Enterprise employee dining and ordering management system based on big data

Abstract

The invention relates to the technical field of meal ordering management, in particular to an enterprise employee meal ordering management system based on big data, which comprises a meal nutrition demand determination module for determining meal nutrition demands through a nutrition analysis model according to historical meal data and personal information of employees, the nutrition analysis model is established on the basis of an improved CNN-GRU-Attention network architecture; the candidate food determining module is used for selecting candidate food from the ordering food nutrition library through an improved clustering algorithm according to the food nutrition requirements; the heat distribution module is used for determining the heat distribution proportion of multiple meals on the day according to the personal information of the employees, and determining a recommended meal sequence in the candidate meal through a meal determination model based on a genetic algorithm; and the meal ordering management module displays the recommended meal sequence when the employees order meals, and uses the meal ordering results of the employees to update the historical meal data. According to the invention, personalized nutritional requirements of employees are met, and the employees are helped to keep diet balance and a better working state.

Description

Technical Field

[0001] The present invention relates to the technical field of meal ordering management, and in particular to a meal ordering management system for enterprise employees based on big data. Background Art

[0002] In my country, chronic diseases such as overweight, obesity, dyslipidemia, hypertension, and diabetes have become part of everyday life. Reports indicate that 260 million people, or nearly one in five, have been diagnosed with a chronic disease. These diseases present both inconvenience and a threat, making them a widespread public health concern. This is particularly true for white-collar workers, many of whom suffer from some degree of chronic illness or overwork.

[0003] The root cause is often an imbalance of nutrients in the human body. A balanced diet ensures a balanced balance of nutrients, allowing the body to function properly, thereby preventing and alleviating chronic diseases among employees. Therefore, companies should provide personalized nutritional recommendations when ordering group meals for employees to help them achieve a balanced diet and a better work environment.

[0004] For example, patent application number 201611221988.X discloses a nutritional meal planning method, including: receiving nutritional survey information sent by a first user terminal, wherein the nutritional survey information includes basic information and attachment information, and sending the attachment information to a second user terminal so that the second user terminal processes the attachment information and obtains processing result information; obtaining nutritional information based on the basic information, and sending the nutritional information to a second user terminal so that the second user terminal obtains nutritional analysis report information based on the nutritional information and the processing result information; and receiving the nutritional analysis report information sent by the second user terminal to improve the efficiency of nutritional meal planning services. However, this patent document fails to provide meal order recommendations based on the personalized nutritional needs of corporate employees.

[0005] Therefore, how to implement meal recommendations that meet employees' personalized nutritional needs in the company's employee meal ordering management system is a technical problem that needs to be solved. Summary of the Invention

[0006] To this end, the present invention provides a meal ordering management system for corporate employees based on big data. Through the CNN-GRU-Attention nutritional analysis model, accurate nutritional demand prediction can be achieved by combining employee personal information and meal cycle patterns, and a recommended meal sequence can be generated through an improved clustering algorithm of operations research. Through the cross-use of machine learning and operations research, corporate group meal ordering recommendations that meet employees' personalized nutritional needs and nutritional guidelines are achieved, thereby helping employees maintain a balanced diet and a better working state.

[0007] To achieve the above objectives, the present invention proposes a meal ordering management system for enterprise employees based on big data, comprising:

[0008] A module for determining nutritional needs for meals, which is used to determine nutritional needs for meals based on employees' historical meal data and personal information using a nutritional analysis model, wherein the nutritional analysis model is based on an improved CNN-GRU-Attention network architecture;

[0009] a candidate meal determination module, configured to select candidate meals from a meal ordering nutrient library using an improved clustering algorithm based on the nutritional requirements of the meal;

[0010] a calorie distribution module, configured to determine a calorie distribution ratio for multiple meals on the same day based on the employee's personal information, and determine a recommended meal sequence from the candidate meals based on the calorie distribution ratio using a meal determination model based on a genetic algorithm;

[0011] The meal ordering management module is used to display the recommended meal sequence when employees order meals, and use the employee's ordering results to update the historical meal data.

[0012] Furthermore, the nutrition analysis model includes a convolutional neural network sub-model, a recurrent neural network sub-model, and an attention mechanism; the personal information includes physical information, work and rest information, and health indicators; and the dining nutritional needs determination module includes a dietary pattern generation unit, a needs generation unit, and a dynamic adjustment unit;

[0013] The dietary pattern generating unit is configured to generate dietary time series features from the historical dining data through the convolutional neural network sub-model;

[0014] The demand generation unit is configured to generate the dietary cyclical regularity by using the dietary time series characteristics, the physical information, the work and rest information, and the health index through the recurrent neural network sub-model to generate the dietary nutritional requirements;

[0015] The dynamic adjustment unit is used to control the update of the hidden state of the recurrent neural network sub-model through the attention mechanism based on the target meal nutritional requirements.

[0016] Furthermore, the hidden state includes a transferred activation state, a current activation state, a previous time step activation state, and an updated activation state, and the dynamic adjustment unit includes a weight generation subunit and a hidden state dynamic adjustment subunit;

[0017] The weight generating subunit is configured to generate an updated weight based on the target meal nutritional requirement and the current activation state through the attention mechanism;

[0018] The hidden state dynamic adjustment subunit is used to update the current activation state according to the update weight, the transfer activation state and the activation state of the previous time step.

[0019] Furthermore, the demand generation unit includes a data fusion subunit, a recurrent neural network operation subunit and an output subunit;

[0020] The data fusion subunit is used to generate input data by encoding and mapping the dietary time series characteristics, the physical information, the work and rest information and the health indicators;

[0021] The recurrent neural network operation subunit is used to perform iterative operations on the input data, the current activation state and the current activation state through the recurrent neural network submodel to generate multiple transfer activation states;

[0022] The output subunit is used to perform feature splicing on the multiple transfer activation states to generate the meal nutritional requirements.

[0023] Furthermore, the demand generation unit further includes an objective function subunit;

[0024] The objective function subunit is used to optimize the parameters of the nutritional analysis model by improving the objective function, wherein the improved objective function is constructed based on cross entropy and sigmoid activation function.

[0025] In the above scheme, the hidden state of the gated recurrent unit is updated by calculating the similarity between the current activation state and the target item through the attention mechanism. Combined with the potential sequence characteristics of timing parameters, it can effectively improve the performance and accuracy of the meal nutritional needs prediction task.

[0026] Furthermore, the improved clustering algorithm is a Kmeans clustering algorithm, and the candidate meal determination module includes a weighted distance calculation unit, a clustering operation unit, and a meal combination generation unit;

[0027] The weighted distance calculation unit is used to calculate the weighted Euclidean distance between the nutritional elements of the meal nutritional requirements and the nutritional elements of each meal in the ordered meal nutritional database to generate a distance function;

[0028] The clustering operation unit is used to generate multiple meal combination clusters in the ordered meal nutrition library by using an improved clustering algorithm based on the distance function calculation;

[0029] The meal combination generating unit is configured to select the candidate meal from the meal combination cluster according to the employee's dining preference.

[0030] Furthermore, the heat distribution module includes a heat calculation unit and a solution unit;

[0031] The calorie calculation unit is used to construct a nutrient intake regression model based on the nutrient content of the candidate meal and the portion size of the candidate meal;

[0032] The solving unit is used to solve the nutrient element regression model through a genetic algorithm to generate a recommended meal sequence.

[0033] In the above solution, through the Kmeans clustering algorithm, regression model construction and genetic algorithm, it is possible to efficiently select scientific and personalized recommended meals from the nutrition library of ordered meals containing a large number of meals.

[0034] Furthermore, the heat distribution module further includes a ratio adjustment unit and a metabolic rate adjustment unit;

[0035] The ratio adjustment unit is used to adjust the preheating heat distribution ratio according to the work and rest information and the health index to generate an adjusted heat distribution ratio;

[0036] The metabolic rate adjustment unit is used to calculate the metabolic rate according to the body information and the health index, and adjust the calorie distribution ratio according to the metabolic rate to generate the calorie distribution ratio.

[0037] Furthermore, the nutritional requirements for meals include carbohydrates, dietary fiber, vitamins and trace elements.

[0038] Furthermore, the dining nutritional requirement determination module, the candidate meal determination module and the calorie allocation module are installed on the assa cloud platform server, and the meal ordering management module is installed on the employee terminal that communicates with the assa cloud platform server.

[0039] Compared with the prior art, the present invention has the following advantages:

[0040] 1. The CNN-GRU-Attention nutritional analysis model enables accurate nutritional needs prediction based on employee personal information and meal cycle patterns. An improved clustering algorithm based on operations research is used to generate recommended meal sequences. By combining machine learning and operations research, we can recommend corporate group meal orders that meet employees' personalized nutritional needs and nutritional guidelines, helping them maintain a balanced diet and improve their work performance.

[0041] 2. The hidden state of the gated recurrent unit is updated by calculating the similarity between the current activation state and the target item through the attention mechanism. Combined with the potential sequence characteristics of timing parameters, it can effectively improve the performance and accuracy of the meal nutritional needs prediction task.

[0042] 3. Through the Kmeans clustering algorithm, regression model construction and genetic algorithm, it is possible to efficiently select scientific and personalized recommended meals from the nutrition library of ordered meals containing a large number of meals. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 This is a structural diagram of a big data-based enterprise employee meal ordering management system according to an embodiment of the present invention;

[0044] Figure 2 This is a flow chart of a big data-based enterprise employee meal ordering management system according to an embodiment of the present invention;

[0045] Figure 3 This is a structural diagram of a nutrition analysis model of a big data-based enterprise employee meal ordering management system according to an embodiment of the present invention;

[0046] Figure 4 This is a structural diagram of the recurrent neural network sub-model and attention mechanism of the nutritional analysis model of the big data-based enterprise employee meal ordering management system in an embodiment of the present invention. DETAILED DESCRIPTION

[0047] In order to make the objects and advantages of the present invention more clearly understood, the present invention is further described below in conjunction with embodiments; it should be understood that the specific embodiments described herein are merely used to explain the present invention and are not intended to limit the present invention.

[0048] The preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood by those skilled in the art that these embodiments are only used to explain the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0049] It should be noted that, in the description of the present invention, terms such as "up", "down", "left", "right", "inside", and "outside" indicating directions or positional relationships are based on the directions or positional relationships shown in the accompanying drawings. This is only for the convenience of description and does not indicate or imply that the device or element must have a specific orientation, be constructed and operated in a specific orientation. Therefore, it cannot be understood as a limitation on the present invention.

[0050] Furthermore, it should be noted that, in the description of the present invention, unless otherwise expressly specified or limited, the terms "mounted," "connected," and "connected" should be understood in a broad sense. For example, they may refer to fixed connections, detachable connections, or integral connections; mechanical connections or electrical connections; direct connections or indirect connections through an intermediate medium; and internal communication between two components. Those skilled in the art will understand the specific meanings of the above terms in the present invention based on specific circumstances.

[0051] like Figures 1 to 4 As shown, the present invention provides a meal ordering management system for enterprise employees based on big data. Through the CNN-GRU-Attention nutritional analysis model, accurate nutritional demand prediction can be achieved by combining employee personal information and meal cycle patterns, and a recommended meal sequence can be generated through an improved clustering algorithm of operations research. Through the cross-use of machine learning and operations research, enterprise group meal ordering recommendations that meet the personalized nutritional needs and nutritional guidelines of employees are achieved, so as to help employees maintain a balanced diet and a better working state.

[0052] like Figures 1 to 4 As shown, this embodiment proposes a meal ordering management system for enterprise employees based on big data, including:

[0053] A module for determining nutritional needs for meals, which is used to determine nutritional needs for meals based on employees' historical meal data and personal information using a nutritional analysis model, wherein the nutritional analysis model is based on an improved CNN-GRU-Attention network architecture;

[0054] a candidate meal determination module, configured to select candidate meals from a meal ordering nutrient library using an improved clustering algorithm based on the nutritional requirements of the meal;

[0055] a calorie distribution module, configured to determine a calorie distribution ratio for multiple meals on the same day based on the employee's personal information, and determine a recommended meal sequence from the candidate meals based on the calorie distribution ratio using a meal determination model based on a genetic algorithm;

[0056] The meal ordering management module is used to display the recommended meal sequence when employees order meals, and use the employee's ordering results to update the historical meal data.

[0057] It is understandable that employees' historical dining data includes the types of meals they have eaten and the content of various nutrients they have consumed during meals, forming big data stored in the information database of the enterprise management platform.

[0058] Understandably, in the CNN-GRU-Attention (convolutional neural network, gated recurrent unit, and attention mechanism) network architecture, the CNN uses convolutional layers to extract local features from employees' historical dining data, including time periods, meal combinations, and nutritional intake. The GRU is used to capture the impact of nutritional intake trends over multiple days and personal information on current needs. The attention mechanism assigns higher weights to key time steps in historical data, increasing the model's sensitivity to time points. This allows the model to integrate employees' static personal information and dynamic dining behaviors, ensuring that predictions are tailored to individual characteristics and behavioral changes.

[0059] Furthermore, if Figure 2 and4 As shown, the nutrition analysis model includes a convolutional neural network sub-model, a recurrent neural network sub-model and an attention mechanism; the personal information includes physical information, work and rest information and health indicators; the dining nutritional demand determination module includes a diet pattern generation unit, a demand generation unit and a dynamic adjustment unit;

[0060] The dietary pattern generating unit is configured to generate dietary time series features from the historical dining data through the convolutional neural network sub-model;

[0061] The demand generation unit is configured to generate the dietary cyclical regularity by using the dietary time series characteristics, the physical information, the work and rest information, and the health index through the recurrent neural network sub-model to generate the dietary nutritional requirements;

[0062] The dynamic adjustment unit is used to control the update of the hidden state of the recurrent neural network sub-model through the attention mechanism based on the target meal nutritional requirements.

[0063] Specifically, if Figure 4 As shown in FIG, the convolutional neural network sub-model includes a multi-scale parallel convolution layer (Multi-Scale Convolution) and an expanded convolution layer to synchronously capture daily / weekly / monthly dietary patterns and expand the temporal receptive field.

[0064] Furthermore, if Figure 2 and 4 As shown, the hidden state includes a transfer activation state, a current activation state, a previous time step activation state and an update activation state, and the dynamic adjustment unit includes a weight generation subunit and a hidden state dynamic adjustment subunit;

[0065] The weight generating subunit is configured to generate an updated weight based on the target meal nutritional requirement and the current activation state through the attention mechanism;

[0066] The hidden state dynamic adjustment subunit is used to update the current activation state according to the update weight, the transfer activation state and the activation state of the previous time step.

[0067] Specifically, if Figure 4 As shown, the weight generation subunit is:

[0068]

[0069] Where, Attention t Indicates the updated weight, softmax(e t ) represents the softmax activation function controlling the target meal nutritional requirement e at time t it Whether to update, h t represents the current activation state, and W represents the learnable weight vector.

[0070] Specifically, if Figure 4 As shown, the hidden state dynamic adjustment subunit is:

[0071]

[0072] Where, Attention t represents the updated weight, h t Indicates the current activation state, h t-1 represents the activation state of the previous time step, h′ t represents the transferred activation state. Therefore, it determines the contribution ratio of the activation state at the previous time step and the transferred activation state to the current activation state. Dynamically adjusting the current hidden state through weighted summation allows the model to synthesize the previous state information according to different update weights to obtain the current state.

[0073] It is understandable that by controlling the update of the current activation state through the attention mechanism, the contribution weight of different modal features to the output of meal nutritional needs can be controlled. For example, if an employee works overtime for many consecutive days (changes in work and rest), the reset gate reduces the weight of historical dietary patterns and focuses on the impact of current work intensity on nutritional needs; the update gate retains the constraints of long-term health indicators (such as blood lipid levels) during holidays to avoid misleading predictions due to short-term dietary fluctuations; temporary overtime requires an increase in the intake ratio of nutrients such as protein to maintain employee working status.

[0074] Furthermore, if Figure 2 and 4 As shown, the demand generation unit includes a data fusion subunit, a recurrent neural network operation subunit and an output subunit;

[0075] The data fusion subunit is used to generate input data by encoding and mapping the dietary time series characteristics, the physical information, the work and rest information and the health indicators;

[0076] The recurrent neural network operation subunit is used to perform iterative operations on the input data, the current activation state and the current activation state through the recurrent neural network submodel to generate multiple transfer activation states;

[0077] The output subunit is used to perform feature splicing on the multiple transfer activation states to generate the meal nutritional requirements.

[0078] Specifically, if Figure 4 As shown, the weight generation subunit is:

[0079]

[0080] Where, Attention t Indicates the updated weight, softmax(e t ) represents the softmax activation function controlling the target meal nutritional requirement e at time t i t Whether to update h′ t ′ represents the pre-updated activation state derived from the current activation state, and W represents the learnable weight vector.

[0081] Specifically, if Figure 4 As shown, the hidden state dynamic adjustment subunit is:

[0082]

[0083] Where, Attention t represents the updated weight, h t Indicates the transfer activation state, h t-1 represents the activation state of the previous time step, h′ t represents the current activation state. Therefore, it determines the contribution ratio of the activation state at the previous time step and the transferred activation state to the current activation state. Dynamically adjusting the current hidden state through weighted summation allows the model to synthesize the previous state information according to different update weights to obtain the current state.

[0084] It is understandable that by controlling the update of the current activation state through the attention mechanism, the contribution weight of different modal features to the output of meal nutritional needs can be controlled. For example, if an employee works overtime for many consecutive days (changes in work and rest), the reset gate reduces the weight of historical dietary patterns and focuses on the impact of current work intensity on nutritional needs; the update gate retains the constraints of long-term health indicators (such as blood lipid levels) during holidays to avoid misleading predictions due to short-term dietary fluctuations; temporary overtime requires an increase in the intake ratio of nutrients such as protein to maintain employee working status.

[0085] Furthermore, if Figure 2 and 4 As shown, the demand generation unit includes a data fusion subunit, a recurrent neural network operation subunit and an output subunit;

[0086] The data fusion subunit is used to generate input data by encoding and mapping the dietary time series characteristics, the physical information, the work and rest information and the health indicators;

[0087] The recurrent neural network operation subunit is used to perform iterative operations on the input data, the current activation state and the current activation state through the recurrent neural network submodel to generate multiple transfer activation states;

[0088] The output subunit is used to perform feature splicing on the multiple transfer activation states to generate the meal nutritional requirements.

[0089] Specifically, if Figure 4 As shown, the data fusion subunit standardizes and pre-processes the physical information (age, gender, BMI), health indicators (blood sugar, blood lipids), work and rest status (sleep time, work intensity), and job code (occupational category) for vector mapping, generates sparse binary vectors through one-hot encoding, and maps the sparse binary vectors corresponding to each data to the target dimension separately through full-connected layer projection to generate multimodal features, and performs feature splicing on the multimodal features to generate the input data.

[0090] Specifically, if Figure 4 As shown, the target meal nutritional requirement is the recommended meal nutritional intake corresponding to the employee's physical information. It is static data and therefore needs to be dynamically adjusted through GRU.

[0091] Specifically, if Figure 4 As shown, the recurrent neural network operator unit is:

[0092]

[0093] z t =σ(W z e t +U z h t-1 +b z )

[0094] r t =σ(W r e t +U r h t-1 +b r )

[0095] Where h t 、h t-1 , h′ t , h′ t ′ respectively represent the transfer activation state, the activation state of the previous time step, the current activation state of the current time step, and the pre-update activation state, e t represents the input embedding at time t, σ is the Sigmoid activation function, z t 、e t 、r t Represent the update gate, input data, and reset gate of the current time step, respectively. h 、U h 、W z 、U z 、Wr 、U r is the learnable weight vector, b h 、b z 、b r Both are learnable bias terms. The previous hidden state and the transfer hidden state are updated according to the update gate, and their summation is used to obtain the pre-update activation state. The value of the transfer hidden state is updated through the transfer subunit. The reset gate is used to determine whether the input embedding and the previous hidden state need to be updated.

[0096] Specifically, if Figure 4 As shown, the output subunit is:

[0097] H=Concat(h t )=(h1,h2,…,h k )

[0098] Where H represents the nutritional requirement of meals, Concat represents feature concatenation, h1,h2,…,h k It represents the transfer activation state of the first time step, the transfer activation state of the second time step, and the transfer activation state of the current k-th time step in sequence.

[0099] It's understandable that CNN-GRU-Attention uses historical data to understand a user's dietary patterns and background, effectively learning the difference between a user's underlying dietary patterns and their dynamic nutritional goals, and thus predicting their nutritional needs. Nutritional needs can vary significantly across different physical conditions and timeframes.

[0100] Furthermore, if Figure 2 As shown, the demand generation unit also includes an objective function subunit;

[0101] The objective function subunit is used to optimize the parameters of the nutritional analysis model by improving the objective function, wherein the improved objective function is constructed based on cross entropy and sigmoid activation function.

[0102] Specifically, the objective function subunit is:

[0103] L=-Σ[ylogp+(1-y)log(1-p)]+λ||W,b|| 2

[0104] Where L is the objective function, y represents the label value, p represents the predicted value, F is the output of the weight generation subunit, W is the learnable weight vector, b is the learnable bias term, and λ represents regularization.

[0105] In the above scheme, the hidden state of the gated recurrent unit is updated by calculating the similarity between the current activation state and the target item through the attention mechanism. Combined with the potential sequence characteristics of timing parameters, it can effectively improve the performance and accuracy of the meal nutritional needs prediction task.

[0106] Furthermore, the improved clustering algorithm is a Kmeans clustering algorithm, and the candidate meal determination module includes a weighted distance calculation unit, a clustering operation unit, and a meal combination generation unit;

[0107] The weighted distance calculation unit is used to calculate the weighted Euclidean distance between the nutritional elements of the meal nutritional requirements and the nutritional elements of each meal in the ordered meal nutritional database to generate a distance function;

[0108] The clustering operation unit is used to generate multiple meal combination clusters in the ordered meal nutrition library by using an improved clustering algorithm based on the distance function calculation;

[0109] The meal combination generating unit is configured to select the candidate meal from the meal combination cluster according to the employee's dining preference.

[0110] Specifically, the weighted distance calculation unit generates a distance function by weighted Euclidean distance according to the nutritional demand priority:

[0111]

[0112] Where Dist represents the distance function, A1, A2, ... represent the nutrient values of the meal's nutritional requirements, B1, B2, ... represent the nutrient values of each meal item in the ordered meal's nutritional database, and ω1, ω2, ... represent the priority weights of each nutrient element. It will be appreciated that these priority weights can be set based on dietary guidelines.

[0113] Specifically, the clustering operation unit initializes cluster centers by randomly selecting from meals that meet the total calorie restriction, and distributes the meals to clusters that meet the constrained nutritional standards through Kmeans clustering calculation.

[0114] Furthermore, the heat distribution module includes a heat calculation unit and a solution unit;

[0115] The calorie calculation unit is used to construct a nutrient intake regression model based on the nutrient content of the candidate meal and the portion size of the candidate meal;

[0116] The solving unit is used to solve the nutrient element regression model through a genetic algorithm to generate a recommended meal sequence.

[0117] Specifically, the regression model is:

[0118] A 1 X1 :A 2 X 2 :A 3 X 3 =C 1 :C 2 :C 3

[0119] Where C 1 :C 2 :C 3 is the calorie distribution ratio, i.e., breakfast calorie distribution: lunch calorie distribution: dinner calorie distribution, A 1 、A 2 、A 3 Represents the nutritional content of breakfast, lunch and dinner respectively, according to the "Nutrient Table", X 1 、X 2 、X 3 Represent the minimum intake weight of breakfast, lunch and dinner respectively, where each parameter is in matrix form, that is, C = [C ij ]、A=[A ij ]、X=[X ij ], i represents the i-th food, and j represents the j-th nutrient element.

[0120] The genetic algorithm, in a regression model, simulates operations such as selection, crossover, and mutation in biological evolution, globally searching for optimal parameters within the solution space of minimum intake weight X. Therefore, based on the meal combinations derived by the clustering operation unit, the meal combination generation unit iterates the genetic algorithm, using minimizing nutritional deviation from user needs as the objective function, to solve for various meal combinations for breakfast, lunch, and dinner, generating a recommended meal sequence and setting meal diversity constraints, such as requiring at least one staple food and a set amount of protein.

[0121] In the above solution, through the Kmeans clustering algorithm, regression model construction and genetic algorithm, it is possible to efficiently select scientific and personalized recommended meals from the nutrition library of ordered meals containing a large number of meals.

[0122] Furthermore, the heat distribution module further includes a ratio adjustment unit and a metabolic rate adjustment unit;

[0123] The ratio adjustment unit is used to adjust the preheating heat distribution ratio according to the work and rest information and the health index to generate an adjusted heat distribution ratio;

[0124] The metabolic rate adjustment unit is used to calculate the metabolic rate according to the body information and the health index, and adjust the calorie distribution ratio according to the metabolic rate to generate the calorie distribution ratio.

[0125] Specifically, the metabolic rate is determined by the Mifflin-StJeor formula for calculating the basal metabolic rate (BMR).

[0126] Specifically, the pre-calorie allocation was 30% for breakfast, 40% for lunch, and 30% for dinner. The ratio adjustment unit intervened, taking into account the employee's habit of exercising between 7 and 8 pm and his health indicators showing an 18% body fat percentage. The resulting allocation was 25% for breakfast, 35% for lunch, 30% for dinner, and 10% for a post-workout snack. The metabolic rate adjustment unit intervened, calculating the employee's metabolic rate at 1600 kcal, which was high, and set the nutrient ratios to 55% carbohydrates, 25% protein, and 20% fat. The final calorie allocation was 2232 kcal total, consisting of 620 kcal for breakfast, 868 kcal for lunch, and 744 kcal for dinner.

[0127] Furthermore, the nutritional requirements for meals include carbohydrates, dietary fiber, vitamins and trace elements.

[0128] Furthermore, the dining nutritional requirement determination module, the candidate meal determination module and the calorie allocation module are installed on the assa cloud platform server, and the meal ordering management module is installed on the employee terminal that communicates with the assa cloud platform server.

[0129] Specifically, the ASSA cloud platform server has a management server that manages the access rights of employees' personal information. The management server uses the user's account number, name, and unique identity code, and uses their name, ID number, and a specific key to calculate a fixed-length string through a one-way irreversible encryption algorithm (such as the domestic encryption algorithm SM3). This unique identity code is unique and irreversible, and its effect is equivalent to its ID number. Using this identity code can not only play the role of identity authentication, but also hide the real ID number, which is more in line with the requirements of personal privacy and security protection.

[0130] In this embodiment, the nutritional analysis model of CNN-GRU-Attention can be used to accurately predict the nutritional needs of employees based on their personal information and meal cycle patterns, and a recommended meal sequence can be generated through an improved clustering algorithm based on operations research. Through the cross-application of machine learning and operations research, corporate group meal ordering recommendations that meet employees' personalized nutritional needs and nutritional guidelines are achieved to help employees maintain a balanced diet and a better working state. By calculating the similarity between the current activation state and the target item through the attention mechanism to update the hidden state of the gated recurrent unit, combined with the potential sequence characteristics of temporal parameters, it can effectively improve the performance and accuracy of the task of predicting nutritional needs for meals. Through the Kmeans clustering algorithm, regression model construction and genetic algorithm, it is possible to efficiently select scientific and personalized recommended meals from a meal ordering meal nutrition library containing a large number of meals.

[0131] Thus far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art may make equivalent changes or substitutions to the relevant technical features, and the technical solutions after such changes or substitutions will fall within the scope of protection of the present invention.

[0132] The foregoing description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Those skilled in the art will readily appreciate that the present invention is susceptible to various modifications and variations. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention are intended to be within the scope of protection of the present invention.

Claims

1. A meal ordering management system for enterprise employees based on big data, characterized in that: include: A module for determining nutritional needs for meals, which is used to determine nutritional needs for meals based on employees' historical meal data and personal information using a nutritional analysis model, wherein the nutritional analysis model is based on an improved CNN-GRU-Attention network architecture; a candidate meal determination module, configured to select candidate meals from a meal ordering nutrient library using an improved clustering algorithm based on the nutritional requirements of the meal; a calorie distribution module, configured to determine a calorie distribution ratio for multiple meals on the same day based on the employee's personal information, and determine a recommended meal sequence from the candidate meals based on the calorie distribution ratio using a meal determination model based on a genetic algorithm; The meal ordering management module is used to display the recommended meal sequence when employees order meals, and use the employee's ordering results to update the historical meal data.

2. The enterprise employee meal ordering management system based on big data according to claim 1 is characterized in that: The nutrition analysis model includes a convolutional neural network sub-model, a recurrent neural network sub-model, and an attention mechanism. The personal information includes physical information, work and rest information, and health indicators. The dining nutritional needs determination module includes a dietary pattern generation unit, a demand generation unit, and a dynamic adjustment unit. The dietary pattern generating unit is configured to generate dietary time series features from the historical dining data through the convolutional neural network sub-model; The demand generation unit is configured to generate the dietary cyclical regularity by using the dietary time series characteristics, the physical information, the work and rest information, and the health index through the recurrent neural network sub-model to generate the dietary nutritional requirements; The dynamic adjustment unit is used to control the update of the hidden state of the recurrent neural network sub-model through the attention mechanism based on the target meal nutritional requirements.

3. The enterprise employee meal ordering management system based on big data according to claim 2 is characterized in that: The hidden state includes a transfer activation state, a current activation state, a previous time step activation state and an updated activation state, and the dynamic adjustment unit includes a weight generation subunit and a hidden state dynamic adjustment subunit; The weight generating subunit is configured to generate an updated weight based on the target meal nutritional requirement and the current activation state through the attention mechanism; The hidden state dynamic adjustment subunit is used to update the current activation state according to the update weight, the transfer activation state and the activation state of the previous time step.

4. The enterprise employee meal ordering management system based on big data according to claim 3 is characterized in that: The demand generation unit includes a data fusion subunit, a recurrent neural network operation subunit and an output subunit; The data fusion subunit is used to generate input data by encoding and mapping the dietary time series characteristics, the physical information, the work and rest information and the health indicators; The recurrent neural network operation subunit is used to perform iterative operations on the input data, the current activation state and the current activation state through the recurrent neural network submodel to generate multiple transfer activation states; The output subunit is used to perform feature splicing on the multiple transfer activation states to generate the meal nutritional requirements.

5. The enterprise employee meal ordering management system based on big data according to claim 2 is characterized in that: The demand generation unit also includes an objective function subunit; The objective function subunit is used to optimize the parameters of the nutritional analysis model by improving the objective function, wherein the improved objective function is constructed based on cross entropy and sigmoid activation function.

6. The enterprise employee meal ordering management system based on big data according to claim 1 is characterized in that: The improved clustering algorithm is a Kmeans clustering algorithm, and the candidate meal determination module includes a weighted distance calculation unit, a clustering operation unit, and a meal combination generation unit; The weighted distance calculation unit is used to calculate the weighted Euclidean distance between the nutritional elements of the meal nutritional requirements and the nutritional elements of each meal in the ordered meal nutritional database to generate a distance function; The clustering operation unit is used to generate multiple meal combination clusters in the ordered meal nutrition library by using an improved clustering algorithm based on the distance function calculation; The meal combination generating unit is configured to select the candidate meal from the meal combination cluster according to the employee's dining preference.

7. The enterprise employee meal ordering management system based on big data according to claim 2 is characterized in that: The heat distribution module includes a heat calculation unit and a solution unit; The calorie calculation unit is used to construct a nutrient intake regression model based on the nutrient content of the candidate meal and the portion size of the candidate meal; The solving unit is used to solve the nutrient element regression model through a genetic algorithm to generate a recommended meal sequence.

8. The enterprise employee meal ordering management system based on big data according to claim 7 is characterized in that: The heat distribution module also includes a ratio adjustment unit and a metabolic rate adjustment unit; The ratio adjustment unit is used to adjust the preheating heat distribution ratio according to the work and rest information and the health index to generate an adjusted heat distribution ratio; The metabolic rate adjustment unit is used to calculate the metabolic rate according to the body information and the health index, and adjust the calorie distribution ratio according to the metabolic rate to generate the calorie distribution ratio.

9. The enterprise employee meal ordering management system based on big data according to any one of claims 1 to 8, characterized in that: The nutritional requirements for meals include carbohydrates, dietary fiber, vitamins and trace elements.

10. The enterprise employee meal ordering management system based on big data according to any one of claims 1 to 8, characterized in that: The dining nutritional requirement determination module, the candidate meal determination module and the calorie allocation module are installed on the assa cloud platform server, and the meal ordering management module is installed on the employee terminal that communicates with the assa cloud platform server.

Images