Intelligent cooking control system based on big data

Abstract

The invention discloses a cooking intelligent control system based on big data. The cooking intelligent control system comprises a cooking control data acquisition module, a data optimization processing module, a healthy dish intelligent recommendation module, a cooking control parameter optimization module and an intelligent cooking execution control module. The invention relates to the technical field of information data processing, in particular to an intelligent cooking control system based on big data, which innovatively proposes a cooking dish recommendation individuation, health control double-layer and cooking process practicability mechanism, and significantly improves the health guidance ability, recommendation adaptation ability and operation effectiveness of intelligent cooking. The construction of a multi-dimensional feature analysis system is innovatively proposed, a multilayer graph neural structure is designed and improved, and an interactive attention mechanism is introduced, so that the accuracy breakthrough of personalized and healthy recommendation is realized; a particle swarm optimization algorithm of a segmented sine and cosine chaotic mapping function, a reduction factor and an improved inertia factor is innovatively adopted, and the intelligent level of the cooking control process is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of information data processing, and in particular to an intelligent cooking control system based on big data. Background Art

[0002] An intelligent cooking control system based on big data refers to an intelligent system that uses big data technology to collect, store, analyze and model multi-dimensional data such as ingredients, cooking processes, environmental parameters and user preferences, and combines the Internet of Things, sensor technology and intelligent algorithms to achieve precise monitoring, real-time regulation and adaptive optimization of the cooking process.

[0003] However, traditional cooking control systems have technical problems such as lack of personalization of cooked dishes, mismatch between the nutritional structure of dishes and the health needs of users, and extensive control of the cooking process, which makes it difficult for the cooked dishes to meet the actual conditions, personalized preferences and health goals of users; existing models for recommending healthy dishes have technical problems such as single recommendation logic and weak multi-dimensional feature fusion capabilities, which leads to serious homogeneity and insufficient accuracy of recommendation results; existing optimization algorithms for optimizing cooking control parameters have technical problems such as single initial population distribution and easy falling into local optimality in high-dimensional search space, which leads to poor optimization effect of healthy cooking control parameters. Summary of the Invention

[0004] In response to the above situation, in order to overcome the defects of the existing technology, the present invention provides an intelligent cooking control system based on big data. It addresses the technical problems in traditional cooking control systems, such as the lack of personalization of cooking dishes, the mismatch between the nutritional structure of dishes and the health needs of users, and the extensive control of the cooking process, which makes it difficult for the cooked dishes to meet the actual conditions, personalized preferences and health goals of users. This solution innovatively proposes a mechanism for personalized cooking dish recommendations, dual-layer health control, and practical cooking process. The personalized cooking dish recommendation realizes intelligent recommendation of the most suitable individual user needs among the culinary dishes by constructing a healthy dish recommendation model, thereby improving the accuracy and adaptability of the recommendation. The dual-layer health control introduces dish health adaptability analysis in the dish selection stage to ensure that the nutritional structure of the recommended dishes matches the health needs of users. In the cooking stage, control parameters are extracted from standard recipes, and health adaptability modeling and parameter optimization are performed in combination with the user's health status to achieve refined and healthy regulation of the cooking process. The practical cooking process effectively avoids the disconnection between recommendation content and actual operability by introducing a food availability judgment mechanism and extracting adjustable cooking control parameters from standard recipes, thereby improving practicality. This solution achieves the organic unity of personalization, health and practicality, significantly improving the health guidance ability, recommendation adaptation ability and operational effectiveness of the intelligent cooking system in real-world application environments. In response to the technical problems of the existing healthy dish recommendation model with a single recommendation logic and weak multi-dimensional feature fusion ability, which leads to serious homogeneity and insufficient accuracy of recommendation results, this solution innovatively proposes to build a multi-dimensional feature analysis system, design a multi-layer graph neural structure with a node propagation mechanism, and introduce an interactive attention mechanism. By building a multi-dimensional feature analysis system for dish health, user health and user dietary preferences, the integrity and personalized expression ability of the recommendation basis are improved. The multi-layer graph neural structure with a node propagation mechanism can deeply explore the potential weak correlation between dish health and user health needs, enhancing the semantic relevance and accuracy of recommendations. The introduced interactive attention mechanism can adaptively adjust the influence weights of neighboring nodes in deep propagation to ensure recommendation diversity and individual differentiation. This method significantly improves the feature modeling ability, health adaptation accuracy and robustness of the recommendation system under multi-source heterogeneous data of the healthy dish recommendation model, achieving a breakthrough in the accuracy of personalized and healthy recommendations.To address the technical issues of existing optimization algorithms for cooking control parameter optimization, such as a single initial population distribution and a tendency to fall into local optimality in high-dimensional search spaces, which in turn leads to poor optimization results for healthy cooking control parameters, this solution innovatively employs an improved particle swarm optimization algorithm based on piecewise sine-cosine chaos mapping functions, a reduction factor, and an improved inertia factor to search for the optimal cooking control parameter combination. This expands the search space coverage, balances search and convergence capabilities, and effectively improves the algorithm's ability to escape local optimality, thereby achieving a globally optimal configuration of cooking control parameters that meets healthy adaptability requirements. This significantly enhances the intelligence and control capabilities of the cooking control process in healthy cooking scenarios.

[0005] The technical solution adopted by the present invention is as follows: The intelligent cooking control system based on big data provided by the present invention includes a cooking control data acquisition module, a data optimization processing module, a healthy dish intelligent recommendation module, a cooking control parameter optimization module and an intelligent cooking execution control module;

[0006] The cooking control data acquisition module is used to collect the raw data required to realize intelligent cooking control, specifically by collecting data through the intelligent kitchen appliance platform and wearable health monitoring equipment to obtain the raw data of intelligent cooking control;

[0007] The data optimization processing module specifically cleans and standardizes the collected raw data; and based on the current status of the ingredients available in the user's kitchen, determines the availability of ingredients and obtains cooking intelligent control optimization data;

[0008] The healthy dish intelligent recommendation module is used to realize personalized healthy dish recommendations from the set of culinary dishes. Specifically, it completes the analysis of dish health adaptation features, user health status analysis and user dietary preference analysis in sequence, and adopts a multi-layer graph neural network and interactive attention mechanism to fuse multiple features. Finally, a healthy dish recommendation model is established through a fully connected neural network. Historical data is used as training data for model training, and real-time data is input into the trained model to obtain personalized healthy dish recommendations.

[0009] The cooking control parameter optimization module is used to optimize and adjust the cooking parameters of personalized recommended healthy dishes. Specifically, the module extracts a cooking control parameter combination based on the standard cooking process of the personalized recommended healthy dishes, calculates a health fitness score based on the user's health status characteristics, and uses the maximization of the health fitness score as the optimization goal. The module uses a particle swarm optimization algorithm with a piecewise sine-cosine chaos mapping function, a reduction factor, and an improved inertia factor to perform a control parameter optimization search, ultimately obtaining an optimized cooking control parameter combination with the best health fitness.

[0010] The intelligent cooking execution control module specifically replaces the original cooking control parameters of the recommended dishes through an optimized combination of cooking control parameters, generates an optimized healthy dish cooking process, generates cooking task instructions, and transmits them to the intelligent cooking device for execution.

[0011] Furthermore, the cooking control data acquisition module specifically collects the original data required for cooking control through the smart kitchen appliance platform and wearable health monitoring equipment to obtain the original data of intelligent cooking control; the original data of intelligent cooking control includes historical cooking control data and real-time cooking control data; the historical cooking control data and real-time cooking control data both include user health information collection data, user dietary preference data, dish information data and cooking ingredient information data; the historical cooking control data also includes user feedback rating data.

[0012] Furthermore, the data optimization processing module includes data cleaning, data standardization processing and food availability judgment, specifically the following steps:

[0013] Data cleaning processing, specifically filling missing values, removing outliers and normalizing fields of the original data;

[0014] Data normalization processing, specifically normalizing the numerical data in the original data through the minimum-maximum normalization method;

[0015] The ingredient availability judgment is specifically to compare the user's current ingredient information with the standard ingredient requirement list of each candidate dish in the dish information data, calculate the dish ingredient matching degree, perform preliminary screening based on the ingredient matching degree, and further judge whether the main ingredient composition conditions of the dish are met. If the main ingredient composition required by the dish is missing, it is judged as unfinishable and removed from the optional dishes to obtain a list of culinary dishes.

[0016] Furthermore, the healthy food intelligent recommendation module specifically includes the following steps:

[0017] Establishing a healthy dish recommendation model specifically includes the following steps:

[0018] The dish health adaptation feature analysis layer uses a deep convolutional neural network to extract health features from the dish information data of the dishes in the culinary list to obtain the dish health adaptation features;

[0019] The user health status analysis layer uses a long-short memory neural network to analyze the user's health status from the user's health information collection data and dietary health goal preference data to obtain the user's health status characteristics;

[0020] The user diet preference analysis layer uses a deep convolutional neural network to analyze user diet preferences from user diet preference data and obtain user diet preference features;

[0021] The dish recommendation feature integration involves using a multi-layer graph neural network and an interactive attention mechanism for feature fusion, including the following steps:

[0022] The graph structure is constructed, specifically including three node types and two edge types, and the adjacency matrix A of the graph is constructed based on the above nodes and edges. The three node types include dish nodes, user health status nodes, and user dietary preference nodes, and their node features are respectively dish health adaptation features, user health status features, and user dietary preference features; the two edge types include the edge between the dish and the user's health goal, and the edge between the dish and the user's dietary preference;

[0023] The multi-layer graph neural network information update is specifically to first calculate the interaction relationship between node features through element-by-element product operation to capture the dependency relationship between nodes, then combine the node's previous layer features and the interaction features between nodes with the weight matrix, and then adjust the propagation information through degree normalization to perform node information propagation operation. Finally, the previous layer node features and the propagation information of all neighboring nodes are weighted and integrated to update the current node features and complete the node update. The formula used is as follows:

[0024] ;

[0025] ;

[0026] Where, represents the propagation information between node i and its neighbor node j, represents the number of neighbor nodes of node i, represents the number of neighbor nodes of node j, and Represent the node features of the lth layer and the neighbor node interaction feature weight matrix, Indicates the The characteristics of node i in the layer, Indicates the The features of node j in layer, represents the element-wise product operation, represents the ReLU activation function, represents the set of neighbor nodes of node i, Indicates the Features of node i in layer;

[0027] Calculate the attention weight to dynamically adjust the influence of each neighbor node on the target node. Specifically, the attention weight between each layer node and its neighbor node features is calculated through the interactive attention mechanism. ;

[0028] Calculate the final features of each layer of nodes. Specifically, the features of neighboring nodes are weighted and aggregated by attention weights. The aggregated features are summed with the features of the previous layer of the current node, and nonlinear processing is performed through the activation function to obtain the final features of each layer of nodes.

[0029] The personalized dish recommendation output layer inputs the final node features into the fully connected neural network layer and processes them through the activation function to obtain personalized healthy dish recommendations;

[0030] Healthy dish recommendation model training, specifically using historical cooking control data as training data to train the healthy dish recommendation model, thereby obtaining a trained healthy dish recommendation model;

[0031] Personalized healthy dish recommendations are used to ensure that each recommendation can meet the user's latest health needs and personalized needs. Specifically, real-time cooking control data is used as input data for the trained healthy dish recommendation model to obtain personalized recommended healthy dishes.

[0032] Furthermore, the cooking control parameter optimization module, specifically based on personalized healthy dishes, selects relevant dishes from existing recipes, extracts their cooking control parameters, optimizes the cooking parameters using an improved optimization algorithm based on the user's health adaptation goals, and obtains an optimized combination of cooking control parameters, including the following steps:

[0033] Cooking control parameter setting, specifically, based on personalized healthy dishes, obtains the standard cooking process of relevant dishes in existing recipes, and dynamically extracts the amount information related to condiments from them as cooking control parameters, and obtains a cooking control parameter combination by combining the parameters;

[0034] Design an optimization goal, specifically, by combining the cooking control parameter combination and the user's health status characteristics to calculate a health fitness score, and maximize the health fitness score as the optimization goal;

[0035] Obtaining an optimized combination of cooking control parameters includes the following steps:

[0036] The particle search population is initialized by taking the cooking control parameter combination as the particle position vector in the particle search algorithm and using the piecewise sine-cosine chaotic mapping function to generate chaotic variables. , and then initialize the particle position vector according to each chaotic variable , thus completing the particle population initialization operation; the formula used is as follows:

[0037] ;

[0038] Where, represents the i-th chaotic variable, represents the i+1th chaotic variable, represents the modulo operator, Represents the control parameter of the chaotic map, and its value range is (0,1], represents the random perturbation term, which is a random number in the range [0,1]. represents the parameter used to control the amplitude of the sinusoidal fluctuation, Indicates the parameter that controls the amplitude of cosine fluctuation, and its value range is (0,0.5], It represents a factor that controls the particle update granularity, which is a constant;

[0039] Calculation of individual particle fitness values, specifically calculating the fitness value f of particles in the particle swarm i , the health fitness score is used as the fitness value of the individual particle;

[0040] Particle update, specifically updating particle velocity and particle position by calculating the reduction factor and improving the inertia factor;

[0041] ;

[0042] ;

[0043] ;

[0044] Where, Indicates that the i-th particle is in the Iteration speed, Indicates that the i-th particle is in the Iteration speed, represents the position of the i-th particle in the t-th iteration, represents the local optimal position of the individual particle, represents the global optimal position of the particle, and represents a random number in the range [0,1], represents the individual learning factor, which is used to control the speed at which particles move to their optimal positions. Represents the group learning factor, which is used to control the speed at which particles move to the global optimal position. t represents the current number of iterations. Indicates the Inertia factor during iteration, represents the initial weight value, Indicates the maximum number of iterations;

[0045] The optimal position of the particle is obtained by re-evaluating the fitness value of the particle after the particle update, and comparing the fitness value of the current particle with its historical individual optimal value and the current population global optimal value. If the current particle fitness is better, the corresponding individual optimal position and global optimal position are updated respectively; at the same time, the local optimal position of the individual particle in the next iteration is obtained. and the global optimal position of the particle in the next iteration ;

[0046] The particle iteration ends when the particle fitness value f i When the fitness threshold is higher than the value and the maximum number of iterations is reached, the search is terminated and the global optimal position of the particle is obtained. The global optimal position of the particle specifically refers to the optimized combination of cooking control parameters.

[0047] Furthermore, the intelligent cooking execution control module specifically replaces the original cooking control parameters of the standard cooking process of personalized recommended healthy dishes according to the optimized combination of the cooking control parameters to obtain an optimized dish cooking process. On this basis, it dynamically generates step-by-step cooking task instructions and transmits the cooking task instructions to the intelligent cooking equipment to achieve personalized, healthy and practical intelligent cooking execution.

[0048] The beneficial effects achieved by the present invention using the above scheme are as follows:

[0049] (1) Aiming at the technical problems in traditional cooking control systems, such as lack of personalization of cooking dishes, mismatch between the nutritional structure of dishes and the health needs of users, and extensive control of the cooking process, which lead to the situation that the cooked dishes are difficult to meet the actual conditions, personalized preferences and health goals of users; this solution innovatively proposes a mechanism of personalized cooking dish recommendation, two-layer health control and practical cooking process. Among them, the personalized cooking dish recommendation realizes the intelligent recommendation of the most suitable individual user needs among the culinary dishes by constructing a healthy dish recommendation model, thereby improving the accuracy and adaptability of the recommendation; the two-layer health control introduces the health adaptability of dishes in the dish selection stage. Analysis is performed to ensure that the nutritional structure of recommended dishes matches the health needs of users. Control parameters are extracted from standard recipes during the cooking stage, and health adaptability modeling and parameter optimization are performed in combination with the user's health status to achieve refined and healthy regulation of the cooking process. The practicality of the cooking process is achieved by introducing a mechanism for judging the availability of ingredients and extracting adjustable cooking control parameters from standard recipes, which effectively avoids the disconnection between recommended content and actual operability and improves practicality. This solution realizes the organic unity of personalization, health and practicality, and significantly improves the health guidance ability, recommendation adaptability and operational effectiveness of the intelligent cooking system in real application environments.

[0050] (2) In response to the technical problems of the existing recommendation logic being single and the multi-dimensional feature fusion capability being weak in the existing healthy dish recommendation model, which leads to serious homogeneity and insufficient accuracy of the recommendation results, this solution innovatively proposes to construct a multi-dimensional feature analysis system, design a multi-layer graph neural structure with a node propagation mechanism, and introduce an interactive attention mechanism. By constructing a multi-dimensional feature analysis system of dish health, user health, and user dietary preferences, the integrity and personalized expression capability of the recommendation basis are improved. The multi-layer graph neural structure with a node propagation mechanism can deeply explore the potential weak correlation between the potential dish health and user health needs, thereby enhancing the semantic relevance and accuracy of the recommendation. The introduced interactive attention mechanism can adaptively adjust the influence weights of neighboring nodes in deep propagation, thereby ensuring the diversity and individual differentiation of recommendations. This method significantly improves the feature modeling capability, health adaptation accuracy, and robustness of the recommendation system under multi-source heterogeneous data of the healthy dish recommendation model, achieving a breakthrough in the accuracy of personalized and healthy recommendations.

[0051] (3) In view of the technical problems of the existing optimization algorithms suitable for cooking control parameter optimization in high-dimensional search space, such as the single initial population distribution and the easy falling into local optimality, which leads to poor optimization effect of healthy cooking control parameters, this scheme innovatively adopts an improved particle swarm optimization algorithm based on piecewise sine-cosine chaos mapping function, reduction factor and improved inertia factor to search for the optimal cooking control parameter combination, expands the coverage of the search space, balances the search ability and convergence ability, and effectively improves the ability of the algorithm to jump out of the local optimality, thereby achieving the global optimal configuration of cooking control parameters while meeting the health adaptability requirements, and significantly improving the intelligence level and regulation ability of the cooking control process in the healthy cooking scenario. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 This is a module diagram of the intelligent cooking control system based on big data provided by the present invention;

[0053] Figure 2 This is a flowchart of the data optimization processing module;

[0054] Figure 3 This is a flowchart of the healthy food intelligent recommendation module;

[0055] Figure 4 A flowchart for establishing a healthy dish recommendation model in the healthy dish intelligent recommendation module;

[0056] Figure 5 Schematic diagram of the flow of cooking control parameter optimization module;

[0057] Figure 6 A schematic diagram of a process for obtaining an optimized combination of cooking control parameters in a cooking control parameter optimization module;

[0058] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. They are used to explain the present invention together with the embodiments of the present invention and do not constitute a limitation of the present invention. DETAILED DESCRIPTION

[0059] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments; based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0060] In the description of the present invention, it should be understood that terms such as "up", "down", "front", "back", "left", "right", "top", "bottom", "inside" and "outside" indicating directions or positional relationships are based on the directions or positional relationships shown in the accompanying drawings. They are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the system or element referred to must have a specific direction, be constructed and operated in a specific direction. Therefore, they should not be understood as limiting the present invention.

[0061] Example 1, see Figure 1 The intelligent cooking control system based on big data provided by the present invention includes a cooking control data acquisition module, a data optimization processing module, a healthy dish intelligent recommendation module, a cooking control parameter optimization module and an intelligent cooking execution control module;

[0062] The cooking control data acquisition module is used to collect the raw data required to realize intelligent cooking control. Specifically, it collects data through the intelligent kitchen appliance platform and wearable health monitoring devices to obtain the raw data of intelligent cooking control, and sends the data to the data optimization processing module;

[0063] The data optimization processing module receives the data sent by the cooking control data acquisition module, specifically by performing data cleaning and standardization on the collected raw data; combining the status of the ingredients currently available in the user's kitchen, it determines the availability of the ingredients, obtains cooking intelligent control optimization data, and sends the data to the healthy dish intelligent recommendation module;

[0064] The healthy dish intelligent recommendation module receives data sent by the data optimization processing module, specifically performs dish health adaptation feature analysis, user health status analysis, and user dietary preference analysis in sequence, and uses a multi-layer graph neural network and an interactive attention mechanism to perform multi-feature fusion. Finally, a healthy dish recommendation model is established through a fully connected neural network. Historical data is used as training data for model training, and real-time data is input into the trained model to obtain personalized healthy dish recommendations. The data is then sent to the cooking control parameter optimization module;

[0065] The cooking control parameter optimization module receives data sent by the healthy dish intelligent recommendation module, specifically extracts cooking control parameter combinations based on the standard cooking process of personalized recommended healthy dishes, calculates the health fitness score based on the user's health status characteristics, and uses the particle swarm optimization algorithm with piecewise sine-cosine chaos mapping function, reduction factor and improved inertia factor to optimize the control parameters and search for the optimal cooking control parameter combination with the best health fitness, and sends the data to the intelligent cooking execution control module;

[0066] The intelligent cooking execution control module specifically replaces the original cooking control parameters of the recommended dishes through an optimized combination of cooking control parameters, generates an optimized healthy dish cooking process, generates cooking task instructions, and transmits them to the intelligent cooking device for execution.

[0067] By performing the above operations, the technical problems of the traditional cooking control system, such as the lack of personalization of cooking dishes, the mismatch between the nutritional structure of dishes and the health needs of users, and the extensive control of the cooking process, which make it difficult for the cooked dishes to meet the actual conditions, personalized preferences and health goals of users, are addressed. This solution innovatively proposes a mechanism for personalized cooking dish recommendations, dual-layer health control and practical cooking process. The personalized cooking dish recommendation realizes the intelligent recommendation of the most suitable individual user needs among the culinary dishes by constructing a healthy dish recommendation model, thereby improving the accuracy and adaptability of the recommendation. The dual-layer health control introduces the health of dishes into the dish selection stage. Adaptability analysis ensures that the nutritional structure of recommended dishes matches the health needs of users. During the cooking stage, control parameters are extracted from standard recipes, and health adaptability modeling and parameter optimization are performed in combination with the user's health status to achieve refined and healthy regulation of the cooking process. The practicality of the cooking process effectively avoids the disconnection between recommended content and actual operability by introducing a mechanism for judging the availability of ingredients and extracting adjustable cooking control parameters from standard recipes, thereby improving practicality. This solution realizes the organic unity of personalization, health and practicality, and significantly improves the health guidance ability, recommendation adaptability and operational effectiveness of the intelligent cooking system in real application environments.

[0068] Example 2, see Figure 1 This embodiment is based on the above embodiment. The cooking control data acquisition module specifically collects the original data required for cooking control through the smart kitchen appliance platform and wearable health monitoring equipment to obtain cooking intelligent control original data; the cooking intelligent control original data includes historical cooking control data and real-time cooking control data; the historical cooking control data and real-time cooking control data both include user health information collection data, user diet preference data, dish information data and cooking ingredient information data; the historical cooking control data also includes user feedback rating data; the user health information collection data includes blood pressure, blood sugar, heart rate, weight, BMI, allergy information, and chronic disease markers; the user diet preference data includes taste tendencies, cuisine preferences, taboo foods and dietary health target preference data; the dish information data includes dish ingredient types, ingredient weight, calories, fat, protein content and dish cooking methods; the cooking ingredient information data includes ingredient types, ingredient weight, and ingredient freshness; the dietary health target preferences include salt-controlled diet, sugar-controlled diet, fat-controlled diet and fat-reducing diet.

[0069] Example 3, see Figure 1 and Figure 2 This embodiment is based on the above embodiment. The data optimization processing module is used to pre-process and structure the raw data collected by the cooking control data acquisition module. Specifically, it performs data cleaning, data standardization processing, and food availability judgment, including the following steps:

[0070] Data cleaning is used to ensure the integrity and consistency of the original data and remove invalid and erroneous data. Specifically, it involves filling missing values, removing outliers, and normalizing fields in the original data.

[0071] The missing value filling is used to achieve data integrity, specifically to numerically fill in the missing data items in the original data through the mean filling method;

[0072] The outlier removal is used to achieve data distribution rationality control, specifically by detecting and removing extreme values ​​and logical outliers in the original data through the Z-Score algorithm;

[0073] The field normalization process is used to ensure the consistency of model input. Specifically, it standardizes data of different formats through unified unit conversion rules to ensure the consistency of model input.

[0074] Data normalization is used to unify the numerical scale of numerical features. Specifically, the numerical data in the original data is normalized by the minimum-maximum normalization method;

[0075] Ingredient availability judgment is used to perform structured analysis on the ingredient information currently available in the user's kitchen and match it with dish requirements, determine whether each candidate dish meets the raw material conditions, and filter out the current set of culinary dishes accordingly; specifically, the user's current ingredient information is compared with the standard ingredient requirement list of each candidate dish in the dish information data, the dish ingredient matching degree is calculated, a preliminary screening is performed based on the ingredient matching degree, and further judgment is made as to whether the main ingredient composition conditions of the dish are met. If the main ingredient composition required for the dish is missing, it is judged as unfinishable and is eliminated from the optional dishes to obtain a list of culinary dishes; the calculation of the ingredient matching degree is specifically the actual number of matching ingredient types divided by the total number of required ingredients; the preliminary screening is specifically to retain dishes with a matching degree higher than the set threshold, and preliminarily exclude dishes below the threshold; the dish specifically refers to the dish name and the main ingredient composition of the dish.

[0076] Example 4, see Figure 1 、 Figure 3 and Figure 4 This embodiment is based on the above embodiment. The healthy dish intelligent recommendation module is used to select dishes that best suit the user's health goals from the menu based on the user's health needs, dietary preferences, and health characteristics of the dishes in the list of culinary dishes that takes into account the actual ingredients of the dishes. Specifically, the module includes the following steps:

[0077] Establishing a healthy dish recommendation model specifically includes the following steps:

[0078] The dish health adaptation feature analysis layer is used to analyze the health features of dishes to ensure that the recommended dishes meet the user's health needs. Specifically, a deep convolutional neural network is used to extract health features from the dish information data of the dishes in the culinary list to obtain the dish health adaptation features. The dish health adaptation features specifically refer to the impact of the dishes on the user's health. The formula used is as follows:

[0079] ;

[0080] Where, Indicates the health and adaptability of the dish. Represents the operation function of the deep convolutional neural network, Represents dish information data;

[0081] The user health status analysis layer is used to analyze the user's health status, determine the user's health status and the health compatibility of dishes, and ensure that the recommended dishes meet the user's health needs. Specifically, a long-short-term memory neural network is used to analyze the user's health status from the user's health information collection data and dietary health goal preference data to obtain the user's health status characteristics. The user health status characteristics represent the user's health status and reflect the current user's health needs for dishes. The formula used is as follows:

[0082] ;

[0083] Where, Indicates the user's health status characteristics, Represents the operation function of the long short-term memory neural network, Indicates user health information collection data, Indicates dietary health goal preference data;

[0084] The user diet preference analysis layer is used to analyze the user's eating habits and taste preferences. Specifically, a deep convolutional neural network is used to analyze the user's diet preference data and obtain the user's diet preference characteristics. The formula used is as follows:

[0085] ;

[0086] Where, Represents the user's dietary preference characteristics, Represents user dietary preference data;

[0087] Dish recommendation feature integration is used to integrate the health adaptation characteristics of dishes, user health needs and dietary preferences. Specifically, feature fusion is performed using a multi-layer graph neural network and an interactive attention mechanism, including the following steps:

[0088] The graph structure is constructed, specifically including three node types and two edge types, and the adjacency matrix A of the graph is constructed based on the above nodes and edges. The three node types include dish nodes, user health status nodes, and user dietary preference nodes, and their node features are respectively dish health adaptation features, user health status features, and user dietary preference features; the two edge types include the edge between the dish and the user's health goal, and the edge between the dish and the user's dietary preference;

[0089] The multi-layer graph neural network information update is specifically to first calculate the interaction relationship between node features through element-by-element product operation to capture the dependency relationship between nodes, then combine the node's previous layer features and the interaction features between nodes with the weight matrix, and then adjust the propagation information through degree normalization to perform node information propagation operation. Finally, the previous layer node features and the propagation information of all neighboring nodes are weighted and integrated to update the current node features and complete the node update. The formula used is as follows:

[0090] ;

[0091] ;

[0092] Where, represents the propagation information between node i and its neighbor node j, represents the number of neighbor nodes of node i, represents the number of neighbor nodes of node j, and Represent the node features of the lth layer and the neighbor node interaction feature weight matrix, Indicates the The characteristics of node i in the layer, Indicates the The features of node j in layer, represents the element-wise product operation, represents the ReLU activation function, represents the set of neighbor nodes of node i, Indicates the Features of node i in layer;

[0093] Calculate the attention weight to dynamically adjust the influence of each neighbor node on the target node. Specifically, the attention weight between each layer node and its neighbor node features is calculated through the interactive attention mechanism. ; The formula used is as follows:

[0094] ;

[0095] Where, represents the attention weight between node i and its neighbor node j, represents the cosine similarity function, Indicates the The features of node j in layer, Indicates the Features of node k in layer;

[0096] Calculate the final features of each layer of nodes. Specifically, the features of neighboring nodes are weighted and aggregated by attention weights. The aggregated features are summed with the features of the previous layer of the current node, and nonlinear processing is performed through the activation function to obtain the final features of each layer of nodes.

[0097] ;

[0098] Where, Indicates the The final features of the layer nodes, Indicates the Final features of layer nodes;

[0099] The personalized dish recommendation output layer inputs the final node features into the fully connected neural network layer and processes them through the activation function to obtain personalized healthy dish recommendations. The formula used is as follows:

[0100] ;

[0101] Where, and Represents the weight and bias parameters of the personalized dish recommendation output layer, Indicates personalized dish recommendations. Represents the final features of the L-th layer node, where L represents the number of layers in the multi-layer graph neural network;

[0102] Healthy dish recommendation model training, specifically using historical cooking control data as training data to train the healthy dish recommendation model, thereby obtaining a trained healthy dish recommendation model;

[0103] Personalized healthy dish recommendations are used to ensure that each recommendation can meet the user's latest health needs and personalized needs. Specifically, real-time cooking control data is used as input data for the trained healthy dish recommendation model to obtain personalized recommended healthy dishes.

[0104] By performing the above operations, in order to address the technical problems of the existing healthy dish recommendation model with single recommendation logic and weak multi-dimensional feature fusion ability, which leads to serious homogeneity and insufficient accuracy of recommendation results, this solution innovatively proposes to build a multi-dimensional feature analysis system, design a multi-layer graph neural structure with a node propagation mechanism, and introduce an interactive attention mechanism. By constructing a multi-dimensional feature analysis system for dish health, user health and user dietary preferences, the integrity and personalized expression ability of the recommendation basis are improved. The multi-layer graph neural structure with a node propagation mechanism can deeply explore the potential weak correlation between potential dish health and user health needs, and enhance the semantic relevance and accuracy of the recommendation. The introduced interactive attention mechanism can adaptively adjust the influence weights of neighbor nodes in deep propagation to ensure recommendation diversity and individual differentiation. This method significantly improves the feature modeling ability of the healthy dish recommendation model, the health adaptation accuracy and the robustness of the recommendation system under multi-source heterogeneous data, and achieves a breakthrough in the accuracy of personalized and healthy recommendations.

[0105] Example 5, see Figure 1 、 Figure 5 and Figure 6 This embodiment is based on the above embodiment. The cooking control parameter optimization module is used to optimize and adjust the cooking control parameters based on the existing recipe structure according to the health goal, ensuring that each recommended dish meets the user's health needs in terms of cooking control and retains the basic flavor of the original recipe. Specifically, based on the personalized recommended healthy dishes, relevant dishes are selected from the existing recipes, and their cooking control parameters are extracted. The cooking parameters are optimized by an improved optimization algorithm based on the user's health adaptation goal to obtain an optimized combination of cooking control parameters. The module includes the following steps:

[0106] Cooking control parameter setting involves obtaining standard cooking processes for personalized healthy dishes from existing recipes, dynamically extracting information about condiment usage from these processes, and combining these information to create a cooking control parameter combination. Condiments specifically refer to various seasonings added during the cooking process, including salt, oil, sugar, sauce, chicken essence, MSG, and chili peppers.

[0107] Design an optimization goal, specifically, by combining the cooking control parameter combination and the user's health status characteristics to calculate a health fitness score, and maximize the health fitness score as the optimization goal;

[0108] The health suitability score calculation method is specifically based on a multi-layer perceptron neural network structure, taking cooking control parameters and user health status characteristics as input data. After calculations in several hidden layers, a health suitability score is obtained, which is used to evaluate the adaptability of the cooking control parameters of the dish to the user's health. The score ranges from 0 to 100, and the higher the score, the better the adaptability of the dish to the user's health goals. The model is trained using the cooking control parameters and user health status characteristics of the historical dish.

[0109] Obtaining an optimized combination of cooking control parameters includes the following steps:

[0110] The particle search population is initialized by taking the cooking control parameter combination as the particle position vector in the particle search algorithm and using the piecewise sine-cosine chaotic mapping function to generate chaotic variables. , and then initialize the particle position vector according to each chaotic variable , thus completing the particle population initialization operation; the formula used is as follows:

[0111] ;

[0112] ;

[0113] Where, represents the i-th chaotic variable, represents the i+1th chaotic variable, represents the modulo operator, Represents the control parameter of the chaotic map, and its value range is (0,1], represents the random perturbation term, which is a random number in the range [0,1]. It represents the parameter used to control the amplitude of the sinusoidal fluctuation, and its value range is (0,0.5], Indicates the parameter that controls the amplitude of cosine fluctuation, and its value range is (0,0.5], It represents a factor that controls the particle update granularity, which is a constant; represents the initial position of the i-th individual, and Respectively represent the lower and upper limits of individual searches;

[0114] Calculation of individual particle fitness values, specifically calculating the fitness value f of particles in the particle swarm i , the health fitness score is used as the fitness value of the individual particle;

[0115] Particle update, specifically updating particle velocity and particle position by calculating the reduction factor and improving the inertia factor;

[0116] ;

[0117] ;

[0118] ;

[0119] ;

[0120] Where, Indicates that the i-th particle is in the Iteration speed, Indicates that the i-th particle is in the Iteration speed, represents the position of the i-th particle in the t-th iteration, represents the local optimal position of the individual particle, represents the global optimal position of the particle, and represents a random number in the range [0,1], represents the individual learning factor, which is used to control the speed at which particles move to their optimal positions. represents the group learning factor, which is used to control the speed at which particles move to the global optimal position. Indicates that the i-th particle is in the The position in the iteration, t represents the current number of iterations, Indicates the Inertia factor during iteration, represents the initial weight value, Indicates the maximum number of iterations;

[0121] The optimal position of the particle is obtained by re-evaluating the fitness value of the particle after the particle update, and comparing the fitness value of the current particle with its historical individual optimal value and the current population global optimal value. If the current particle fitness is better, the corresponding individual optimal position and global optimal position are updated respectively; at the same time, the local optimal position of the individual particle in the next iteration is obtained. and the global optimal position of the particle in the next iteration ;

[0122] The particle iteration ends when the particle fitness value f i When the fitness threshold is higher than the value and the maximum number of iterations is reached, the search is terminated and the global optimal position of the particle is obtained. The global optimal position of the particle specifically refers to the optimized combination of cooking control parameters.

[0123] By performing the above operations, in order to address the technical problems of the existing optimization algorithms suitable for cooking control parameter optimization in the high-dimensional search space, such as the single initial population distribution and the susceptibility to falling into local optimality, which leads to poor optimization effect of healthy cooking control parameters, this solution innovatively adopts an improved particle swarm optimization algorithm based on piecewise sine-cosine chaos mapping function, reduction factor and improved inertia factor to search for the optimal cooking control parameter combination, expands the coverage of the search space, balances the search ability and convergence ability, and effectively improves the algorithm's ability to escape from the local optimality, thereby achieving the global optimal configuration of cooking control parameters while meeting the health adaptability requirements, significantly improving the intelligence level and regulation ability of the cooking control process in the healthy cooking scenario.

[0124] Example 6, see Figure 1 This embodiment is based on the above embodiment. In order to solve the technical problems that the existing optimization algorithm suitable for cooking control parameter optimization has a single initial population distribution and is easy to fall into local optimality in the high-dimensional search space, which leads to poor optimization effect of healthy cooking control parameters, this solution innovatively adopts an improved particle swarm optimization algorithm based on piecewise sine-cosine chaos mapping function, reduction factor and improved inertia factor to search for the optimal cooking control parameter combination, expands the coverage of the search space, balances the search ability and convergence ability, and effectively improves the algorithm's ability to jump out of the local optimality, thereby achieving the global optimal configuration of cooking control parameters while meeting the health adaptability requirements, and significantly improving the intelligence level and regulation ability of the cooking control process in the healthy cooking scenario.

[0125] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus.

[0126] While the embodiments of the present invention have been shown and described, it will be apparent to those skilled in the art that various changes, modifications, substitutions, and alterations can be made to the embodiments without departing from the principles and spirit of the invention.

[0127] The present invention and its embodiments are described above. This description is not restrictive. The drawings show only one embodiment of the present invention, and the actual structure is not limited thereto. In short, if a person skilled in the art is inspired by this and, without departing from the purpose of the present invention, designs structures and embodiments similar to this technical solution without inventiveness, they shall fall within the scope of protection of the present invention.

Claims

1. Intelligent cooking control system based on big data, characterized by: It includes cooking control data acquisition module, data optimization processing module, healthy dish intelligent recommendation module, cooking control parameter optimization module and intelligent cooking execution control module; The cooking control data acquisition module specifically obtains raw data of intelligent cooking control through data acquisition operations; The data optimization processing module specifically cleans and standardizes the original data; and based on the current status of the available ingredients in the user's kitchen, determines the availability of ingredients and obtains cooking intelligent control optimization data; The healthy dish intelligent recommendation module specifically establishes a multidimensional feature analysis system through analysis of dish health adaptation features, user health status analysis, and user dietary preference analysis, and uses a multi-layer graph neural network and interactive attention mechanism to integrate dish recommendation features to establish a healthy dish recommendation model. Historical data is used as training data for model training, and real-time data is input into the trained model to obtain personalized healthy dish recommendations; The cooking control parameter optimization module is used to optimize and adjust the cooking parameters of personalized recommended healthy dishes. Specifically, based on the personalized recommended healthy dishes, a cooking control parameter combination is extracted from the standard cooking process, a health fitness score is calculated in combination with the user's health status characteristics, and the health fitness score is maximized as the optimization goal. A particle swarm optimization algorithm with a piecewise sine-cosine chaos mapping function, a reduction factor, and an improved inertia factor is used to obtain an optimized cooking control parameter combination. The intelligent cooking execution control module specifically replaces the original cooking control parameters of the recommended dishes through an optimized combination of cooking control parameters to obtain an optimized healthy dish cooking process, generates cooking task instructions, and transmits them to the intelligent cooking device for execution.

2. The intelligent cooking control system based on big data according to claim 1 is characterized by: The data optimization processing module obtains cooking intelligent control optimization data through data cleaning, standardization processing and food availability judgment, and includes the following steps: Data cleaning processing, specifically filling missing values, removing outliers and normalizing fields of the original data; Data normalization processing, specifically normalizing the numerical data in the original data through the minimum-maximum normalization method; The ingredient availability judgment is specifically to compare the user's current ingredient information with the standard ingredient requirement list of each candidate dish in the dish information data, calculate the dish ingredient matching degree, perform preliminary screening based on the ingredient matching degree, and further judge whether the main ingredient composition conditions of the dish are met. If the main ingredient composition required by the dish is missing, it is judged as unfinishable and removed from the optional dishes to obtain a list of culinary dishes.

3. The intelligent cooking control system based on big data according to claim 1 is characterized by: The healthy dish intelligent recommendation module is used to recommend personalized and healthy dishes from the list of culinary dishes, and specifically includes the following steps: Establishing a healthy dish recommendation model specifically includes the following steps: The dish health adaptation feature analysis layer uses a deep convolutional neural network to extract health features from the dish information data of the dishes in the culinary list to obtain the dish health adaptation features; The user health status analysis layer uses a long-short memory neural network to analyze the user's health status from the user's health information collection data and dietary health goal preference data to obtain the user's health status characteristics; The user diet preference analysis layer uses a deep convolutional neural network to analyze user diet preferences from user diet preference data and obtain user diet preference features; Feature integration for dish recommendations, specifically using a multi-layer graph neural architecture with a node propagation mechanism and an interactive attention mechanism for feature fusion; The personalized dish recommendation output layer inputs the final node features into the fully connected neural network layer and processes them through the activation function to obtain personalized healthy dish recommendations; Healthy dish recommendation model training, specifically using historical cooking control data as training data to train the healthy dish recommendation model, thereby obtaining a trained healthy dish recommendation model; Personalized healthy dish recommendations are used to ensure that each recommendation can meet the user's latest health needs and personalized needs. Specifically, real-time cooking control data is used as input data for the trained healthy dish recommendation model to obtain personalized recommended healthy dishes.

4. The intelligent cooking control system based on big data according to claim 1, characterized in that: The dish recommendation feature integration specifically includes the following steps: The graph structure is constructed, specifically including three node types and two edge types, and the adjacency matrix A of the graph is constructed based on the above nodes and edges. The three node types include dish nodes, user health status nodes, and user dietary preference nodes, and their node features are respectively dish health adaptation features, user health status features, and user dietary preference features; the two edge types include the edge between the dish and the user's health goal, and the edge between the dish and the user's dietary preference; The multi-layer graph neural network information update is specifically to first calculate the interaction relationship between node features through element-by-element product operation to capture the dependency relationship between nodes, then combine the node's previous layer features and the interaction features between nodes with the weight matrix, and then adjust the propagation information through degree normalization to perform node information propagation operation. Finally, the previous layer node features and the propagation information of all neighboring nodes are weighted and integrated to update the current node features and complete the node update. The formula used is as follows: ； ； Where, represents the propagation information between node i and its neighbor node j, represents the number of neighbor nodes of node i, represents the number of neighbor nodes of node j, and Represent the node features of the lth layer and the neighbor node interaction feature weight matrix, Indicates the The characteristics of node i in layer, Indicates the The features of node j in layer, represents the element-wise product operation, represents the ReLU activation function, represents the set of neighbor nodes of node i, Indicates the Features of node i in layer; Calculate the attention weight to dynamically adjust the influence of each neighbor node on the target node. Specifically, the attention weight between each layer node and its neighbor node features is calculated through the interactive attention mechanism. ; The final features of each layer of nodes are calculated by weighting and aggregating the features of neighboring nodes through attention weights, adding the aggregated features with the features of the previous layer of the current node, and performing nonlinear processing through activation functions to obtain the final features of each layer of nodes.

5. The intelligent cooking control system based on big data according to claim 1 is characterized by: The cooking control parameter optimization module is specifically based on personalized healthy dishes, selects relevant dishes from existing recipes, extracts their cooking control parameters, and optimizes the cooking parameters using an improved optimization algorithm based on the user's health adaptation goals to obtain an optimized combination of cooking control parameters, including the following steps: Cooking control parameter setting, specifically, based on personalized healthy dishes, obtains the standard cooking process of relevant dishes in existing recipes, and dynamically extracts the amount information related to condiments from them as cooking control parameters, and obtains a cooking control parameter combination by combining the parameters; Design an optimization goal, specifically, by combining the cooking control parameter combination and the user's health status characteristics to calculate a health fitness score, and maximize the health fitness score as the optimization goal; The optimal combination of cooking control parameters is obtained, specifically by optimizing the cooking control parameters through the improved particle swarm optimization algorithm, and finally obtaining the optimal combination of cooking control parameters with the best health fitness.

6. The intelligent cooking control system based on big data according to claim 1, characterized in that: The step of obtaining the optimal combination of cooking control parameters specifically includes the following steps: The particle search population is initialized by taking the cooking control parameter combination as the particle position vector in the particle search algorithm and using the piecewise sine-cosine chaotic mapping function to generate chaotic variables. , and then initialize the particle position vector according to each chaotic variable , thus completing the particle population initialization operation; the formula used is as follows: ； Where, represents the i-th chaotic variable, represents the i+1th chaotic variable, represents the modulo operator, Represents the control parameter of the chaotic map, and its value range is (0,1], represents the random perturbation term, which is a random number in the range [0,1]. represents the parameter used to control the amplitude of the sinusoidal fluctuation, It represents the parameter that controls the amplitude of cosine fluctuation, and its value range is (0,0.5], It represents a factor that controls the particle update granularity, which is a constant; Calculation of individual particle fitness values, specifically calculating the fitness value f of particles in the particle swarm i , the health fitness score is used as the fitness value of the individual particle; Particle update, specifically updating particle velocity and particle position by calculating the reduction factor and improving the inertia factor; ； ； ； Where, Indicates that the i-th particle is in the Iteration speed, Indicates that the i-th particle is in the Iteration speed, represents the position of the i-th particle in the t-th iteration, represents the local optimal position of the individual particle, represents the global optimal position of the particle, and represents a random number in the range [0,1], represents the individual learning factor, which is used to control the speed at which particles move to their optimal positions. Represents the group learning factor, which is used to control the speed at which particles move to the global optimal position. t represents the current number of iterations. Indicates the Inertia factor during iteration, represents the initial weight value, Indicates the maximum number of iterations; The optimal position of the particle is obtained by re-evaluating the fitness value of the particle after the particle update, and comparing the fitness value of the current particle with its historical individual optimal value and the current population global optimal value. If the current particle fitness is better, the corresponding individual optimal position and global optimal position are updated respectively; at the same time, the local optimal position of the individual particle in the next iteration is obtained. and the global optimal position of the particle in the next iteration ; The particle iteration ends when the particle fitness value f i When the fitness threshold is higher than the value and the maximum number of iterations is reached, the search is terminated and the global optimal position of the particle is obtained. The global optimal position of the particle specifically refers to the optimized combination of cooking control parameters.

7. The intelligent cooking control system based on big data according to claim 1 is characterized by: The intelligent cooking execution control module specifically replaces the original cooking control parameters of the standard cooking process of personalized recommended healthy dishes according to the optimized combination of the cooking control parameters to obtain an optimized dish cooking process. On this basis, it dynamically generates step-by-step cooking task instructions and transmits the cooking task instructions to the intelligent cooking equipment to achieve personalized, healthy and practical intelligent cooking execution.

8. The intelligent cooking control system based on big data according to claim 1 is characterized by: The cooking control data acquisition module specifically collects the original data required for cooking control through the smart kitchen appliance platform and wearable health monitoring equipment to obtain the original data of intelligent cooking control; the original data of intelligent cooking control includes historical cooking control data and real-time cooking control data; the historical cooking control data and real-time cooking control data both include user health information collection data, user dietary preference data, dish information data and cooking ingredient information data; the historical cooking control data also includes user feedback rating data.

Images