gradually allocate ingredients to ensure the partition capacity V k Not overloaded;

[0134] Calculate the target temperature and humidity T for each zone opt (Z k ) and H opt (Z k ), adjust the current environment parameter T k ,H k , so that it gradually approaches the target value.

[0135] If some partitioned devices cannot reach the target value, re-evaluate the low P i Food distribution ensures that the freshness requirements of high-priority ingredients are met.

[0136] Furthermore, a storage mapping table for each ingredient is generated based on the optimization results.

[0137] Z k : Indicates the partition number to which ingredient i is assigned;

[0138] T k ,H k : The actual adjusted temperature and humidity parameters of the partition;

[0139] P i : Storage priority for subsequent monitoring reference.

[0140] Furthermore, the partition environmental parameters {T k ,H k}, calculate and T opt (Z k ) and H opt (Z k ) deviation.

[0141] If the deviation ΔT k or ΔH k When the preset threshold is exceeded, the following mechanisms are triggered:

[0142] Call the partition optimization algorithm again to reallocate some low-priority ingredients;

[0143] Issue abnormal reminders and suggest users to make environmental corrections or adjust food storage strategies.

[0144] The above solution ensures the long-term preservation of high-priority ingredients by dynamically optimizing the partitioned storage environment, combining the freshness protection regularization term and the partitioned energy consumption constraint, and lays a precise storage foundation for subsequent steps (real-time monitoring and dietary recommendations).

[0145] S3. Regularly obtain the real-time environmental parameters of each partition through the partition sensor and calculate the deviation between the actual environment of the previous partition and the target storage conditions, and update the remaining shelf life of each food. Calculate the food freshness index based on the deviation and the remaining shelf life to evaluate the status of each food in real time.

[0146] Specifically, the following food label information is introduced from steps 1 and 2: exp ,W i ,T opt ,H opt}, respectively, remaining shelf life, food weight, optimal storage temperature and humidity requirements. These input data provide the basic information required for real-time monitoring, including current environmental conditions, food storage characteristics, and partition mapping to ensure the efficient operation of the monitoring system.

[0147] Furthermore, at each time interval Δt, the partition sensor system obtains the Z of each partition k Real-time environment parameters: T real,k (t),H real,k (t), calculate the partition environment deviation, which is used to measure the difference between the actual environment of the current partition and the target storage condition:

[0148] ΔT k =|T real,k -T k |,ΔHk =|H real,k -H k |

[0149] Where, ΔT k Represents partition Z k Temperature deviation; ΔH k Represents partition Z k humidity deviation.

[0150] Collect the real-time weight W of each ingredient i i (t), record the change in the amount of food taken each time ΔW i :

[0151] ΔW i =W i (t)-W i (t-Δt)

[0152] Update the remaining shelf life of each ingredient rem :

[0153] D rem =D exp -(T cur -T start )

[0154] Among them, T cur Indicates the current time; T start Indicates the start time of food storage.

[0155] Furthermore, this solution proposes a nonlinear freshness index model that combines environmental deviation and shelf life to evaluate the status of each ingredient in real time:

[0156]

[0157] Among them, α, β represent the weight of the deviation on the index, which is used to control the influence of temperature and humidity; ∈ represents a small value to prevent the denominator from being zero; It indicates the nonlinear effect of temperature deviation on freshness preservation. The greater the deviation, the greater the effect on B i The more significant the reduction; Used to reflect the direct impact of the remaining shelf life on the freshness status.

[0158] Furthermore, an innovative regularization term is specially designed to consider the environmental conflict effects between multiple ingredients and give priority to protecting high priority (P i High) ingredients:

[0159]

[0160] This regularization item assigns a higher environmental resource adjustment weight to ingredients with greater storage priority conflicts in the same partition to avoid negative impacts on key ingredients.

[0161] Furthermore, the partition deviation is monitored to detect whether it exceeds the preset threshold δ T and δ H :

[0162] ΔT k >δ T or ΔH k >δ H

[0163] If an abnormality is triggered, the system enters the following adjustment process:

[0164] Low P i The ingredients are preferentially transferred to the partitions with higher environmental adaptability to release the partition resources;

[0165] Adjusted Allocation Update and partition parameters {T k ,H k}.

[0166] Use dynamic control algorithms to adjust the partition equipment and optimize the temperature and humidity to the target value T k ,H k Get closer.

[0167] The screen or App will prompt users of abnormal conditions and provide suggestions (such as reducing the storage of new ingredients, adjusting storage methods, etc.).

[0168] Furthermore, if the problem cannot be solved after adjustment, the system records the current abnormal state and gives priority to it in the next optimization step.

[0169] Understandably, output the current partition storage status {T real,k ,H real,k ,ΔT k ,ΔH k} and food freshness index {B i}, update the central database;

[0170] Record all adjustments and abnormal event logs to provide real-time food status support for the subsequent meal recommendation module (step 5).

[0171] This solution uses an innovative freshness index model and regular term design to monitor the storage status of food in the refrigerator in real time, and combines dynamic adjustment and anomaly detection mechanisms to provide core support for long-term preservation and intelligent management.

[0172] S4. When the user takes food, the real-time weight of each time the food is taken is recorded and the change in the amount of food taken each time is calculated. Each taking behavior is recorded according to the change in the amount of food taken, and the taking timestamp and the name of the food are associated to form an intake log. The nutritional value of each intake is calculated according to the change in the amount of food taken and the nutritional components of the corresponding food. The nutritional value of each intake is input into a predefined user health model to analyze the user's dietary status in real time.

[0173] Specifically, the food nutrition label {η provided in step 1 is introduced i}, including the basic nutritional content of each ingredient:

[0174] η i ={η cal ,η pro ,η fat ,η carb}, representing the gram content of calories, protein, fat and carbohydrates respectively.

[0175] Initialize the health model Used to track the user's healthy intake status. The initial value is determined by historical intake data.

[0176] The initialization goal is to integrate real-time monitoring data and food label information to provide accurate basic data for user intake behavior analysis.

[0177] Furthermore, according to {ΔW i}Record each access behavior and associate it with the access timestamp T use and ingredient name i , forming the ingestion log:

[0178]

[0179] Each log record corresponds to an intake behavior and is saved in the intake log database for analyzing the user's long-term eating habits.

[0180] Furthermore, through the user's usage behavior ΔW i Update remaining weight of ingredients W i :

[0181] W i (t) = W i (t-Δt)-ΔW i

[0182] The update ensures that inventory data is synchronized in real time, providing accurate ingredient status for subsequent meal recommendations.

[0183] Furthermore, according to the current time T cur and storage start time T start Update remaining shelf life Drem :

[0184] D rem =D exp -(T cur -T start )

[0185] Simultaneously monitor D rem The critical value of D rem ≤δ exp , marking the ingredient as about to expire, triggering a priority recommendation strategy.

[0186] Further, according to the weight ΔW taken i And the nutritional content of the corresponding ingredients η i , calculate the nutritional value of each intake:

[0187]

[0188] in, Include total calories and total amount of macronutrients (protein, fat, carbohydrates).

[0189] Furthermore, the present invention introduces a freshness adjustment factor γ i , used to simulate the effect of food freshness on nutrient absorption:

[0190] γ i =exp(-λ·B i )

[0191] Among them, λ is the adjustment parameter, which controls the weight of the freshness index on the nutritional correction; γ i The smaller it is, the lower the freshness and the greater the impact on the nutritional value.

[0192] Furthermore, based on γ i Correct the nutritional value to get the actual nutritional intake:

[0193]

[0194] The revised model reflects the potential impact of refrigerator storage management on nutritional intake and enhances the relevance of the system to user health management.

[0195] Furthermore, the user health model The following core parameters are included:

[0196] Total intake: (calories), (protein), (Fat), (carbohydrate);

[0197] Food category intake ratio: {ρveg ,ρ meat ,ρ grain}, representing the intake proportion of vegetables, meat and grains respectively;

[0198] Nutritional intake deviation Reflects the difference between current intake status and target health status.

[0199] Further, according to the modified nutritional value And the current time interval Δt, update the health model:

[0200]

[0201] The updated health model is used to analyze the user's dietary status in real time and provide input for dietary recommendations.

[0202] Furthermore, by ingesting logs and health model Detects the following unusual behaviors:

[0203] The intake ratio of a single category is too high or too low:

[0204] ρ meat >θ meat ,ρ veg <θ veg

[0205] where θ meat ,θ veg is the health range threshold;

[0206] Total calorie intake deviates from target value:

[0207]

[0208] are the upper and lower bounds of the target intake range.

[0209] Furthermore, if an abnormality is detected, the system provides specific suggestions through the refrigerator screen or mobile app:

[0210] Adjust food intake to increase the intake of under-consumed categories;

[0211] Optimize the combination of ingredients to avoid excessive or monotonous intake.

[0212] Understandably, output user health intake model and ingestion logs Provide accurate input for the next dietary recommendation (step 5);

[0213] Data is stored in a central database to support long-term health analysis and system optimization.

[0214] This solution comprehensively analyzes the user's intake behavior through freshness correction, dynamic update of nutrition model and anomaly detection mechanism, and is closely integrated with the storage management function of the refrigerator to provide scientific support for user health management and personalized dietary recommendations.

[0215] S5. Generate personalized dietary recommendations and reminders based on user health needs and storage status.

[0216] Specifically, the storage status data of step 3 is introduced:

[0217] Food freshness index i} and remaining shelf life {D rem}, used to give priority to recommending ingredients that are about to expire;

[0218] Current inventory weight {W i} and storage location Used to generate meal suggestions and location directions.

[0219] Furthermore, the optimization goal of dietary recommendations is defined: while meeting the health needs of users, maximize the use of existing ingredients and reduce waste.

[0220] Furthermore, the constraints:

[0221] Ensure nutritional balance of recommended recipes and user health model The target value in near:

[0222]

[0223] Among them, ∈ cal ,∈ pro is the acceptable deviation range.

[0224] Give priority to ingredients that are about to expire:

[0225]

[0226] where ∈ is a small value that prevents the denominator from being zero.

[0227] To avoid insufficient stock of ingredients, add the following recommendations:

[0228]

[0229] W min The minimum stock required for the recommended dish.

[0230] Furthermore, the objective function Φ of dietary recommendation is designed based on the user's health needs, food freshness and inventory status:

[0231]

[0232] Among them, η cal,i is the heat of food i; ΔW i is the recommended amount of food; α and β are weight coefficients, which control the influence of freshness and shelf life respectively.

[0233] Furthermore, the Lagrange multiplier method is used to integrate constraints such as dietary balance and waste reduction into the objective function, and the recommended solution is optimized:

[0234]

[0235] in, As an indicator function, ensure that ingredients with sufficient inventory are recommended first.

[0236] Furthermore, according to the optimization result {ΔW i}, generate a meal plan with specific recipes and amounts include:

[0237] Recommended recipe name (e.g., "Chicken Breast Salad");

[0238] Required ingredients and recommended usage ΔW i ;

[0239] Reason for the recommendation (e.g., “Prefer carrots that are about to expire”).

[0240] System combined with storage location Provide food location guidance (such as "second layer on the left side of the refrigerator area") to improve user convenience.

[0241] Furthermore, the system reminds the user through the refrigerator screen or mobile app:

[0242] Ingredients that are about to expire and their priority use recommendations (e.g., "milk will expire in 2 days and can be used to make oatmeal milk porridge");

[0243] Meal recommendations and preparation instructions (e.g. “Take 200g of chicken breast from the second shelf of the refrigerator to make chicken breast salad”).

[0244] Furthermore, after finishing their meals, users can provide feedback on their actual usage and satisfaction through the App. The system dynamically adjusts and optimizes the model based on the feedback to continuously improve the accuracy of recommendations.

[0245] Further, output dietary recommendations and reminder information Provide users with personalized operation suggestions;

[0246] Store user feedback and usage data to provide support for long-term optimization of dietary recommendation models and health analysis.

[0247] This solution generates personalized meal plans by optimizing the meal recommendation objective function, combining an innovative user health model and ingredient freshness regularization term, reducing waste and meeting users' health needs, providing core support for the proactive services of smart refrigerators.

[0248] In summary, the present invention has made comprehensive improvements to the defects of the prior art from three aspects: dynamic storage optimization, personalized health management and intelligent interaction. It not only extends the shelf life of food, but also brings users a more convenient and scientific user experience through intelligent health management functions.

[0249] The embodiment of the present invention further provides a smart refrigerator article storage management system, the system comprising:

[0250] A refrigerator data collection unit is used to collect the internal data of the refrigerator and food nutrition labels, wherein the internal data includes item images, item RFID tag information, item weight, and refrigerator partition environment parameters, and recognize the item images according to a preset multi-label classification model, and generate an intelligent storage label data set based on the recognition results combined with the item RFID tag information and item weight;

[0251] Among them, the recognition result includes the name of the food, the category of the food and the recognition confidence; the RFID tag information of the item includes the shelf life and production date marked on the package, and the remaining shelf life calculated in combination with the current time, and the recommended storage temperature range and recommended storage humidity range for food storage are obtained from the domain food database; the intelligent storage tag data set includes the name of the food, the category of the food, the recommended storage temperature range, the recommended storage humidity range, the remaining shelf life, the item weight and the storage priority; the refrigerator partition environmental parameters include the current temperature and humidity;

[0252] A refrigerator partition storage unit is used to initialize refrigerator partitions and design a dynamic allocation model for refrigerator partitions. The dynamic allocation model is used to maximize the storage and preservation effect of food ingredients and minimize the overall energy consumption of the refrigerator. The intelligent storage label data set and the initialized refrigerator partitions are used as inputs of the dynamic allocation model to perform partition optimization calculations to obtain optimized partition results. According to the optimized partition results, a storage mapping table for each food ingredient is generated to obtain the partition number to which the food ingredient is assigned, the actual adjusted temperature and humidity parameters of the partition, and the storage priority.

[0253] The refrigerator partition storage optimization unit is used to regularly obtain the real-time environmental parameters of each partition through the partition sensor and calculate the deviation between the actual environment of the front partition and the target storage conditions, and update the remaining shelf life of each food. The food preservation index is calculated based on the deviation and the remaining shelf life to evaluate the status of each food in real time;

[0254] If the partition sensor detects an abnormal deviation, and the abnormal deviation exceeds a preset threshold, an abnormal adjustment process is triggered to generate a new food storage mapping table and partition environment parameters;

[0255] The refrigerator partition access change adaptation unit is used to record the real-time weight of each time the food is taken and calculate the change in the amount of food taken each time when the user takes food, record each access behavior according to the change in the amount of food taken, associate the access timestamp and the name of the food, form an intake log, calculate the nutritional value of each intake according to the change in the amount of food taken and the nutritional components of the corresponding food, and input the nutritional value of each intake into a predefined user health model to analyze the user's dietary status in real time;

[0256] The personalized setting unit is used to generate personalized dietary recommendations and reminders based on the user's health needs and storage status.

[0257] Unless otherwise specifically stated, the relative steps, numerical expressions and values ​​of the components and steps set forth in these embodiments do not limit the scope of the present application.

[0258] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a non-volatile computer-readable storage medium that is executable by a processor. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0259] In the description of the present application, it should be noted that the terms "upper", "lower", etc. indicate orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, or the orientations or positional relationships in which the inventive product is usually placed when in use. These are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as a limitation on the present application.

[0260] In the description of this application, it should also be noted that, unless otherwise clearly specified and limited, the terms "set", "install", "connect", and "connect" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two elements. For ordinary technicians in this field, the specific meanings of the above terms in this application can be understood according to specific circumstances.

[0261] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit it. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for managing storage of items in a smart refrigerator, characterized in that: The method comprises: S1. Collect the internal data of the refrigerator and the food nutrition label, wherein the internal data includes the item image, the item RFID tag information, the item weight and the refrigerator partition environment parameters, and identify the item image according to the preset multi-label classification model, and generate the intelligent storage label data set according to the identification result combined with the item RFID tag information and the item weight; Among them, the recognition result includes the name of the food, the category of the food and the recognition confidence; the RFID tag information of the item includes the shelf life and production date marked on the package, and the remaining shelf life calculated in combination with the current time, and the recommended storage temperature range and recommended storage humidity range for food storage are obtained from the domain food database; the intelligent storage tag data set includes the name of the food, the category of the food, the recommended storage temperature range, the recommended storage humidity range, the remaining shelf life, the item weight and the storage priority; the refrigerator partition environmental parameters include the current temperature and humidity; S2. Initialize refrigerator partitions and design a dynamic allocation model for refrigerator partitions. The dynamic allocation model is used to maximize the storage and preservation effect of food and minimize the overall energy consumption of the refrigerator. The intelligent storage label data set and the initialized refrigerator partitions are used as inputs of the dynamic allocation model to perform partition optimization calculations to obtain optimized partition results. According to the optimized partition results, a storage mapping table for each food is generated to obtain the partition number to which the food is assigned, the actual adjusted temperature and humidity parameters of the partition, and the storage priority; S3, regularly obtain the real-time environmental parameters of each partition through the partition sensor and calculate the deviation between the actual environment of the front partition and the target storage condition, and update the remaining shelf life of each food. Calculate the food freshness index based on the deviation and the remaining shelf life to evaluate the status of each food in real time; If the partition sensor detects an abnormal deviation, and the abnormal deviation exceeds a preset threshold, an abnormal adjustment process is triggered to generate a new food storage mapping table and partition environment parameters; S4. When the user takes food, the real-time weight of each time the food is taken is recorded and the change in the amount of food taken each time is calculated. Each taking behavior is recorded according to the change in the amount of food taken, and the taking timestamp and the name of the food are associated to form an intake log. The nutritional value of each intake is calculated according to the change in the amount of food taken and the nutritional components of the corresponding food, and the nutritional value of each intake is input into a predefined user health model to analyze the user's dietary status in real time; S5. Generate personalized dietary recommendations and reminders based on user health needs and storage status.

2. The method for storing and managing items in a smart refrigerator according to claim 1, characterized in that: The loss function of the preset multi-label classification model It is expressed as: in, is the cross entropy loss; S j For the confidence of each category, low confidence categories contribute more loss; λ is the regularization weight, which controls the regularization effect; ∈ prevents the denominator from being zero; n is the total number of categories; The storage priority P i Calculated based on the remaining shelf life and category weight, expressed as: in, is the class weight, α controls the impact of the class weight, D exp For the remaining shelf life.

3. The method for storing and managing items in a smart refrigerator according to claim 1, characterized in that: The objective function of the dynamic allocation model is as follows: Among them, Φ is the objective function, the overall optimization value; is an indicator function, indicating whether the i-th ingredient is assigned to partition Z k ; ΔT k =|T k -T opt (Z k )|, ΔT k is the current partition temperature T k and demand T opt (Z k ) deviation; ΔH k =|H k -H opt (Z k )|, ΔH k is the current zone humidity H k With demand H opt (Z k ) deviation; β,γ respectively represent the influence weights of temperature and humidity deviation on the optimization target; Each partition Z k The target temperature and humidity are calculated using the weighted average of ingredients: Among them, T opt ,H opt Recommended storage temperature and humidity, which come from the smart storage label dataset for each ingredient i is the current ingredient i; W i is the weight of the food; At the same time, the freshness protection regularization term is introduced into the dynamic allocation model Prioritize the allocation of ingredients with the shortest shelf life to the best environment: The regularization term increases the remaining shelf life D exp The inverse of the square of , so that the remaining ingredients with short shelf life are allocated to the most suitable partition first to avoid waste.

4. The method for managing storage of items in a smart refrigerator according to claim 3, characterized in that: The partition optimization calculation comprises: Store all ingredients according to the priority P i Arrange in descending order, giving priority to ingredients with higher priority; Use the greedy strategy combined with the objective function to gradually allocate ingredients to ensure the partition capacity V k Not overloaded; Calculate the target temperature and humidity T for each zone opt (Z k ) and H opt (Z k ), adjust the current refrigerator partition environmental parameters T k ,H k , so that it gradually approaches the target value; If the devices in some partitions cannot reach the target value, re-evaluate the low storage priority P i Food distribution ensures that the freshness requirements of high-priority ingredients are met.

5. The method for storing and managing items in a smart refrigerator according to claim 1, characterized in that: The method of calculating the food freshness index according to the deviation and the remaining shelf life to evaluate the status of each food in real time includes: A nonlinear freshness index model combining environmental deviation and shelf life is used to evaluate the status of each ingredient in real time: Among them, α, β represent the weight of the deviation on the index, which is used to control the influence of temperature and humidity; ∈ represents a small value to prevent the denominator from being zero; It indicates the nonlinear effect of temperature deviation on freshness preservation. The greater the deviation, the greater the effect on B i The more significant the reduction; Used to reflect the direct impact of the remaining shelf life on the freshness status. Introducing regularization terms Consider the environmental conflict effects between multiple ingredients and give priority to protecting high-priority ingredients: The regular term Assign higher environmental resource adjustment weights to ingredients with greater storage priority conflicts in the same partition to avoid negative impacts on key ingredients.

6. The method for storing and managing items in a smart refrigerator according to claim 1, characterized in that: The food nutrition label includes the basic nutritional ingredients of each food, including the gram content of calories, protein, fat and carbohydrates; After the timestamp and the name of the food are combined to form the intake log, the change in the amount of food consumed ΔW is calculated. i Update remaining weight of ingredients W i : W i (t)=W i (t-Δt)-ΔW i At the same time, according to the current time T cur and storage start time T start Update remaining shelf life D rem : D rem =D exp -(T cur -T start ) Simultaneously monitor D rem The critical value of D rem ≤δ exp , marking the ingredient as about to expire, triggering a priority recommendation strategy.

7. A method for managing storage of items in a smart refrigerator according to claim 6, characterized in that: The nutritional value of each intake is calculated as follows: The change in the amount of food consumed ΔW i And the nutritional content of the corresponding ingredients η i , calculate the nutritional value of each intake: in, Include total calories and total macronutrients; Introducing freshness adjustment factor γ i , used to simulate the effect of food freshness on nutrient absorption: c i =exp(-λ·B i ) Among them, λ is the adjustment parameter, which controls the weight of the freshness index on the nutritional correction; γ i The smaller it is, the lower the freshness is and the greater the impact on the nutritional value is; Based on gamma i Correct the nutritional value to get the actual nutritional intake The user health model The update rules are as follows: Where f is the update function and Δt is the change time.

8. The method for managing storage of items in a smart refrigerator according to claim 1, characterized in that: The S5 specifically includes: Define optimization goals for dietary recommendations; Combining user health needs, food freshness and inventory status, the objective function Φ of dietary recommendation is designed: Among them, η cal,i is the heat of food i; ΔW i is the recommended amount of ingredients; α, β are weight coefficients, which control the influence of freshness and shelf life respectively, B i It is the food freshness index; The Lagrange multiplier method is used to integrate constraints such as dietary balance and waste reduction into the objective function, and the recommended solution is obtained through optimization: in As an indicator function, ensure that ingredients with sufficient inventory are recommended first; Generate a meal plan with specific recipes and usage amounts based on the optimization results It also generates smart reminders to remind users through the refrigerator screen or mobile app, providing users with personalized operation suggestions.

9. The method for managing storage of items in a smart refrigerator according to claim 8, characterized in that: The constraints of the optimization objective are: Ensure nutritional balance of recommended recipes and user health model The target value in Closest to: Among them, ∈ cal ,∈ pro is the acceptable deviation range; Give priority to ingredients that are about to expire: Among them, ∈ is a small value to prevent the denominator from being zero; To avoid insufficient stock of ingredients, add the following recommendations: Among them, W min The minimum stock required for the recommended dish.

10. An intelligent refrigerator item storage management system, characterized in that: The system comprises: A refrigerator data collection unit is used to collect the internal data of the refrigerator and food nutrition labels, wherein the internal data includes item images, item RFID tag information, item weight, and refrigerator partition environment parameters, and recognize the item images according to a preset multi-label classification model, and generate an intelligent storage label data set based on the recognition results combined with the item RFID tag information and item weight; Among them, the recognition result includes the name of the food, the category of the food and the recognition confidence; the RFID tag information of the item includes the shelf life and production date marked on the package, and the remaining shelf life calculated in combination with the current time, and the recommended storage temperature range and recommended storage humidity range for food storage are obtained from the domain food database; the intelligent storage tag data set includes the name of the food, the category of the food, the recommended storage temperature range, the recommended storage humidity range, the remaining shelf life, the item weight and the storage priority; the refrigerator partition environmental parameters include the current temperature and humidity; A refrigerator partition storage unit is used to initialize refrigerator partitions and design a dynamic allocation model for refrigerator partitions. The dynamic allocation model is used to maximize the storage and preservation effect of food ingredients and minimize the overall energy consumption of the refrigerator. The intelligent storage label data set and the initialized refrigerator partitions are used as inputs of the dynamic allocation model to perform partition optimization calculations to obtain optimized partition results. According to the optimized partition results, a storage mapping table for each food ingredient is generated to obtain the partition number to which the food ingredient is assigned, the actual adjusted temperature and humidity parameters of the partition, and the storage priority. The refrigerator partition storage optimization unit is used to regularly obtain the real-time environmental parameters of each partition through the partition sensor and calculate the deviation between the actual environment of the front partition and the target storage conditions, and update the remaining shelf life of each food. The food preservation index is calculated based on the deviation and the remaining shelf life to evaluate the status of each food in real time; If the partition sensor detects an abnormal deviation, and the abnormal deviation exceeds a preset threshold, an abnormal adjustment process is triggered to generate a new food storage mapping table and partition environment parameters; The refrigerator partition access change adaptation unit is used to record the real-time weight of each time the food is taken and calculate the change in the amount of food taken each time when the user takes food, record each access behavior according to the change in the amount of food taken, associate the access timestamp and the name of the food, form an intake log, calculate the nutritional value of each intake according to the change in the amount of food taken and the nutritional components of the corresponding food, and input the nutritional value of each intake into a predefined user health model to analyze the user's dietary status in real time; The personalized setting unit is used to generate personalized dietary recommendations and reminders based on the user's health needs and storage status.

Images