Intelligent Cooking Optimization System and Method Based on Machine Learning

Abstract

The present invention belongs to the technical field of intelligent kitchens, and discloses an intelligent cooking optimization system and method based on machine learning, including: a data acquisition module for acquiring user characteristic data; an ingredient determination module for analyzing the user characteristic data to determine the ingredient combinations corresponding to each user group; a data acquisition module for acquiring user diet data; a cooking formulation module for fusing the user diet data and the ingredient combinations to formulate personalized cooking plans for different user groups; a parameter control module for dynamically controlling cooking parameters during the cooking process of each dish in combination with the user diet data and the cooking plan; the present invention not only improves cooking efficiency and dish quality, but also meets the user's requirements for nutritional balance and taste diversification, thereby effectively solving the problem that traditional cooking methods cannot meet personalized needs, promoting the realization of healthy eating, and improving the dish quality and service level in a collective dining environment.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent kitchens, and more specifically, to an intelligent cooking optimization system and method based on machine learning. Background Art

[0002] With the rapid development of society and the improvement of people's living standards, catering demands in large canteens (such as school canteens, enterprise canteens, etc.) have become increasingly complex and diverse; canteens not only need to meet basic dietary needs, but also pay attention to nutritional balance, diverse flavors, and healthy eating, becoming an important place to promote collective health; in this context, canteen managers are faced with the challenge of providing personalized and nutritious meals for different groups (such as students, employees, etc.); however, traditional canteen cooking methods often struggle to meet these personalized needs; due to the scale of the canteen and the diversity of the population, a single menu and fixed cooking methods often cannot meet the taste and health requirements of each diner; especially during peak dining hours, how to quickly and efficiently provide meals suitable for different needs while ensuring food quality has become a major problem in canteen management.

[0003] A patent application with the publication number CN108897245A discloses an intelligent cooking system; it includes: after collecting, storing, and analyzing the information of ingredients or food to calculate reference cooking parameters and a reference cooking curve, the information of the ingredients to be cooked and cooking requirements are uploaded to the cloud for storage, and the reference cooking parameters and reference cooking curve are revised by combining the characteristics difference between the ingredients or food to be cooked and the stored ingredients or food to obtain the current cooking parameters and current cooking curve for this cooking; the cooking tool downloads the current cooking parameters and current cooking curve to the cooking device to start the cooking process, and after cooking is completed, the feedbacker feeds back the information, and after the analysis module compares the feedback information with the current cooking parameters and current cooking curve, the cooking parameters and cooking curve are optimized, and the optimized cooking parameters and cooking curve are used as reference data for the next cooking; it can be widely used in the working process of intelligent small household appliances to realize the diversification and private customization of the functions of intelligent small household appliances.

[0004] However, although the above technology realizes intelligent cooking, it overly relies on cloud processing and storage, resulting in network latency problems; moreover, it mainly focuses on the differential calculation of ingredient characteristics while ignoring the personalized consideration of users' health conditions and taste preferences; in addition, the cooking process control is relatively fixed, lacking the ability to flexibly control cooking parameters.

[0005] In view of this, the present invention proposes an intelligent cooking optimization system and method based on machine learning to solve the above problems. Summary of the Invention

[0006] To overcome the above-mentioned defects of the prior art and achieve the above object, the present invention provides the following technical solutions: An intelligent cooking optimization method based on machine learning, including:

[0007] Collect user characteristic data, where the user characteristic data includes the characteristic information of each user;

[0008] Analyze the user characteristic data, use a clustering algorithm to divide user groups, and use deep learning to determine the ingredient combinations corresponding to each user group;

[0009] Obtain user diet data, where the user diet data includes the historical diet records of each user;

[0010] Fuse the user diet data and the ingredient combinations, use a similarity calculation method to screen dishes, and formulate personalized cooking plans for different user groups;

[0011] Combine the user diet data and the cooking plan, and use an improved optimization algorithm to dynamically control cooking parameters during the cooking process of each dish.

[0012] Further, the characteristic information includes age, gender, body mass index, and exercise volume; the method for obtaining the body mass index is: obtain the height and weight of each user, square the height of each user to obtain the squared height; divide the weight of each user by the corresponding squared height to obtain the body mass index of each user;

[0013] The steps of dividing user groups include:

[0014] Step A101: Set different digital labels for different genders and mark them as gender labels, and replace all genders in the user characteristic data with the corresponding gender labels; regard each group of characteristic information in the user characteristic data as a sample point, and the sample points correspond to the characteristic information one by one;

[0015] Step A102: Calculate the total number of groups N and set the number of groups , ;

[0016] Step A103: Randomly select sample points as the center points, and sequentially incrementally mark each center point as , ;

[0017] Step A104: Mark the sample points that are not used as center points as division points, and sequentially incrementally mark each division point as , , is the number of groups of characteristic information in the user characteristic data;

[0018] Step A105: According to For each center point, establish the corresponding user groups, and calculate the point distance from each division point to each center point in sequence;

[0019] Step A106: Compare the point distances from the division point to each center point, and assign the division point to the user group corresponding to the center point with the minimum point distance;

[0020] Step A107: Let , and return to Step A106;

[0021] Step A108: Loop Steps A106 to A107 until the loop ends, and enter Step A109;

[0022] Step A109: Recalculate the new center point corresponding to each user group;

[0023] Step A110: Repeat Steps A105 to A109 until the new center points of each user group recalculated in Step A109 are the same as the new center points of the corresponding user groups calculated in the previous loop, then the loop ends, and obtain the user groups and the corresponding division points, and use them as the group set;

[0024] Step A111: Let , and return to Step A103;

[0025] Step A112: Loop Steps A103 to A111 until the loop ends, obtain the group sets, , and enter Step A113;

[0026] Step A113: Calculate the average distance corresponding to each group set, and use the group set with the maximum average distance as the optimal set, and obtain the user groups in the optimal set.

[0027] Furthermore, in Step A102, the method for calculating the total number of groups N is as follows: Preset 4 classification criteria, and the classification criteria correspond one by one to the data in the feature information; According to the classification criteria, perform group division on each type of data in the feature information to obtain the number of groups corresponding to each type of data in the feature information; Multiply each number of groups in sequence to obtain the total number of groups N;

[0028] In Step A105, the expression of the point distance is: ; In the formula, is the division point to the center point The distance of the point is the center point in the value of the nth dimension, is the dividing point in the value of the nth dimension, ; where different dimensions represent different data in the feature information;

[0029] In the step A109, the calculation method of the new center point of each user group includes:

[0030] ;

[0031] In the formula, is the new center point corresponding to the mth user group, is the mth user group is the mth user group in the nth dividing point, , is the mth user group in the nth dividing point corresponding values in different dimensions, is the mth user group number of dividing points, .

[0032] Furthermore, in A113, the method for calculating the average distance corresponding to each group set includes:

[0033] Calculate the distance coefficient corresponding to each sample point in each group set, add up the distance coefficients corresponding to each group set in sequence, and then divide by n to obtain the average distance corresponding to each group set; the expression of the distance coefficient is: ; in the formula, is the distance coefficient of the ith sample point, is the out-group distance of the ith sample point, is the in-group distance of the ith sample point, is the maximum value function, is the ith sample point, ; ;

[0034] The calculation method of the in-group distance corresponding to the ith sample point is: Mark the user group corresponding to the ith sample point as the current group, and mark all sample points in the current group except the ith sample point as other points; Calculate the ith sample point corresponding user group is marked as the current group, and all sample points in the current group except the ith sample point are marked as other points; Calculate the The point distances from each sample point to every other point are calculated and labeled as intra-cluster distances; the intra-cluster distances are summed up in sequence and then divided by the number of intra-cluster distances to obtain the within-group distance of the th sample point.

[0035] The method for calculating the out-group distance corresponding to the th sample point is as follows: calculate the point distances from the th sample point to each center point and label them as adjacent distances; sort each adjacent distance from largest to smallest, label the user group corresponding to the center point of the second-largest adjacent distance as the nearest group, and label all sample points within the nearest group as adjacent points; calculate the point distances from the th sample point to each adjacent point and label them as out-of-cluster distances; sum up the out-of-cluster distances in sequence and then divide by the number of out-of-cluster distances to obtain the out-group distance of the th sample point.

[0036] Furthermore, the method for determining the food ingredient combination includes:

[0037] Set different digital labels for different user groups and label them as group labels; use the group labels corresponding to each user group as analysis data, input the analysis data into a trained food ingredient analysis model, and predict the corresponding combination labels; the combination labels are the digital labels corresponding to the food ingredient combinations, and the digital labels corresponding to different food ingredient combinations are all different; obtain the corresponding food ingredient combinations according to the predicted combination labels.

[0038] The training process of the food ingredient analysis model includes:

[0039] Pre-collect groups of analysis data, set corresponding combination labels for groups of analysis data, where is an integer greater than 1, convert the analysis data and the corresponding combination labels into a corresponding set of feature vectors; use each set of feature vectors as the input of the food ingredient analysis model, the food ingredient analysis model takes a set of predicted combination labels corresponding to each set of analysis data as the output, takes the actual combination label corresponding to each set of analysis data as the prediction target, and the actual combination label is the pre-set combination label corresponding to the analysis data; use minimizing the sum of prediction errors of all analysis data as the training target; train the food ingredient analysis model until the sum of prediction errors converges and then stop training; the food ingredient analysis model is a deep neural network model.

[0040] Furthermore, the historical diet records include the dishes consumed by the user and the dishes the user is taboo to eat; the dishes consumed by the user are the dishes selected by the user during historical dining, and the dishes the user is taboo to eat are the dishes that the user cannot eat;

[0041] The method for formulating a cooking plan includes:

[0042] Based on the user's diet data, obtain the number of times each user selects each dish; calculate the similarity between every two dishes according to the number of selections; take the dishes with a selection count of 0 and that are not the user's taboo dishes as the unattempted dishes for the corresponding user, and construct a corresponding set of similar dishes for each unattempted dish; preset a similarity threshold, compare the similarity of each unattempted dish with the similarity threshold, and take the dishes corresponding to the similarity values greater than or equal to the similarity threshold as the similar dishes of the corresponding unattempted dish and add them to the set of similar dishes of the corresponding unattempted dish; the dishes corresponding to the similarity values less than or equal to the similarity threshold are not used as the similar dishes of the corresponding unattempted dish;

[0043] According to the similarity and the set of similar dishes, calculate the predicted number of times for each unattempted dish; add up all the selection counts and predicted numbers of times corresponding to each dish in turn to obtain the total number of times for each dish; sort all the dishes in descending order according to the total number of times to obtain a dish ranking list; obtain the ingredients corresponding to each dish and compare the ingredients corresponding to each dish with the ingredient combinations; if the ingredient combination does not contain the ingredients corresponding to the dish, delete the corresponding dish from the dish ranking list; if the ingredient combination contains the ingredients corresponding to the dish, retain the corresponding dish in the dish ranking list; obtain the number of users, where the number of users is equal to the number of groups of feature information in the user feature data; preset a variety coefficient, multiply the number of users by the variety coefficient to obtain the number of dish varieties; select the dishes in the dish ranking list in ascending order according to the number of dish varieties as the cooking plan.

[0044] Furthermore, the expression for similarity is: ; in the formula, is the similarity between the th dish and the th dish, is the number of times the th user selects the th dish, is the number of times the th user selects the th dish, , is the number of users, , , , , is the number of dishes;

[0045] The expression for the predicted number of times is: ; in the formula, is the predicted number of times, is the unattempted dish and the The similarity of similar dishes, the th similar dish is the th similar dish in the set of similar dishes corresponding to the unattempted dish, is the number of times the user corresponding to the unattempted dish selects the th similar dish, , is the number of similar dishes in the set of similar dishes corresponding to the unattempted dish.

[0046] Furthermore, the cooking parameters include static parameters and dynamic parameters. The static parameters include the amount of ingredients used, the amount of seasonings used, the cooking time, and the cooking method. The dynamic parameters include the size of the firepower;

[0047] The method for dynamically controlling cooking parameters includes:

[0048] Construct M sets of parameter sets, set sequentially increasing numerical labels for the M sets of parameter sets, and mark them as set labels. The range of the set labels is ; Set the initial aperture center and the initial aperture radius , and set the number of iterations ; Among them, , ; Define a loop process. The loop process is: generate m candidate solutions within the aperture range, and reduce the aperture radius , , and the candidate solutions correspond one-to-one with the set labels; calculate the cooking effect corresponding to each candidate solution, and move the aperture center to the candidate solution with the maximum cooking effect; repeat the loop process, and each time the loop process is repeated, the number of iterations +1; Preset an iteration threshold G and a radius threshold H; until or , stop repeating the loop process, obtain the candidate solution corresponding to the aperture center, and mark it as the optimal solution; according to the parameter set corresponding to the set label corresponding to the optimal solution, and mark it as the optimal set;

[0049] Obtain the cooking time in the optimal set, and mark it as the optimal time; preset a step size set, and the step size set includes dishes and the time step sizes corresponding to the dishes; according to the current dish, obtain the corresponding time step size from the step size set, and mark it as the current step size; divide the optimal time by the current step size to obtain the number of time moments ; Input the optimal set into the trained firepower prediction model to predict the firepower magnitude at the first future moment, which is marked as the first predicted firepower. The first future moment is the next moment after the current moment; Replace the initial firepower magnitude in the optimal set with the first predicted firepower and re-enter it into the trained firepower prediction model to predict the firepower magnitude at the second future moment. The second future moment is the next moment after the first future moment; Perform cyclic prediction on the firepower magnitude until the firepower prediction model predicts the firepower magnitude at the firepower magnitude at the future moment, which is marked as the

[0050] predicted firepower; Use all the predicted firepower magnitudes as the firepower set, and optimize and control the cooking parameters based on the optimal set and the firepower set; The training process of the firepower prediction model is the same as that of the ingredient analysis model, and both are deep neural network models.

[0051] The calculation method of the cooking effect is as follows: Set different digital labels for different dishes and mark them as dish labels; Take the dish being cooked as the current dish, obtain the dish label corresponding to the current dish, and mark it as the current label; Obtain the parameter set corresponding to the set label of the candidate solution, use the parameter set and the current label as evaluation parameters, and input the evaluation parameters into the trained effect evaluation model to evaluate the corresponding cooking effect; The training process of the effect evaluation model is the same as that of the ingredient analysis model, and both are deep neural network models;

[0052] The expression of the candidate solution is: ; In the formula, is the th candidate solution, is the center of the aperture corresponding to the t-th repeated loop process, is the aperture radius corresponding to the t-th repeated loop process, is a random coefficient, , ;

[0053] The expression for reducing the aperture radius is: ; In the formula, is the reduced aperture radius, is the contraction coefficient, .

[0054] An intelligent cooking optimization system based on machine learning, for implementing the intelligent cooking optimization method based on machine learning, includes:

[0055] A data acquisition module, for acquiring user characteristic data, where the user characteristic data includes the characteristic information of each user;

[0056] An ingredient determination module, for analyzing the user characteristic data, dividing user groups by using a clustering algorithm, and determining the ingredient combination corresponding to each user group by using deep learning;

[0057] A data acquisition module, for acquiring user diet data, where the user diet data includes the historical diet records of each user;

[0058] A cooking formulation module, for integrating the user diet data and the ingredient combination, screening dishes by using a similarity calculation method, and formulating personalized cooking plans for different user groups;

[0059] A parameter control module, for combining the user diet data and the cooking plan, and dynamically controlling cooking parameters during the cooking process of each dish by using an improved optimization algorithm.

[0060] The technical effects and advantages of the intelligent cooking optimization system and method based on machine learning of the present invention:

[0061] By collecting user characteristic data and performing accurate analysis, it is possible to divide different users into groups, and automatically determine the ingredient combinations that meet the needs of different user groups; at the same time, by mining user diet data, it is possible to formulate personalized cooking plans for different user groups; in addition, an optimization algorithm is used to dynamically control the cooking process to achieve optimized control of cooking parameters; it not only improves cooking efficiency and dish quality, but also meets the user's requirements for nutritional balance and taste diversification, thus effectively solving the problem that traditional cooking methods cannot meet personalized needs, promoting the realization of healthy eating, and improving the dish quality and service level in the collective dining environment. Brief Description of the Drawings

[0062] Figure 1 It is a schematic diagram of the intelligent cooking optimization system based on machine learning according to Embodiment 1 of the present invention;

[0063] Figure 2 It is a flowchart of the user group division method according to Embodiment 1 of the present invention;

[0064] Figure 3 It is a flowchart of the intelligent cooking optimization method based on machine learning according to Embodiment 2 of the present invention. Detailed Embodiments

[0065] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0066] Embodiment 1

[0067] Please refer to Figure 1 As shown, the intelligent cooking optimization system based on machine learning in this embodiment includes a data collection module, an ingredient determination module, a data acquisition module, a cooking formulation module, and a parameter control module; each module is connected by wired and / or wireless means to achieve data transmission between modules.

[0068] The data collection module is used to collect user characteristic data.

[0069] The user characteristic data includes the characteristic information of each user. The users are diners in the cafeteria (such as students dining in the school cafeteria, employees dining in the enterprise cafeteria, etc.); the characteristic information includes age, gender, body mass index, and exercise amount. The exercise amount is the calorie consumption on the current day; the age and gender in the characteristic information are both obtained through the relevant systems within the organization; for example: the age and gender of each student can be obtained through the student information management system within the corresponding school, and the age and gender of each employee can be obtained through the human resources management system within the corresponding enterprise; the exercise amount in the characteristic information is obtained through wearable devices (such as smart watches, smart bracelets, etc.) worn by the users; the method for obtaining the body mass index is: obtain the height and weight of each user, square the height of each user to obtain the square of the height; divide the weight of each user by the corresponding square of the height to obtain the body mass index corresponding to each user; among them, the height and weight of each user are also obtained through the relevant systems within the organization.

[0070] It should be noted that the requirements for ingredient combinations vary among different users; among them, users of different ages have significant differences in nutritional needs; for children and adolescents, more calcium, protein, and vitamins are needed to support growth and development; for adults, a balanced diet is concerned, and ingredients with high fiber and low fat are needed; for the elderly, the intake of calcium, vitamin D, and B12 needs to be increased to maintain bone and nerve health; gender differences also affect nutritional needs and ingredient choices; generally, men need more protein and energy; for women, the intake of iron and calcium needs to be concerned; the body mass index is used to evaluate the health status of users; the larger the body mass index, the more overweight or obese the user is, so low-fat and low-calorie ingredients are needed; the smaller the body mass index, the more underweight the user is, and high-energy and high-nutrient-density ingredients need to be increased; the amount of exercise directly affects energy and nutritional needs; for high-intensity exercise, more carbohydrates and protein are needed to supplement energy and promote muscle recovery; for low-intensity exercise, the intake of high-energy ingredients needs to be reduced, and the proportion of vegetables and fruits needs to be increased to control calorie intake.

[0071] An ingredient determination module for analyzing user characteristic data, dividing user groups using a clustering algorithm, and determining the ingredient combination corresponding to each user group using deep learning.

[0072] As Figure 2 shown, the steps of dividing user groups include:

[0073] Step A101: Set different digital tags for different genders and mark them as gender tags, and replace the genders in the user characteristic data with the corresponding gender tags; regard each group of characteristic information in the user characteristic data as a sample point, and the sample point corresponds to the characteristic information one by one;

[0074] Step A102: Calculate the total number of groups N and set the number of groups , ;

[0075] Step A103: Randomly select sample points as the center points, and sequentially incrementally mark each center point as , ; that is, mark the first center point as , mark the second center point as , and mark the th center point as ;

[0076] Step A104: Mark the sample points that are not used as center points as division points, and sequentially incrementally mark each division point as , , is the number of groups of feature information in the user feature data; that is, mark the first division point as , mark the second division point as , and mark the th center point as ;

[0077] Step A105: Establish corresponding user groups based on the center points, and calculate the point distances from each division point to each center point in sequence;

[0078] Step A106: Compare the point distances from the division point to each center point, and assign the division point to the user group corresponding to the center point with the minimum point distance;

[0079] Step A107: Let , and return to Step A106;

[0080] Step A108: Loop Steps A106 to A107 until when the loop ends, and enter Step A109;

[0081] Step A109: Recalculate the new center points corresponding to each user group;

[0082] Step A110: Repeat Steps A105 to A109 until the new center points of each user group recalculated in Step A109 are the same as the new center points of the corresponding user groups calculated in the previous loop, then the loop ends, and obtain user groups and the corresponding division points, and use them as the group set;

[0083] Step A111: Let , and return to Step A103;

[0084] Step A112: Loop Steps A103 to A111 until when the loop ends, obtain group sets, , and enter Step A113;

[0085] Step A113: Calculate the average distance corresponding to each group set, and use the group set with the largest average distance as the optimal set, and obtain the user groups in the optimal set.

[0086] In the above step A102, the method for calculating the total population N is as follows: Four classification criteria are preset, and the classification criteria correspond one by one to the data in the feature information. The classification criteria are preset by those skilled in the art according to the actual situation. According to the classification criteria, each type of data in the feature information is divided into groups, and the population quantity corresponding to each type of data in the feature information is obtained. The population quantities of each group are multiplied in sequence to obtain the total population N. Exemplarily, the classification criteria corresponding to age are that those aged 0 - 12 are children, those aged 13 - 18 are teenagers, those aged 19 - 60 are adults, and those aged over 60 are the elderly. Therefore, according to the classification criteria, the population quantity obtained by dividing age into groups is 4.

[0087] In the above step A105, the expression of the point distance is: ; in the formula, is the dividing point to the center point of the point distance, is the value of the th dimension in the center point , is the dividing point in the th dimension value, ; among them, different dimensions represent different data in the feature information. For example, age is one dimension, and the amount of exercise is one dimension.

[0088] In the above step A109, the method for calculating the new center point of each user group includes:

[0089] ;

[0090] In the formula, is the new center point corresponding to the th user group, is the th dividing point in the th user group, , is the value corresponding to the th dividing point in the th dimension of the th user group, is the number of dividing points in the th user group.

[0091] In step A113, the method for calculating the average distance corresponding to each group set includes:

[0092] Calculate the distance coefficient corresponding to each sample point in each group set, add up each distance coefficient corresponding to each group set in sequence, and then divide by n to obtain the average distance corresponding to each group set; the expression of the distance coefficient is: ; In the formula, is the distance coefficient of the th sample point, is the out-group distance of the th sample point, is the in-group distance of the th sample point, is the maximum value function, ;

[0093] The calculation method of the in-group distance corresponding to the th sample point is: Mark the user group corresponding to the th sample point as the current group, and mark all sample points in the current group except the th sample point as other points; Calculate the point distance from the th sample point to each other point, and mark it as the intra-cluster distance; Add up each intra-cluster distance in sequence, and then divide by the number of intra-cluster distances to obtain the in-group distance of the th sample point;

[0094] The calculation method of the out-group distance corresponding to the th sample point is: Calculate the point distance from the th sample point to each center point, and mark it as the adjacent distance; Sort each adjacent distance from large to small, mark the user group corresponding to the center point of the second adjacent distance as the nearest group, and mark all sample points in the nearest group as adjacent points; Calculate the point distance from the th sample point to each adjacent point, and mark it as the out-of-cluster distance; Add up each out-of-cluster distance in sequence, and then divide by the number of out-of-cluster distances to obtain the out-group distance of the th sample point.

[0095] The method for determining the food ingredient combination includes:

[0096] Set different digital tags for different user groups and mark them as group tags; use the group tags corresponding to each user group as analysis data, and input the analysis data into a trained food ingredient analysis model to predict the corresponding combined tags; the combined tags are the digital tags corresponding to food ingredient combinations, and the digital tags corresponding to different food ingredient combinations are all different; according to the predicted combined tags, obtain the corresponding food ingredient combinations; the construction method of the food ingredient combinations is as follows: those skilled in the art obtain various different types of food ingredients through food ingredient databases and websites (such as NutritionData, MyFitnessPal, Food Ingredient Network, etc.), and randomly combine all the food ingredients to construct multiple different food ingredient combinations, where each food ingredient combination includes at least one food ingredient.

[0097] The training process of the food ingredient analysis model includes:

[0098] Pre-collect a set of analysis data, for each set of analysis data, set the corresponding combined tag, where \(n\) is an integer greater than 1, convert the analysis data and the corresponding combined tag into a corresponding set of feature vectors; the combined tag corresponding to the analysis data is collected by those skilled in the art during the determination process of historical food ingredient combinations. a set of analysis data, and in combination with practical experience, analyze each set of analysis data in turn to determine the combined tag corresponding to each set of analysis data, and set the corresponding combined tags for each set of analysis data in turn;

[0099] Use each set of feature vectors as the input of the food ingredient analysis model, and the food ingredient analysis model takes a set of predicted combined tags corresponding to each set of analysis data as the output, and takes the actual combined tag corresponding to each set of analysis data as the prediction target. The actual combined tag is the combined tag preset corresponding to the analysis data; take minimizing the sum of the prediction errors of all analysis data as the training target; among them, the calculation formula of the prediction error is where \(e_i\) is the prediction error, \(i\) is the group number of the feature vector corresponding to the analysis data, \(\hat{y}_i\) is the predicted combined tag corresponding to the \(i\)-th set of analysis data, \(y_i\) is the actual combined tag corresponding to the \(i\)-th set of analysis data; train the food ingredient analysis model until the sum of the prediction errors reaches convergence and then stop training.

[0100] The above-mentioned food ingredient analysis model is specifically a deep neural network model, which includes an input layer, a hidden layer, and an output layer. Each hidden layer contains multiple neurons, and there are connections between each neuron and the neurons in the next layer. These connections contain weights that determine the importance and influence of data transmission in the neural network. An activation function is applied to each neuron between the hidden layer and the output layer. The activation function introduces non-linearity, allowing the network to learn more complex patterns and features.

[0101] A data acquisition module for acquiring user diet data.

[0102] User diet data includes the historical diet records of each user. The historical diet records include the dishes consumed by the user and the dishes the user is taboo from eating. Dishes consumed by the user are the dishes selected by the user during historical dining. Dishes the user is taboo from eating are the dishes the user cannot consume, such as dishes the user is allergic to. User diet data is obtained through the consumption record form in the cafeteria consumption management system. It should be understood that the purpose of collecting user diet data is to effectively identify the user's diet preferences and habits by analyzing the user's consumption behavior and diet choices. It can not only help to deeply understand the user's taste preferences, consumption frequency, and dish selection, but also reveal the diet trends in different time periods and seasons. Therefore, the cafeteria can formulate cooking plans more accurately, further improving the service quality and user experience.

[0103] A cooking plan formulation module for integrating user diet data and ingredient combinations, screening dishes using a similarity calculation method, and formulating personalized cooking plans for different user groups.

[0104] The method for formulating a cooking plan includes:

[0105] Based on the user diet data, obtain the number of times each user selects each dish. According to the number of selections, calculate the similarity between every two dishes. Consider the dishes with a selection count of 0 and that are not dishes the user is taboo from eating as unattempted dishes for the corresponding user, and construct a corresponding set of similar dishes for each unattempted dish. Preset a similarity threshold, which is pre-set by those skilled in the art according to the actual situation. Compare the similarity of each unattempted dish with the similarity threshold. The dishes corresponding to the similarity values greater than or equal to the similarity threshold are used as the similar dishes of the corresponding unattempted dish and added to the set of similar dishes of the corresponding unattempted dish. The dishes corresponding to the similarity values less than or equal to the similarity threshold are not used as the similar dishes of the corresponding unattempted dish.

[0106] Calculate the predicted number of times for each untried dish based on the similarity and the set of similar dishes; add up all the selected times and predicted times corresponding to each dish in sequence to obtain the total number of times for each dish; sort all the dishes in descending order according to the total number of times to obtain a dish sorting list; those skilled in the art obtain the ingredients corresponding to each dish by referring to recipe-related literature or websites, and compare the ingredients corresponding to each dish with the ingredient combinations; if the ingredient combination does not contain the ingredients corresponding to the dish, delete the corresponding dish from the dish sorting list; if the ingredient combination contains the ingredients corresponding to the dish, retain the corresponding dish in the dish sorting list; obtain the number of users, where the number of users is equal to the number of groups of feature information in the user feature data; preset a species coefficient, which is preset by those skilled in the art according to the actual situation; multiply the number of users by the species coefficient to obtain the number of dish species; select the dishes in the dish sorting list in ascending order according to the number of dish species and use them as the cooking plan.

[0107] The expression for similarity is: ; where is the similarity between the th dish and the th dish, is the number of times the th user selects the th dish, is the number of times the th user selects the th dish, , is the number of users, , , , , is the number of dishes, and the number of dishes is obtained from the user's diet data;

[0108] The expression for the predicted number of times is: ; where is the predicted number of times, is the similarity between the untried dish and the th similar dish, and the th similar dish is the th similar dish in the set of similar dishes corresponding to the untried dish, is the number of times the user corresponding to the untried dish selects the th similar dish, , is the number of similar dishes in the set of similar dishes corresponding to the untried dish.

[0109] A parameter control module, which is used to combine user diet data and cooking recipes, and adopt an improved optimization algorithm to dynamically control cooking parameters during the cooking process of each dish.

[0110] The dish to be cooked is regarded as the current dish; the cooking parameters include static parameters and dynamic parameters. The static parameters include the amount of ingredients used, the amount of seasonings used, the cooking time, and the cooking method. The dynamic parameters include the size of the cooking fire. The amount of ingredients used includes the weights of various ingredients used when cooking the current dish. The amount of seasonings used includes the weights of various seasonings used when cooking the current dish, such as salt, sugar, oil, etc. The cooking time is the time elapsed during the cooking process of the current dish. The cooking method is, for example, stir-frying, boiling, steaming, etc. The size of the cooking fire is the intensity of heat energy generated by the cooking equipment when cooking the current dish, and the cooking equipment is, for example, a stove, an oven, a steam pot, etc.

[0111] The method for dynamically controlling cooking parameters includes:

[0112] Construct M sets of parameter sets, set sequentially increasing digital labels for the M sets of parameter sets, and mark them as set labels. The range of the set labels is ; Set the initial aperture center and the initial aperture radius , and set the number of iterations ; Among them, , ; Define a loop process. The loop process is: generate m candidate solutions within the aperture range, and reduce the aperture radius , , and the candidate solutions correspond one-to-one with the set labels; calculate the cooking effect corresponding to each candidate solution, and move the aperture center to the candidate solution with the largest cooking effect; repeat the loop process, and each time the loop process is repeated, the number of iterations +1; Preset an iteration threshold G and a radius threshold H. The iteration threshold G and the radius threshold H are preset by those skilled in the art according to the actual situation; until or , stop repeating the loop process, obtain the candidate solution corresponding to the aperture center, and mark it as the optimal solution; according to the parameter set corresponding to the set label corresponding to the optimal solution, and mark it as the optimal set.

[0113] The method for constructing M sets of parameter sets is as follows: Those skilled in the art preset a parameter range according to the technical parameters of the cooking device and the ingredients corresponding to the dish. The parameter range includes the range corresponding to each parameter in the research parameters. The research parameters include the amount of ingredients, the amount of seasonings, the cooking time, the method label, and the initial fire power. Randomly select a value from each range within the parameter range and construct a set of parameters. A total of M sets of parameter sets are constructed, and the M sets of parameter sets are all different. Among them, the method label is the digital label corresponding to the cooking method, and the method labels corresponding to different cooking methods are all different.

[0114] The calculation method of the cooking effect is as follows: Set different digital labels for different dishes and mark them as dish labels. Obtain the dish label corresponding to the current dish and mark it as the current label. Obtain the parameter set corresponding to the set label of the candidate solution. Use the parameter set and the current label as evaluation parameters, and input the evaluation parameters into the trained effect evaluation model to evaluate the corresponding cooking effect. The training process of the effect evaluation model is the same as that of the ingredient analysis model, and both are deep neural network models. The cooking effect is the comprehensive evaluation of the taste and appearance of the current dish cooked under the condition of the parameter set by the user. To evaluate the cooking effect corresponding to the evaluation parameters, those skilled in the art collect multiple groups of evaluation data during the cooking process of historical dishes, cook the dishes under the conditions of each group of evaluation data, let multiple users taste the cooked dishes, and give the corresponding comprehensive evaluation. Take the average value of the comprehensive evaluations of each user as the cooking effect corresponding to the evaluation data.

[0115] The expression of the candidate solution is: ; In the formula, is the th candidate solution, is the center of the aperture corresponding to the t-th repeated loop process, is the aperture radius corresponding to the t-th repeated loop process, is a random coefficient, , .

[0116] The expression for reducing the aperture radius is: ; In the formula, is the reduced aperture radius, is the contraction coefficient, .

[0117] Obtain the cooking time in the best set and mark it as the best time. Preset a set of step sizes, which includes the dishes and the corresponding time steps for the dishes. The set of step sizes is preset by those skilled in the art according to the actual situation. According to the current dish, obtain the corresponding time step from the set of step sizes and mark it as the current step. Divide the best time by the current step to obtain the number of moments ; Input the optimal set into the trained firepower prediction model to predict the firepower at the first future moment, which is marked as the first predicted firepower, and the first future moment is the next moment of the current moment; Replace the initial firepower in the optimal set with the first predicted firepower and re-enter it into the trained firepower prediction model to predict the firepower at the second future moment, and the second future moment is the next moment of the first future moment; Perform cyclic prediction on the firepower until the firepower prediction model predicts the firepower at the th future moment, which is marked as the th predicted firepower; Use all the predicted firepower values as the firepower set, and optimize and control the cooking parameters according to the optimal set and the firepower set; The training process of the firepower prediction model is the same as that of the ingredient analysis model, and both are deep neural network models.

[0118] In this embodiment, by collecting user feature data and performing precise analysis, different users can be grouped, and the ingredient combinations that meet the needs of different user groups can be automatically determined; At the same time, by mining user diet data, personalized cooking plans can be developed for different user groups; In addition, an optimization algorithm is used to dynamically control the cooking process to achieve optimized control of cooking parameters; It not only improves cooking efficiency and dish quality, but also meets users' requirements for nutritional balance and taste diversification, thus effectively solving the problem that traditional cooking methods cannot meet personalized needs, promoting the realization of healthy eating, and improving the dish quality and service level in the collective dining environment.

[0119] Embodiment 2

[0120] Please refer to Figure 3 as shown. For the parts not described in detail in this embodiment, refer to the description in Embodiment 1. Provide an intelligent cooking optimization method based on machine learning, and the method includes:

[0121] Collect user feature data;

[0122] Analyze the user feature data, use a clustering algorithm to divide user groups, and use deep learning to determine the ingredient combinations corresponding to each user group;

[0123] Obtain user diet data;

[0124] Fuse the user diet data and the ingredient combinations, use a similarity calculation method to screen dishes, and develop personalized cooking plans for different user groups;

[0125] Combine the user diet data and the cooking plan, and use an improved optimization algorithm to dynamically control the cooking parameters during the cooking process of each dish.

[0126] Embodiment 3

[0127] The present application also provides an electronic device. The electronic device may include one or more processors and one or more memories. Among them, computer-readable code is stored in the memory, and when the computer-readable code is run by one or more processors, the above-mentioned intelligent cooking optimization method based on machine learning can be executed.

[0128] The method or system according to an embodiment of the present application can also be implemented by means of the architecture of the electronic device shown in the present application. The electronic device may include a bus, one or more CPUs, a ROM, a RAM, a communication port connected to a network, an input / output, a hard disk, etc. The storage device in the electronic device, such as a ROM or a hard disk, can store the intelligent cooking optimization method based on machine learning provided by the present application. Further, the electronic device may further include a user interface. Of course, the architecture shown in the present application is only exemplary, and when implementing different devices, one or more components shown in the electronic device of the present application can be omitted according to actual needs.

[0129] Embodiment 4

[0130] An embodiment of the present application discloses a computer-readable storage medium. Computer-readable instructions are stored on the computer-readable storage medium. When the computer-readable instructions are run by a processor, the intelligent cooking optimization method based on machine learning according to an embodiment of the present application described with reference to the above drawings can be executed. The storage medium includes, but is not limited to, for example, volatile memory and / or non-volatile memory. Volatile memory may include, for example, random access memory (RAM) and cache memory, etc. Non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc.

[0131] In addition, according to an embodiment of the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the present application provides a non-transitory machine-readable storage medium storing machine-readable instructions that can be run by a processor to execute instructions corresponding to the method steps provided by the present application, such as: an intelligent cooking optimization method based on machine learning. When the computer program is executed by a central processing unit (CPU), the above-mentioned functions defined in the method of the present application are executed.

[0132] As mentioned above, the above are only specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, and all should be covered by the protection scope of the present invention. Therefore, the protection scope of the present invention should be subject to the protection scope of the claims.

[0133] Finally, the above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention shall be included within the protection scope of the present invention.

Claims

1. An intelligent cooking optimization method based on machine learning, characterized in that, Including: Collect user characteristic data, where the user characteristic data includes the characteristic information of each user; Analyze the user characteristic data, use a clustering algorithm to divide user groups, and use deep learning to determine the ingredient combinations corresponding to each user group; Obtain user diet data, where the user diet data includes the historical diet records of each user; the historical diet records include the dishes consumed by the user and the dishes the user prohibits; the dishes consumed by the user are the dishes selected by the user during historical dining, and the dishes the user prohibits are the dishes that the user cannot eat; Fuse the user diet data and the ingredient combinations, use a similarity calculation method to screen dishes, and formulate personalized cooking plans for different user groups; The method for formulating a personalized cooking plan includes: According to the user diet data, obtain the number of times each user selects each dish; according to the number of selections, calculate the similarity between every two dishes; regard the dishes with a selection number of 0 and that are not dishes the user prohibits as the unattempted dishes for the corresponding user, and construct a corresponding set of similar dishes for each unattempted dish according to the comparison result between the similarity corresponding to each unattempted dish and a preset similarity threshold; According to the similarity and the set of similar dishes, calculate the predicted number of times corresponding to each unattempted dish; add up all the selection numbers and predicted numbers corresponding to each dish in sequence to obtain the total number of times corresponding to each dish; sort all the dishes in descending order according to the total number of times corresponding to them to obtain a dish ranking list; obtain the ingredients corresponding to each dish, and compare the ingredients corresponding to each dish with the ingredient combinations; if the ingredient combinations do not contain the ingredients corresponding to the dish, delete the corresponding dish from the dish ranking list; if the ingredient combinations contain the ingredients corresponding to the dish, retain the corresponding dish in the dish ranking list; obtain the number of users, where the number of users is equal to the number of groups of characteristic information in the user characteristic data; preset a species coefficient, multiply the number of users by the species coefficient to obtain the number of dish species; select the dishes in the dish ranking list in ascending order according to the number of dish species and use them as the cooking plan; The expression for the number of predictions is as follows: In the formula, yc is the number of predictions, sim(u, wc) is the similarity between the untried dish and the u-th similar dish, and the u-th similar dish is the u-th similar dish in the set of similar dishes corresponding to the untried dish. xc u is the number of times the user corresponding to the untried dish has selected the u-th similar dish, where u ∈ [1, U], and U is the number of similar dishes in the set of similar dishes corresponding to the untried dish; Combining the user diet data and the cooking plan, use an improved optimization algorithm to dynamically control cooking parameters during the cooking process of each dish; The cooking parameters include static parameters and dynamic parameters. The static parameters include the amount of ingredients used, the amount of seasonings used, the cooking time, and the cooking method. The dynamic parameter includes the size of the firepower; construct M sets of parameter sets according to the cooking parameters, and determine the best set according to the maximum cooking effect from the M sets of parameter sets; input the best set into the trained firepower prediction model to perform cyclic prediction on the size of the firepower, and regard all the predicted firepower sizes as the firepower set; optimize and control the cooking parameters according to the best set and the firepower set.

2. The intelligent cooking optimization method based on machine learning according to claim 1, characterized in that, The characteristic information includes age, gender, body mass index, and exercise amount; the method for obtaining the body mass index is: obtain the height and weight of each user, square the height of each user to obtain the square of the height; Divide the weight of each user by the corresponding square of the height to obtain the body mass index of each user; The steps for dividing user groups include: Step A101: Set different numerical tags for different genders, mark them as gender tags, and replace all genders in the user feature data with the corresponding gender tags; regard each group of feature information in the user feature data as a sample point, and the sample points correspond to the feature information one by one; Step A102: Calculate the total number of the population N, set the number of populations a = 1, a ∈ [1, N]; Step A103: Randomly select a sample points as the center points, and sequentially and incrementally label each center point as ω b , b ∈ [1, a]; Step A104: Mark the sample points that are not the center points as division points, and sequentially and incrementally mark each division point as ψ c , c ∈ [1, n - a], where n is the number of groups of feature information in the user feature data; Step A105: Establish a corresponding a user populations based on the a center points, and calculate the point distance from each dividing point to each center point in turn; Step A106: Compare the point distances from the division point ψ c to each center point, and assign the division point ψ c to the user group corresponding to the center point with the smallest point distance; Step A107: Let c = c + 1, and return to Step A106; Step A108: Loop Steps A106 to A107 until the loop ends when c = n - a, and enter Step A109; Step A109: Recalculate the new center point corresponding to each user population; Step A110: Repeat Steps A105 to A109 until the new center points of each user population recalculated in Step A109 are the same as the new center points of the corresponding user populations calculated in the previous loop, then the loop ends, obtain a user populations and the corresponding dividing points, and regard them as the population set; Step A111: Let a = a + 1, and return to Step A103; Step A112: Loop Steps A103 to A111 until the loop ends when a = N', obtain N' population sets, N′ = N, and enter Step A113; Step A113: Calculate the average distance corresponding to each population set, regard the population set with the largest average distance as the optimal set, and obtain the user populations in the optimal set.

3. The intelligent cooking optimization method based on machine learning according to claim 2, characterized in that In the above Step A102, the method for calculating the total number of the population N is as follows: preset 4 classification criteria, and the classification criteria correspond to the data in the feature information one by one; according to the classification criteria, conduct population division on each type of data in the feature information, and obtain the number of populations corresponding to each type of data in the feature information; multiply each population number in turn to obtain the total number of the population N; In the step A105, the expression of the point distance is as follows: In the formula, D cb is the point distance from the division point ψ c to the center point ω b , ω bd is the value of the d-th dimension in the center point ω b , ψ cd is the value of the d-th dimension in the division point ψ c , where d ∈ [1, 4]; among them, different dimensions represent different data in the feature information. In the above Step A109, the method for calculating the new center point of each user population includes: where ω' a is the new center point corresponding to the a-th user group, and ψ ar is the r-th partition point in the a-th user group, ψ ar =(ψ ar1 , ψ ar2 , ψ ar3 , ψ ar4 ), ψ ar =(ψ ar1 , ψ ar2 , ψ ar3 , ψ ar4 ) are the values corresponding to the r-th partition point in the a-th user group under different dimensions, R a is the number of partition points in the a-th user group, and r ∈ [1, R a .

4. The intelligent cooking optimization method based on machine learning according to claim 3, wherein In the above A113, the method for calculating the average distance corresponding to each population set includes: Calculate the distance coefficient corresponding to each sample point in each group set, add up the distance coefficients corresponding to each group set in sequence, and then divide by n to obtain the average distance corresponding to each group set; the expression of the distance coefficient is: In the formula, E g is the distance coefficient of the g-th sample point, cw g is the out-group distance of the g-th sample point, cj g is the in-group distance of the g-th sample point, max is the maximum value function, g ∈ [1, n]; The method for calculating the within-group distance corresponding to the g-th sample point is as follows: mark the user population corresponding to the g-th sample point as the current population, and mark all sample points in the current population except the g-th sample point as other points; calculate the point distance from the g-th sample point to each other point, and mark it as the within-cluster distance; add up each within-cluster distance in turn, and then divide by the number of within-cluster distances to obtain the within-group distance of the g-th sample point; The method for calculating the between-group distance corresponding to the g-th sample point is as follows: calculate the point distance from the g-th sample point to each center point, and mark it as the adjacent distance; sort each adjacent distance from large to small, mark the user population corresponding to the center point of the adjacent distance ranked second as the nearest population, and mark all sample points in the nearest population as adjacent points; calculate the point distance from the g-th sample point to each adjacent point, and mark it as the between-cluster distance; add up each between-cluster distance in turn, and then divide by the number of between-cluster distances to obtain the between-group distance of the g-th sample point.

5. The intelligent cooking optimization method based on machine learning according to claim 4, wherein The method for determining food ingredient combinations includes: Setting different digital tags for different user groups and marking them as group tags; taking the group tags corresponding to each user group as analysis data, inputting the analysis data into a trained food ingredient analysis model to predict the corresponding combination tags; the combination tags are the digital tags corresponding to the food ingredient combinations, and the digital tags corresponding to different food ingredient combinations are all different; obtaining the corresponding food ingredient combinations according to the predicted combination tags; The training process of the food ingredient analysis model includes: Pre-collecting h sets of analysis data, setting corresponding combination tags for each of the h sets of analysis data, where h is an integer greater than 1, converting the analysis data and the corresponding combination tags into a corresponding set of feature vectors; taking each set of feature vectors as the input of the food ingredient analysis model, the food ingredient analysis model outputs a set of predicted combination tags corresponding to each set of analysis data, taking the actual combination tag corresponding to each set of analysis data as the prediction target, and the actual combination tag is the pre-set combination tag corresponding to the analysis data; taking minimizing the sum of the prediction errors of all analysis data as the training target; training the food ingredient analysis model until the sum of the prediction errors converges and then stopping the training; the food ingredient analysis model is a deep neural network model.

6. The intelligent cooking optimization method based on machine learning according to claim 5, characterized in that, It also includes: Presetting a similarity threshold, comparing the similarity of each untried dish with the similarity threshold, and taking the dishes corresponding to the similarity values greater than or equal to the similarity threshold as the similar dishes of the corresponding untried dish and adding them to the set of similar dishes of the corresponding untried dish; the dishes corresponding to the similarity values less than or equal to the similarity threshold are not taken as the similar dishes of the corresponding untried dish.

7. The intelligent cooking optimization method based on machine learning according to claim 6, wherein, The expression for similarity is as follows: In the formula, sim(p,q) is the similarity between the p-th dish and the q-th dish, xc kp is the number of times the k-th user selects the p-th dish, xc kq is the number of times the k-th user selects the q-th dish, k ∈ [1, K], K is the number of users, K = n, p ∈ [1, F], q ∈ [1, F], p ≠ q, and F is the number of dishes.

8. The intelligent cooking optimization method based on machine learning according to claim 7, characterized in that, It also includes: Setting sequentially increasing digital tags for M sets of parameter sets and marking them as set tags, and the range of the set tags is [1, M]; Set the initial aperture center C0 and the initial aperture radius L0, and set the iteration number t = 0; among them, Define a loop process, and the loop process is: generate m candidate solutions within the aperture range and reduce the aperture radius L t , 1 < m < M, and the candidate solutions correspond one-to-one with the set labels; calculate the cooking effect corresponding to each candidate solution, and move the aperture center to the candidate solution with the largest cooking effect; repeat the loop process, and each time the loop process is repeated, let the iteration number t = t + 1; preset the iteration threshold G and the radius threshold H; until t ≥ G or L t < H, stop repeating the loop process, obtain the candidate solution corresponding to the aperture center, and mark it as the optimal solution; according to the parameter set corresponding to the set label corresponding to the optimal solution, and mark it as the optimal set; Obtaining the cooking time in the best set and marking it as the best time; presetting a step size set, where the step size set includes dishes and the corresponding time step sizes for the dishes; obtaining the corresponding time step size from the step size set according to the current dish and marking it as the current step size; dividing the best time by the current step size to obtain the number of moments f; inputting the best set into a trained fire power prediction model to predict the fire power magnitude at the first future moment and marking it as the first predicted fire power, and the first future moment is the next moment of the current moment; replacing the initial fire power magnitude in the best set with the first predicted fire power and re-inputting it into the trained fire power prediction model to predict the fire power magnitude at the second future moment, and the second future moment is the next moment of the first future moment; performing cyclic prediction on the fire power magnitude until the fire power prediction model predicts the fire power magnitude at the f-th future moment and marking it as the f-th predicted fire power; taking all the predicted fire power magnitudes as the fire power set, and optimizing and controlling the cooking parameters according to the best set and the fire power set; the training process of the fire power prediction model is the same as that of the food ingredient analysis model, and both are deep neural network models.

9. The intelligent cooking optimization method based on machine learning according to claim 8, wherein, The method for constructing M sets of parameter sets is as follows: preset a parameter range, where the parameter range includes the range corresponding to each parameter in the research parameters; the research parameters include the amount of ingredients, the amount of seasonings, the cooking time, the method label, and the initial fire power; randomly select a value from each range within the parameter range, and construct a set of parameter sets, with a total of M sets of parameter sets, and the M sets of parameter sets are all different; among them, the method label is the digital label corresponding to the cooking method, and the method labels corresponding to different cooking methods are all different; The calculation method of the cooking effect is as follows: set different digital labels for different dishes and mark them as dish labels; take the dish to be cooked as the current dish, obtain the dish label corresponding to the current dish and mark it as the current label; obtain the parameter set corresponding to the set label of the candidate solution, use the parameter set and the current label as evaluation parameters, and input the evaluation parameters into the trained effect evaluation model to evaluate the corresponding cooking effect; the training process of the effect evaluation model is the same as that of the ingredient analysis model, and both are deep neural network models; The expression of the candidate solution is: x i = C t + L t v; where x i is the i-th candidate solution, C t is the corresponding aperture center during the t-th repeated loop process, L t is the corresponding aperture radius during the t-th repeated loop process, v is a random coefficient, i ∈ [1, m], v ∈ [-1, 1]; The expression for reducing the aperture radius is: L t+1 = L t y; where L t+1 is the reduced aperture radius, y is the contraction coefficient, and y ∈ [0, 1].

10. An intelligent cooking optimization system based on machine learning, for implementing the intelligent cooking optimization method based on machine learning according to any one of claims 1-9, characterized in that, It includes: A data collection module for collecting user characteristic data, where the user characteristic data includes the characteristic information of each user; An ingredient determination module for analyzing the user characteristic data, using a clustering algorithm to divide user groups, and using deep learning to determine the ingredient combinations corresponding to each user group; A data acquisition module for acquiring user diet data, where the user diet data includes the historical diet records of each user; A cooking formulation module for integrating the user diet data and the ingredient combinations, using a similarity calculation method to screen dishes, and formulating personalized cooking plans for different user groups; A parameter control module for combining the user diet data and the cooking plan, using an improved optimization algorithm to dynamically control the cooking parameters during the cooking process of each dish.

Images