Diet dinner plate for diabetics

Abstract

The diet dinner plate comprises a dinner plate body, an opening is formed in the upper end of the dinner plate body, a protective cover used for sealing the opening is rotationally installed at the opening of the dinner plate body through a second rotating shaft, a rotating plate is rotationally installed in the protective cover through a first rotating shaft, and a heat detection module is arranged in the rotating plate; a partition plate is rotatably mounted on the rotating plate through a third rotating shaft, partition blocks are fixedly connected to the inner wall of the dinner plate body and used for dividing an inner cavity of the dinner plate body into a plurality of containing cavities, and a locking structure is arranged on the dinner plate body and used for locking or unlocking the protective cover and the dinner plate body. The dinner plate can be folded through rotation of the protective cover and the dinner plate body in combination with limiting and fixing of the locking structure, heat detection is conducted on food in all areas in the dinner plate body through the heat detection module, and quantitative diet data support is provided for a patient.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of diabetic diet care, in particular to a diet plate for diabetic patients. BACKGROUND

[0002] With the continuous rise of global diabetes prevalence, diabetes has become one of the major chronic diseases affecting human health. As the core of comprehensive treatment of diabetes, diet management is directly related to the blood glucose control effect and the risk of complications. Clinical studies have shown that the diet management of diabetic patients needs to precisely control the intake of calories and the proportion of carbohydrates, proteins and fats, and avoid the dramatic fluctuations in blood glucose caused by unbalanced diet. Therefore, special diet utensils with quantitative detection and diet guidance functions have become a key requirement for clinical nursing and daily health management.

[0003] At present, the plates used by diabetic patients only have basic holding and partition functions, and the diet proportion is guided by partition. Since there are differences in individual calorie needs and nutrient tolerance among diabetic patients, and daily meals are mostly mixed and stacked foods, patients cannot know the actual calories and specific contents of carbohydrates, proteins and fats in meals through existing plates, and can only rely on experience estimation or manual query for diet control, resulting in excessive or insufficient intake of each meal, which cannot meet the needs of fine diet management of diabetic patients. SUMMARY

[0004] In view of the deficiencies of the prior art, the present application provides a diet plate for diabetic patients, which solves the problem that the existing plates used by diabetic patients can only rely on experience estimation or manual query for diet control, resulting in excessive or insufficient intake of each meal, which cannot meet the needs of fine diet management of diabetic patients.

[0005] To achieve the above purpose, the present application is realized by the following technical scheme: a diet plate for diabetic patients, comprising a plate body, characterized in that the upper end of the plate body is open, a protective cover for closing the opening is rotatably installed on the opening of the plate body through a second rotating shaft, a rotating plate is rotatably installed in the protective cover through a first rotating shaft, a heat detection module is arranged in the interior of the rotating plate, and the heat detection module is used for heat detection of food in the plate body. A partition plate is rotatably installed on the rotating plate through a third rotating shaft, a partition block is fixedly connected to the inner wall of the plate body, the partition block is used for dividing the inner cavity of the plate body into a plurality of accommodating cavities, a locking structure is arranged on the plate body, and the locking structure is used for locking or unlocking the protective cover and the plate body.

[0006] Further, the first rotating shaft is a first damping rotating shaft, the first damping rotating shaft is two and symmetrically installed on the protection cover, the second rotating shaft is a second damping rotating shaft, the number of the second damping rotating shaft is two and symmetrically installed on the dining plate body, the third rotating shaft is a third damping rotating shaft, the third damping rotating shaft is two and symmetrically installed on the rotating plate, and the partition block is T-shaped.

[0007] Further, one side of the dining plate body and one side of the protection cover are provided with a placing groove, the interior of the dining plate body and the interior of the protection cover are symmetrically provided with a sliding groove, one side of the protection cover is provided with a mounting groove, the other side is symmetrically provided with a tableware groove, one side of the rotating plate is provided with a through hole, the locking structure comprises a handle and a pressing column, the outer wall of the pressing column is slidably connected to the interior of the protection cover, the handle is rotatably connected with a rotating shaft, the two ends of the rotating shaft are respectively provided with sliding blocks, and the sliding blocks are slidably arranged in the sliding grooves of the dining plate body and the protection cover.

[0008] Further, one side of the pressing column is symmetrically fixedly connected with a spring, one end of the spring is fixedly connected to the interior of the protection cover, the bottom of the pressing column is provided with an L-shaped limiting block, and one end of the L-shaped limiting block is provided with a first wedge surface.

[0009] Further, the middle part of the rotating shaft is fixedly connected with a limiting block, the limiting block is provided with a limiting groove on one side, the L-shaped limiting block has a hook portion, the first wedge surface is arranged on the hook portion, the hook portion can be hooked with the limiting groove, the top of the limiting block is provided with a second wedge surface, and the first wedge surface of the limiting block is matched with the second wedge surface of the L-shaped limiting block.

[0010] Further, the heat detection module comprises a data acquisition unit, a modeling solving unit and a heat calculation unit, the data acquisition unit is used for acquiring the spectrum signal, the mass data and the environmental temperature data of the food in the dining plate body, and the acquisition data is obtained after pretreatment; The modeling solving unit is used for establishing a coupling model based on the acquisition data, and solving the hidden variable of the food through a variational inference optimization algorithm; The heat calculation unit is used for calculating the mass of carbohydrates, proteins and fats in the food according to the hidden variable, and then obtaining the total heat, and displaying the total heat.

[0011] Further, the spectrum signal comprises three characteristic band signals of near-infrared characteristic absorption peaks of carbohydrates, proteins and fats in the food, and the pretreatment comprises filtering processing, temperature drift correction processing, absorbance conversion processing and data format standardization processing.

[0012] Further, the establishment of the coupling model specifically comprises the following steps: Based on the collected data, a multi-component and multi-level absorbance assumption model is established by using the Lambert-Beer law, and the absorbance assumption model includes the quantitative relationship between the absorbance of each characteristic wave band and the concentration of food components, the thickness proportion of each layer and the total thickness of the layers; A scattering correction term based on the Rayleigh-Gans approximation is introduced into the absorbance assumption model to compensate for the scattering distortion of the spectral signal, and a compensated scattering correction term is obtained; Based on the collected data, a mass constraint model is established by using the law of conservation of mass, and the mass constraint model is used to associate the total mass of the food with the concentration of the components, the total thickness and the density of the nutrients; The absorbance assumption model, the compensated scattering correction term and the mass constraint model are combined to form a coupled model.

[0013] Further, the hidden variables of the food are solved by the variational inference optimization algorithm, and the method specifically includes the following steps: Based on the coupled model, a conjugate prior distribution is set for each hidden variable, and the hidden variables include the thickness proportion of the carbohydrate layer, the thickness proportion of the protein layer, the thickness proportion of the fat layer, the average concentration of the carbohydrate, the average concentration of the protein, the average concentration of the fat and the total thickness; A variational distribution conjugate to the prior distribution is constructed to obtain variational distribution parameters, and the variational distribution is decomposed into the product of each sub-distribution of the hidden variables; The variational distribution parameters are optimized by maximizing the lower bound of the evidence, the expectation of the likelihood term is calculated by using Monte Carlo sampling, the KL divergence term is calculated in an analytical manner, and the optimization is iterated until the lower bound of the evidence converges, and the expectation of the variational distribution is taken as the optimal estimation value of the hidden variable.

[0014] Further, the total heat is obtained, and the method specifically includes the following steps: Based on the optimal estimation value of the hidden variable, the dietary partition area of the dining table ontology and the standard density of the nutrients, the mass of the carbohydrates, the proteins and the fats in the food is calculated, and the nutrient-related data includes the mass and proportion of the carbohydrates, the proteins and the fats in the food; According to the mass, the standard unit heat coefficients of the carbohydrates, the proteins and the fats are used to obtain the total heat of the food by linear superposition; The total heat and the nutrient-related data are displayed through video display and voice broadcast.

[0015] The beneficial effects of the present application are: 1. The present application divides the inner part of the dinner plate body into independent areas conforming to the golden ratio of diabetes diet through the partition block, provides a targeted detection basis for the heat detection module, and enables the heat detection module to face the food area when in use and to be stored and protected when idle by opening and closing the rotating plate. The heat detection module in the rotating plate is physically protected by the partition plate, and the dinner plate can be folded to improve portability while protecting the internal structure by rotating the protective cover and the dinner plate body and limiting and fixing the locking structure. The heat detection module detects the heat of the food in each area of the dinner plate body, thereby realizing the detection and output of food heat and providing quantitative diet data support for patients.

[0016] 2. The present application drives the rotating shaft through the handle to make the sliding block slide in the sliding groove, thereby driving the limiting block to move up and down. When the limiting block moves upward and touches the L-shaped limiting block, the matching wedge surfaces of the limiting block and the L-shaped limiting block make one end of the L-shaped limiting block enter the limiting slot of the limiting block, thereby limiting the protective cover on the dinner plate body. When the protective cover needs to be opened, the L-shaped limiting block can be removed from the limiting slot by pressing the pressing column, and the protective cover can be opened, thereby facilitating the protective cover to protect the dinner plate body and the structure inside the dinner plate body, and enhancing the practicality and convenience of use.

[0017] 3. The present application establishes a multi-component and multi-level absorbance assumption model based on Lambert-Beer's law, introduces a Rayleigh-Gans approximate scattering correction term to compensate for signal distortion, and constructs a coupling model combined with the law of conservation of mass. Then, the implicit variable is efficiently solved by a variational inference optimization algorithm, which not only adapts to limited computing power but also ensures the estimation accuracy of parameters such as layer thickness ratio and component concentration. Based on the optimal estimation value of the implicit variable, the mass of each nutrient and the total heat are accurately deduced, and the result reliability is guaranteed by mass consistency verification. The result is output in the form of display and voice broadcast, thereby realizing the precision, real-time and localization of heat detection. BRIEF DESCRIPTION OF DRAWINGS

[0018] Figure 1 The present application is a three-dimensional structure schematic diagram.

[0019] Figure 2 The present application is a dinner plate body internal structure schematic diagram.

[0020] Figure 3 The present application is a protective cover local structure schematic diagram.

[0021] Figure 4 The present application is Figure 2 The present application is an enlarged schematic diagram of position A.

[0022] Figure 5 The present application is a rotating plate local structure schematic diagram.

[0023] Figure 6 The partial structure of the limiting block of the present application is shown in the exploded schematic view.

[0024] Wherein, 1, the dinner plate body; 2, the protective cover; 3, the handle; 4, the placing groove; 5, the rotating shaft; 6, the pressing column; 7, the limiting block; 8, the partition block; 9, the accommodation groove; 10, the tableware groove; 11, the partition plate; 12, the rotating plate; 13, the first damping rotating shaft; 14, the sliding block; 15, the sliding groove; 16, the limiting groove; 17, the spring; 18, the second damping rotating shaft; 19, the heat detection module; 20, the third damping rotating shaft; 21, the L-shaped limiting block; 22, the through hole. DETAILED DESCRIPTION

[0025] The technical solutions of the present application will be described clearly and completely below in combination with the drawings of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all the other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0026] Please refer to the drawings of the present application Figure 1 - the drawings of the present application Figure 6 The embodiment of the present application provides a diet dinner plate for diabetic patients, which comprises a dinner plate body 1, a second damping rotating shaft 18 is symmetrically installed on one side of the dinner plate body 1, a protective cover 2 is installed on one side of the second damping rotating shaft 18, a first damping rotating shaft 13 is symmetrically installed on one side of the protective cover 2, a rotating plate 12 is installed on one side of the first damping rotating shaft 13, a heat detection module 19 is arranged in the rotating plate 12, the heat detection module 19 is used for detecting the heat of food in the dinner plate body 1, a third damping rotating shaft 20 is symmetrically installed on one side of the rotating plate 12, a partition plate 11 is installed on one side of the third damping rotating shaft 20, a partition block 8 is fixedly connected to the inner wall of the dinner plate body 1, the partition block 8 is T-shaped, and a locking structure is installed on the other side of the dinner plate body 1.

[0027] Specifically, the food is placed in the tray body 1, and the protective cover 2 can be flexibly rotated to a closed or open state according to the use requirement through the rotation characteristics of the second damping shaft 18. When closed, the tray body 1 and the related structure can be protected, avoiding damage to the components caused by dust, water stains and the like when carrying or idling. When opened, the placing area of the tray body 1 is not completely blocked, and the patient can take and place food without affecting the patient. Through the setting of the first damping shaft 13, the rotating plate 12 can be freely opened and closed relative to the protective cover 2. When opened, the heat detection module 19 can be directly opposite the food in the tray body 1, and when closed, the heat detection module 19 can be stored between the protective cover 2 and the rotating plate 12, reducing the interference of the external environment on the detection module. Through the heat detection module 19 directly acting on the food in the tray body 1, the collection and calculation of the heat and related nutrient composition data of the food are realized, and real-time and quantitative dietary data are provided for the patient to provide reference for fine diet management.

[0028] Through the setting of the third damping shaft 20, the partition plate 11 can be rotated on the rotating plate 12 according to the requirement, so that the partition plate 11 can be conveniently opened and closed. When opened, the opening and closing operation of the rotating plate 12 and the work of the heat detection module 19 are not affected, and when closed, a physical separation can be formed between the rotating plate 12 and the protective cover 2, avoiding the contact and wear of the heat detection module 19 by the sundries possibly existing in the protective cover 2. Through the structural design of the partition block 8, the internal space of the tray body 1 can be clearly divided into three independent areas of vegetables, proteins and staple foods according to the golden ratio of diabetic diet, so that the patient can conveniently place different types of food in proportion, and the heat detection module 19 can detect the food in each area in a targeted manner, avoiding detection errors caused by the mixing of different types of food, and improving the detection accuracy.

[0029] By setting the locking structure, the protective cover 2 can be limited on the tray body 1 according to the needs, so that the tray forms a closed structure, and at this time, the rotating plate 12 and the partition plate 11 are also accommodated in the inside of the protective cover 2, so that the tray is folded for taking, which is convenient for carrying, so that the tray body 1 and the protective cover 2, the protective cover 2 and the rotating plate 12, and the rotating plate 12 and the partition plate 11 are folded in turn through the second damping shaft 18, the first damping shaft 13 and the third damping shaft 20, and the protective cover 2 is limited on the tray body 1 through the locking structure, so that the components are unfolded for use, and the whole is folded for carrying, which improves the portability of the tray. Through the cooperation of various structures, the food is classified and stored, the heat detection module 19 is protected, and the tray is portable, and the heat detection module 19 detects the heat of the food, provides quantitative diet data support for diabetic patients, solves the problem that the existing tray for diabetic patients can only rely on experience estimation or manual query for diet control, leading to the situation that the intake of patients for each meal is excessive or insufficient, and cannot meet the needs of fine diet management of diabetic patients.

[0030] Please refer to the accompanying drawings Figure 2 , the accompanying drawings Figure 3 , the accompanying drawings Figure 5 , and the side of the tray body 1 and the side of the protective cover 2 are provided with a placing groove 4, the inside of the tray body 1 and the inside of the protective cover 2 are symmetrically provided with a sliding groove 15, one side of the protective cover 2 is provided with a mounting groove 9, the other side of the protective cover 2 is symmetrically provided with a tableware groove 10, and one side of the rotating plate 12 is provided with a through hole 22.

[0031] Specifically, the locking structure is placed through the placing groove 4, the portability is improved, the locking structure is limited through the sliding groove 15, the napkin, blood glucose test paper and the like are placed in the mounting groove 9 according to needs, and the mounting groove 9 is limited in the mounting groove 9 of the protective cover 2 by the partition plate 11, the tableware is placed through the tableware groove 10, and the rotating plate 12 is rotated on the protective cover 2 through the through hole 22, so that the utility is enhanced.

[0032] Please refer to the accompanying drawings Figure 2 - the accompanying drawings Figure 4 , the accompanying drawings Figure 6The locking structure comprises a handle 3 and a pressing column 6, the outer wall of the pressing column 6 is slidingly connected to the inside of the protective cover 2, one side of the handle 3 is rotationally connected with a rotating shaft 5, both ends of the rotating shaft 5 are fixedly connected with sliding blocks 14, the outer wall of the sliding blocks 14 is slidingly connected in sliding grooves 15 of the dining plate body 1 and the protective cover 2. One side of the pressing column 6 is symmetrically fixedly connected with springs 17, one end of the spring 17 is fixedly connected to the inside of the protective cover 2, the bottom of the pressing column 6 is fixedly connected with an L-shaped limiting block 21, one end of the L-shaped limiting block 21 is provided with a first wedge surface. The middle part of the rotating shaft 5 is fixedly connected with a limiting block 7, a limiting groove 16 is formed in one side of the limiting block 7, the top of the limiting block 7 is provided with a second wedge surface, the first wedge surface of the limiting block 7 is matched with the second wedge surface of the L-shaped limiting block 21. The L-shaped limiting block 21 has a hook portion, the first wedge surface is arranged on the hook portion, and the hook portion can be hooked with the limiting groove 16.

[0033] Specifically, by rotating the handle 3 and the rotating shaft 5, the rotating shaft 5 is limited on the dining plate body 1 and the protective cover 2 through the sliding blocks 14, so that the dining plate can be carried by the handle 3, and the handle 3 can be placed in the placing groove 4, which is convenient to put into a bag. The sliding blocks 14 slide in the sliding grooves 15 by driving the rotating shaft 5 by the handle 3, and then the limiting block 7 moves up and down. When the limiting block 7 moves up and touches the L-shaped limiting block 21, the L-shaped limiting block 21 slides in the protective cover 2 by the matching wedge surfaces of the limiting block 7 and the L-shaped limiting block 21, and then the pressing column 6 slides and the spring 17 is compressed, so that the hook portion of the L-shaped limiting block 21 is clamped into the limiting groove 16 of the limiting block 7, thereby the protective cover 2 is fixed and limited on the dining plate body 1. When the protective cover 2 needs to be opened, the hook portion of the L-shaped limiting block 21 is separated from the limiting groove 16 by pressing the pressing column 6, so that the protective cover 2 can be opened, thereby the protective cover 2 protects the dining plate body 1 and the structure in the dining plate body 1, and the practicality and convenience of use are enhanced.

[0034] The heat detection module 19 comprises a data acquisition unit, a modeling solving unit and a heat calculation unit. The data acquisition unit is used for acquiring the spectral signal, mass data and environmental temperature data of the food in the dining plate body 1, and obtaining the acquisition data after preprocessing.

[0035] Further, the spectral signal comprises three characteristic band signals of near-infrared characteristic absorption peaks of carbohydrates, proteins and fats in the food, and the preprocessing comprises filtering processing, temperature drift correction processing, absorbance conversion processing and data format standardization processing.

[0036] Specifically, the data acquisition unit synchronously acquires the spectral signal, mass data and environmental temperature data of the food in the dining plate body, and converts the original data into acquisition data meeting the subsequent modeling solving requirements through a series of preprocessing operations, thereby providing data support for the composition quantification and heat calculation of the stacked mixed food.

[0037] Generally, spectral signal acquisition is based on the near-infrared spectroscopy detection principle. It utilizes the physical property that carbohydrates, proteins, and fats exhibit characteristic absorption peaks in specific near-infrared bands. By detecting the degree of absorption of near-infrared light by food in different bands, its composition information can be inferred. As an option, the three characteristic bands can be selected as 1.73 μm, 2.18 μm, and 2.31 μm, respectively. The 1.73 μm band corresponds to the OH bond stretching vibration absorption peak of carbohydrates, the 2.18 μm band corresponds to the NH bond stretching vibration absorption peak of proteins, and the 2.31 μm band corresponds to the CH bond stretching vibration absorption peak of fats, which helps improve detection accuracy.

[0038] Quality data is collected for the food zones on the plate itself. Generally, the weight sensor is a MEMS piezoresistive sensor, and the ambient temperature data is collected through a temperature sensor. In order to eliminate noise interference in the raw data, correct deviations caused by environmental factors, and unify different types of raw data into a standardized format, the data acquisition unit needs to preprocess the collected spectral signals, quality data and ambient temperature data.

[0039] Filtering is used to remove random noise from the original signal. Moving average filtering algorithms are applied to both the spectral signal and the mass data, and their calculation formulas are as follows: ,in, The filtered first The value of each data point. The first one before filtering The value of each data point. The length of the sliding window is determined based on the sampling frequency and noise characteristics. In this invention, the sliding window length... Option 4 can be selected, which means using the average of 5 adjacent sampling points as the filtered result, thus effectively suppressing noise in the original data.

[0040] Temperature drift correction is used to correct the impact of ambient temperature changes on spectral detection parameters. Specifically, it corrects the molar absorptivity of each characteristic band, and its calculation formula is as follows: ,in, The ambient temperature is Time Nutrients in Molar absorptivity of each characteristic band At standard temperature of 25℃, the first Nutrients in Molar absorptivity of each characteristic band For temperature coefficient, This refers to the real-time ambient temperature collected by an environmental temperature sensor. As an alternative, the temperature coefficient of carbohydrates... Values The temperature coefficient of proteins Values Temperature coefficient of fat Values This formula can uniformly correct the molar absorptivity at different ambient temperatures to the standard state, avoiding detection deviations caused by temperature fluctuations and improving the stability of spectral detection.

[0041] Absorbance conversion processing is used to convert the light intensity data of the spectral signal into absorbance data that reflects the absorption characteristics of food components. The calculation formula is as follows: ,in, For the first Absorbance of each characteristic wavelength band, For the first The incident light intensity of each characteristic wavelength band For the first The intensity of transmitted light in each characteristic wavelength band. Specifically, The intensity of near-infrared light emitted by the light source was calibrated when no food was placed in it. The absorbance is the real-time detection value of the detector after light penetrates the food. The absorbance calculated by this formula can directly reflect the degree of absorption of near-infrared light in a specific wavelength band by the food, and thus establish a quantitative relationship with component concentration and layer thickness ratio.

[0042] Data format standardization converts filtered mass data and temperature-corrected absorbance data into a predefined format for subsequent calculations. Standardization converts all data to single-precision floating-point format, which reduces storage and computational overhead while maintaining data accuracy. Data format standardization also includes range normalization, mapping the numerical ranges of different data types to the same interval, preventing interference with modeling and solving due to differences in data magnitude.

[0043] In some embodiments, the data acquisition unit further includes a power supply subunit for providing power to each unit, and the data acquisition unit can control the operation of each sensor in a time-sharing wake-up manner, waking up the ambient temperature sensor, weight sensor and spectral sensor in sequence, with the wake-up interval between adjacent sensors set to a preset duration. This method can avoid the problem of excessive power consumption peak caused by multiple sensors working at the same time, and balance the synchronization of data acquisition with the requirement of low power consumption.

[0044] The modeling and solving unit is used to build a coupled model based on the collected data and solve the latent variables of food through variational inference optimization algorithm; Further, a coupling model is established, specifically including the following steps: based on the collected data, a multi-component and multi-level absorbance assumption model is established by using the Lambert-Beer law, the absorbance assumption model contains the quantitative relationship between the absorbance of each characteristic wave band and the food component concentration, layer thickness proportion and total layer thickness; a scattering correction term based on the Rayleigh-Gans approximation is introduced into the absorbance assumption model to compensate for the scattering distortion of the spectral signal, and a compensated scattering correction term is obtained; based on the collected data, a mass constraint model is established by using the law of conservation of mass, the mass constraint model is used to associate the total mass of the food with the relationship between the component concentration, total layer thickness and nutrient density; the absorbance assumption model, the compensated scattering correction term and the mass constraint model are combined to form the coupling model.

[0045] Further, the hidden variables of the food are solved by a variational inference optimization algorithm, specifically including the following steps: Based on the coupling model, a conjugate prior distribution is set for each hidden variable, and the hidden variables include the carbohydrate layer thickness proportion, the protein layer thickness proportion, the fat layer thickness proportion, the average concentration of carbohydrates, the average concentration of proteins, the average concentration of fats and the total layer thickness. A variational distribution conjugate to the prior distribution is constructed to obtain variational distribution parameters, and the variational distribution is decomposed into the product of each sub-distribution of the hidden variable. The variational distribution parameters are optimized by maximizing the lower bound of the evidence, the likelihood term expectation is calculated by Monte Carlo sampling, and the KL divergence term is calculated in an analytical way. Iterative optimization is performed until the lower bound of the evidence converges, and the expectation of the variational distribution is taken as the optimal estimation value of the hidden variable.

[0046] Specifically, the modeling and solving unit realizes accurate solving of the hidden variables of the stacked mixed food based on the collected data output by the data acquisition unit, by establishing a multi-physical field coupling model and combining a variational inference optimization algorithm, thereby providing parameter support for subsequent calculation of nutrient mass and total heat.

[0047] Generally, the component distribution and layer thickness structure of the stacked mixed food have complexity, and a single model cannot accurately describe the quantitative relationship between the spectral signal, the quality data and the component parameters, so a coupling model needs to be constructed to integrate multiple physical mechanisms. The establishment process of the coupling model is based on the collected data, and the absorbance assumption model is constructed, the scattering correction term is introduced, and the mass constraint model is established, and finally a complete coupling system is formed through model fusion to ensure the logical consistency and quantitative correlation between the physical quantities.

[0048] The absorbance assumption model is constructed based on the Lambert-Beer law, which is the basic principle of near-infrared spectrum detection and can describe the quantitative relationship between the absorption degree of a substance to a specific wavelength of light and the component concentration and optical path. Considering that the food in the present application is a stacked mixed structure of carbohydrates, proteins and fats, the Lambert-Beer law needs to be extended to form a multi-component and multi-level absorbance assumption model, and its calculation formula is: ,in, For the first data collection Absorbance of each characteristic wavelength band, The first after temperature drift correction Nutrients in Molar absorptivity of each characteristic band Represents carbohydrates, Represents protein, Represents fat. For the first The average concentration of various nutrients The total thickness of the food layer, For the first The layer thickness ratio of each nutrient, and meets the following requirements. This formula achieves a quantitative description of the total absorbance of stacked food by linearly superimposing the absorption contributions of each nutrient. The input is the absorbance in the collected data and the temperature-corrected molar absorptivity, and the output is the quantitative relationship between absorbance and latent variables, establishing a direct correlation between spectral signals and component parameters and layer thickness parameters.

[0049] Due to the particle characteristics and interlayer interface effects of stacked food, near-infrared light undergoes scattering during penetration, leading to biases in simple absorption stabilization models. Therefore, a scattering correction term based on the Rayleigh-GANS approximation is needed for compensation. The Rayleigh-GANS approximation is applicable to scenarios where the scattering particle size is much smaller than the incident light wavelength, consistent with the physical characteristics of food particles. The formula for calculating the scattering correction term is: ,in, For the first Scattering correction terms for each characteristic band, The wavelength-dependent scattering coefficient is obtained through experimental calibration. The average density of food, To collect the total quality of the corresponding dietary zones in the data, Let be the area of ​​the food zone. This scattering correction term is superimposed on the light absorption stabilization model to obtain the compensated light absorption stabilization model, whose expression is: This correction process can counteract the spectral signal distortion caused by scattering, ensuring that the quantitative relationship between absorbance and composition parameters and layer thickness parameters is more consistent with the actual physical process.

[0050] The law of conservation of mass is the fundamental basis for quantitative analysis of substances. To ensure consistency between component parameters and mass data, a mass constraint model needs to be established based on the mass data collected. The total mass of food equals the sum of the masses of each nutrient, and the mass of each nutrient is determined by its concentration, volume, and density. Based on this, the calculation formula for the mass constraint model is derived as follows: ,in, For the first The standard density of a nutrient. The formula takes the total mass of the regions in the collected data as input and outputs the quantitative constraint relationship between the total mass and the latent variables, realizing the data fusion of spectral detection and mass detection.

[0051] The compensated absorbance-fixed model is combined with the mass-constrained model to form a complete coupled model. This model takes absorbance and total mass of the regions from the collected data as inputs and latent variables as outputs, covering three major physical mechanisms: spectral absorption, scattering compensation, and mass conservation. Its expression is as follows: This coupled model associates the latent variables through constraints. After the coupled model is established, a variational inference optimization algorithm is used to solve for the latent variables. This algorithm approximates the true posterior distribution through an approximate posterior distribution, and can achieve efficient estimation of latent variables under limited computing power. The latent variables include the proportion of carbohydrate layer thickness, the proportion of protein layer thickness, the proportion of fat layer thickness, the average concentration of carbohydrates, the average concentration of protein, the average concentration of fat, and the total layer thickness. Its solution process follows the Bayesian inference framework, and sequentially completes the prior distribution setting, variational distribution construction, maximization of the lower bound of evidence, and parameter iterative optimization.

[0052] To simplify the subsequent calculation of KL divergence and ensure the convergence of the optimization process, conjugate prior distributions need to be set for each latent variable. The layer thickness percentage, as a probability distribution parameter, adopts the Dirichlet distribution as its prior distribution. This distribution is the conjugate prior of the multinomial distribution and naturally satisfies the constraint that the sum of the percentages is 1. The average nutrient concentration, as a non-negative continuous variable, adopts the Gamma distribution as its prior distribution. This distribution is the conjugate prior of the exponential distribution and is suitable for describing concentration-type physical quantities. The total layer thickness, as a continuous variable, adopts a normal distribution as its prior distribution, which can accommodate a reasonable range of layer thickness values. The prior distribution expressions for each latent variable are as follows: in, This is the shape parameter of the Dirichlet distribution, which is typically set to 1 to ensure the non-informative nature of the prior distribution. and The shape parameter and rate parameter of the Gamma distribution are respectively. As an option, the prior shape parameter of carbohydrates, proteins and fats are all set to 0.5, and the rate parameter is all set to 3, which is consistent with the concentration distribution range of common foods. and These represent the mean and variance of a normal distribution, with the mean taken as 1 cm and the variance taken as 0.25 cm. 2 It covers a range of layer thicknesses for common foods.

[0053] A variational distribution conjugate to the prior distribution is constructed. Using the mean-field variational inference hypothesis, the variational distribution is decomposed into the product of the latent variable sub-distributions. This hypothesis reduces the dimensionality of the optimization variables and adapts to computational power requirements. The expression for the variational distribution is: The variational distribution is conjugate to the prior distribution, i.e. It follows a Dirichlet distribution. It follows a Gamma distribution. If the distribution is normal, the parameters of each sub-distribution are the variational parameters to be optimized, and the specific expression is as follows: in, Let be the shape parameter of the variational Dirichlet distribution. For the shape and rate parameters of the variational Gamma distribution, The mean and variance of the variational normal distribution are given above, and these parameters together constitute the set of variational distribution parameters.

[0054] The goal of variational inference is to maximize the lower bound of evidence. The lower bound of evidence quantifies how closely the variational distribution approximates the true posterior distribution, and its calculation formula is as follows: ,in, As the lower bound of evidence, For a set of latent variables, This represents the expectation of the variational distribution. Let be the likelihood function, describing the degree of matching between the collected data and the latent variables. For variational distribution and prior distribution Divergence is used to constrain the rationality of variational distributions.

[0055] In the likelihood function calculation, due to the nonlinear characteristics of the coupled model, the expectation cannot be solved analytically. Therefore, the Monte Carlo sampling method is used for approximate calculation. M latent variable samples are collected from the variational distribution, and the expectation is approximated by the sample mean. The calculation formula is as follows: ,in, For the first There are several latent variable samples, with M representing the number of samples. As an option, M is set to 20 to control the computational load while ensuring computational accuracy. The specific form of the likelihood function is determined by the coupled model. The input is the latent variable samples and the collected data, and the output is the likelihood value of the data under that sample. An approximate estimate of the expectation of the likelihood term is achieved through sample averaging.

[0056] Since the KL divergence term is conjugate between the variational distribution and the prior distribution, it can be calculated analytically without numerical integration, thus reducing computational complexity. Taking the KL divergence of layer thickness percentage as an example, its calculation formula is as follows: ,in, Gamma function, Double Gamma function, the formula is directly derived using the conjugate property of Dirichlet distribution, the input is the variational parameter and the prior parameter, and the output is the KL divergence value. Similarly, the KL divergence of nutrient concentration and total layer thickness can be calculated analytically by the conjugate property of Gamma distribution and normal distribution, and the final KL divergence term is the sum of the KL divergence of each latent variable sub-distribution.

[0057] In the iterative optimization process, the momentum stochastic gradient descent algorithm is used to update the variational parameters, and the gradient of the evidence lower bound is used as the optimization direction to gradually adjust the variational parameters to maximize the evidence lower bound. First, calculate the gradient of the evidence lower bound with respect to each variational parameter, the gradient is composed of the gradient of the likelihood term and the gradient of the KL divergence term, then update the parameters combined with the momentum coefficient, the iterative formula is: , wherein, is the variational parameter set of the th iteration, is the learning rate, generally taking a value of 0.01, is the momentum coefficient, taking a value of 0.9, is the gradient of the previous iteration, is the gradient of the current iteration of the evidence lower bound.

[0058] The convergence condition of the iterative optimization is set to the change of the evidence lower bound in the last 5 rounds being less than a preset threshold, the threshold value is 10⁻ 6 , to ensure the stability of the solution. When the convergence condition is met, the iteration is terminated, and the expectation of each variational distribution is taken as the optimal estimate of the latent variable.

[0059] The heat calculation unit is used to calculate the mass of carbohydrates, proteins and fats in food according to the latent variable, and then obtain the total heat, and display the total heat.

[0060] Further, obtaining the total heat includes the following steps: Based on the optimal estimate of the latent variable, the dietary partition area of the plate body 1 and the standard density of nutrients, the mass of carbohydrates, proteins and fats in food is calculated respectively, and the nutrient-related data includes the mass and proportion of carbohydrates, proteins and fats in food; according to the mass, the standard unit heat coefficient of carbohydrates, proteins and fats is used to obtain the total heat of food by linear superposition; The total heat and the nutrient-related data are displayed through video display and voice broadcast.

[0061] Specifically, the heat calculation unit obtains the optimal estimate of the latent variable based on the modeling and solving unit, combines the preset nutrition parameters and plate structure parameters, and sequentially completes the mass calculation of carbohydrates, proteins and fats, total heat accounting, and presents the results to the user through a multi-modal display mode.

[0062] Generally, the calculation of nutrient quality takes the optimal estimate of latent variables as input, combines the area of the diet partition of the plate ontology and the standard density of nutrients to construct a quantitative relationship. The proportion of the layer thickness of total layer thickness carbohydrates, proteins and fats in the latent variables and the average concentration directly determine the volume proportion and material content of each nutrient. Then, the mass data is obtained by density conversion, and the calculation formula is: wherein, is the mass of the nth nutrient, corresponding to carbohydrates, proteins and fats, is the area of the diet partition corresponding to the plate ontology, is the optimal estimate of the total layer thickness of food in the latent variables, is the optimal estimate of the layer thickness proportion of the nth nutrient, is the optimal estimate of the average concentration of the nth nutrient, is the standard density of the nth nutrient. The formula input is the optimal estimate of the latent variables, the area of the diet partition and the standard density, and the output is the mass of a single nutrient, which realizes the quantitative conversion from physical parameters to nutritional quality and lays a foundation for subsequent calorie calculation. After obtaining the mass of the three nutrients, further calculation of the proportion parameters in the nutrient related data is carried out, and the calculation formula is: wherein, is the mass proportion of the nth nutrient, is the mass of carbohydrates, proteins and fats, respectively, and the sum of the three is consistent with the total mass of the diet partition in the collected data. Through this formula, the nutritional composition ratio of food can be intuitively reflected, providing a reference for the user's diet structure.

[0063] The calculation of total calories is based on the energy coefficient principle in nutrition, and adopts a linear superposition method, i.e. the mass of each nutrient is multiplied by its corresponding standard unit heat coefficient, and then summed to obtain the total heat of food. The calculation formula is: wherein, is the total heat of food, are the standard unit heat coefficients of carbohydrates, proteins and fats, respectively. The formula input is the mass of the three nutrients and the corresponding standard unit heat coefficient, and the output is the total heat of food, which realizes the direct conversion of nutritional quality to energy value, ensuring the accuracy and authority of calorie calculation.

[0064] The calculation of total calories is based on the energy coefficient principle in nutrition, and adopts a linear superposition method, i.e. the mass of each nutrient is multiplied by its corresponding standard unit heat coefficient, and then summed to obtain the total heat of food. The calculation formula is: wherein, is the total heat of food, are the standard unit heat coefficients of carbohydrates, proteins and fats, respectively. The formula input is the mass of the three nutrients and the corresponding standard unit heat coefficient, and the output is the total heat of food, which realizes the direct conversion of nutritional quality to energy value, ensuring the accuracy and authority of calorie calculation.

[0065] ​The display of total heat and nutrient related data is realized by video display and voice broadcast, ensuring that the user can conveniently obtain information in different scenarios. Generally, the video display module adopts an OLED touch screen, and the voice broadcast module adopts a micro speaker.

[0066] Working principle: in use, food is placed in the tray body 1, the protective cover 2 is flexibly rotated to the closed or open state according to the use requirement, and the rotating plate 12 is freely opened and closed relative to the protective cover 2, so that the heat detection module 19 is opposite the food in the tray body 1 when opened, and the heat detection module 19 is accommodated between the protective cover 2 and the rotating plate 12 when closed, so that the heat detection module 19 collects and calculates the heat and related nutrient composition data of the food in the tray body 1, providing real-time and quantitative dietary data reference for patients and performing fine diet management.

[0067] The tray is carried through the handle 3, and the handle 3 can be placed in the placing groove 4, facilitating the tray to be put into a bag. The handle 3 drives the rotating shaft 5 to make the sliding block slide in the sliding groove 15, thereby driving the limiting block 7 to move up and down, so that the hook part of the L-shaped limiting block 21 is clamped into the limiting groove 16 of the limiting block 7, thereby limiting the protective cover 2 on the tray body 1. When the protective cover 2 needs to be opened, the hook part of the L-shaped limiting block 21 is separated from the limiting groove 16 by pressing the pressing column 6, so that the protective cover 2 can be opened, thereby conveniently protecting the tray body 1 and the structure in the tray body 1 by the protective cover 2, and enhancing the practicability and convenience of use.

[0068] Although embodiments of the present application have been shown and described, it is to be understood that various modifications, substitutions, replacements and changes can be made to these embodiments without departing from the principles and spirit of the present application, and the scope of the present application is defined by the appended claims and their equivalents.

Claims

1. A dining plate for diabetic patients, comprising a plate body (1), characterized in that, The upper end of the plate body (1) is open. A protective cover (2) for closing the opening is installed at the opening of the plate body (1) via a second rotating shaft. A rotating plate (12) is installed inside the protective cover (2) via a first rotating shaft. A heat detection module (19) is installed inside the rotating plate (12). The heat detection module (19) is used to detect the heat of the food in the plate body (1). A partition plate (11) is rotatably mounted on the rotating plate (12) via a third rotating shaft. A partition block (8) is fixedly connected to the inner wall of the plate body (1). The partition block (8) is used to divide the inner cavity of the plate body (1) into multiple accommodating cavities. A locking structure is provided on the plate body (1). The locking structure is used to lock or unlock the protective cover (2) and the plate body (1).

2. A dining plate for diabetic patients according to claim 1, characterized in that: The first rotating shaft is the first damping rotating shaft (13), and there are two first damping rotating shafts (13) symmetrically installed on the protective cover (2). The second rotating shaft is the second damping rotating shaft (18), and there are two second damping rotating shafts (18) symmetrically installed on the plate body (1). The third rotating shaft is the third damping rotating shaft (20), and there are two third damping rotating shafts (20) symmetrically installed on the rotating plate (12). The partition block (8) is T-shaped.

3. A dining plate for diabetic patients according to claim 2, characterized in that: The plate body (1) and the protective cover (2) are provided with a placement groove (4) on one side and a sliding groove (15) on the other side. The plate body (1) and the protective cover (2) are provided with a sliding groove (15) symmetrically. The protective cover (2) is provided with a placement groove (9) on one side and a tableware groove (10) symmetrically on the other side. The rotating plate (12) is provided with a through hole (22) on one side. The locking structure includes a handle (3) and a pressing post (6). The outer wall of the pressing post (6) is slidably connected to the inside of the protective cover (2). The handle (3) is rotatably connected to a rotating shaft (5). The two ends of the rotating shaft (5) are respectively provided with sliders (14). The sliders (14) are slidably set in the sliding grooves (15) of the plate body (1) and the protective cover (2).

4. A dining plate for diabetic patients according to claim 3, characterized in that: A spring (17) is symmetrically fixedly connected to one side of the push post (6). One end of the spring (17) is fixedly connected to the inside of the protective cover (2). An L-shaped limiting block (21) is provided at the bottom of the push post (6). One end of the L-shaped limiting block (21) is provided with a first wedge-shaped surface.

5. A dining plate for diabetic patients according to claim 4, characterized in that: The middle part of the rotating shaft (5) is fixedly connected to a limiting block (7). A limiting groove (16) is opened on one side of the limiting block (7). The L-shaped limiting block (21) has a hook. A first wedge surface is set on the hook. The hook can hook with the limiting groove (16). A second wedge surface is set on the top of the limiting block (7). The first wedge surface of the limiting block (7) matches the second wedge surface of the L-shaped limiting block (21).

6. A dining plate for diabetic patients according to any one of claims 1-5, characterized in that: The heat detection module (19) includes a data acquisition unit, a modeling and solving unit and a heat calculation unit. The data acquisition unit is used to collect the spectral signal, mass data and ambient temperature data of the food in the plate body (1), and after preprocessing, the collected data is obtained. The modeling and solving unit is used to build a coupled model based on the collected data and solve the latent variables of food through variational inference optimization algorithm; The calorie calculation unit is used to calculate the mass of carbohydrates, proteins, and fats in food based on latent variables, thereby obtaining the total calories, and then displaying the total calories.

7. A dining plate for diabetic patients according to claim 6, characterized in that: The spectral signal includes three characteristic band signals corresponding to the near-infrared characteristic absorption peaks of carbohydrates, proteins and fats in food. The preprocessing includes filtering, temperature drift correction, absorbance conversion and data format standardization.

8. A dining plate for diabetic patients according to claim 6, characterized in that: The establishment of the coupling model specifically includes the following steps: Based on the collected data, a multi-component, multi-level light absorption stabilization model was established using the Lambert-Beer law. The light absorption stabilization model includes the quantitative relationship between the absorbance of each characteristic band and the concentration of food components, the proportion of layer thickness, and the total layer thickness. A scattering correction term based on the Rayleigh-Gans approximation is introduced into the light absorption stabilization model to compensate for the scattering distortion of the spectral signal, resulting in a compensated scattering correction term. Based on the collected data, a mass constraint model is established using the law of conservation of mass. The mass constraint model is used to correlate the total mass of food with component concentration, total layer thickness and nutrient density. The light absorption fixed model, the compensation scattering correction term, and the mass constraint model are combined to form a coupled model.

9. A dining plate for diabetic patients according to claim 6, characterized in that: The method of solving for the latent variables of food using variational inference optimization algorithm specifically includes the following steps: Based on the coupled model, a conjugate prior distribution is set for each latent variable, which includes the proportion of carbohydrate layer thickness, the proportion of protein layer thickness, the proportion of fat layer thickness, the average concentration of carbohydrates, the average concentration of protein, the average concentration of fat, and the total layer thickness. Construct a variational distribution conjugate to the prior distribution, obtain the variational distribution parameters, and decompose the variational distribution into the product of the latent variable sub-distributions; The variational distribution parameters are optimized by maximizing the lower bound of evidence. The expectation of the likelihood term is calculated using Monte Carlo sampling, and the KL divergence term is calculated analytically. The optimization is iterative until the lower bound of evidence converges, and the expectation of the variational distribution is taken as the optimal estimate of the latent variable.

10. A dining plate for diabetic patients according to claim 6, characterized in that: The process of obtaining the total heat specifically includes the following steps: Based on the optimal estimate of the latent variables, the area of ​​the food partition of the plate (1) and the standard density of nutrients, the mass of carbohydrates, proteins and fats in the food are calculated respectively. Nutrient-related data include the mass and proportion of carbohydrates, proteins and fats in the food. Based on mass, the total calories of food are obtained by linearly superimposing the standard unit calorie coefficients of carbohydrates, proteins and fats. Total calorie and nutrient data are displayed through video and voice announcements.

Images