Recipe generation method based on AI analysis

Abstract

The invention discloses a recipe generation method based on AI analysis. The method comprises the following steps: S1, demand confirmation: a demand acquisition module of a recipe generation system determines the total number N of dish types based on the acquired number of diners in a canteen on the same day and the number of employees in the canteen; s2, generation of a dish demand list: a type analysis module of the recipe generation system generates the dish demand list based on the dish type total number N, the dish type proportion requirement and the environment temperature; s3, generation of a dish recommendation list: inputting a dish demand list to a dish screening module, classifying and sorting dishes in a dish library by the dish screening module based on the constructed dish scoring model, and generating the dish recommendation list after filling the dish demand list; s4, recipe confirmation; and S5, generating a purchase plan. Compared with the prior art, the problem of high dish repetition rate caused by inflexible recipe generation scheme and low richness of an existing recipe generation system is solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of smart canteens, and in particular to a recipe generation method based on AI analysis. BACKGROUND

[0002] Schools, government agencies, enterprises and institutions provide meals for employees or other personnel by setting up canteens. The recipe customization scheme of traditional canteens relies on the chef of the canteen to develop appropriate recipes according to the dining habits of the diners and their own proficiency in dishes. Not only is the level of the chef high, but it also takes a lot of effort for the chef to develop a recipe that is nutritionally balanced, has a low dish repetition rate, and is easy to cook, resulting in an inevitable decline in the quality of the canteen's recipes.

[0003] There are currently some recipe generation systems, but they usually serve special groups of people such as patients, fitness enthusiasts, and weight loss enthusiasts, and develop corresponding recipes with certain functionality based on special purposes. The richness of the recipes is limited by the special nature of the personnel.

[0004] For the short-term recipe needs of special groups of people, the existing recipe generation system can be accepted for its low richness in generating recipes. However, for ordinary canteens, personnel dine in the canteen for a long time, and the personnel are diverse and have more diverse needs. The current recipe generation system has a low flexibility and richness in generating recipes, resulting in a high repetition rate of dishes and an inability to meet the dining needs. SUMMARY

[0005] The present application aims to provide a recipe generation method based on AI analysis to solve the problem of high dish repetition rate caused by the inflexible and low richness of the recipe generation scheme of the existing recipe generation system.

[0006] To achieve the above-mentioned purpose, the present application adopts the following technical scheme: a recipe generation method based on AI analysis, comprising the following steps: S1, demand confirmation: the demand collection module of the recipe generation system determines the total number N of dish types based on the number of diners and the number of canteen staff collected on the day; S2, dish demand list generation: the type analysis module of the recipe generation system generates a dish demand list based on the total number N of dish types, dish type proportion requirements, and environmental temperature. The dish demand list includes dish types and the number of individual dish types; S3, dish recommendation list generation: input the dish demand list to the dish screening module. The dish screening module sorts the dishes in the dish library based on the constructed dish scoring model and generates a dish recommendation list after filling the dish demand list; S4, recipe confirmation: the chef manually confirms the dish recommendation list to obtain the recipe for the day; S5, procurement plan generation: the procurement module of the recipe generation system calculates the dish list based on the daily recipe and the number of diners in the dining hall on the day, and generates a food procurement plan according to the dish list and the food inventory.

[0007] As a further description of the above technical solution: In step S1, the calculation formula of the total number of dish types N is: N=Nb+K*(R1*0.7+R2*0.5-1), where Nb is the reference dish number, K is the scaling factor, R1 is the proportion of diners, and R2 is the attendance rate of dining hall staff. The calculation result is rounded to an integer to obtain the total number of dish types N.

[0008] As a further description of the above technical solution: In step S2, the dish types include high-calorie, high-protein, high-fiber, low-fat light food, and composite carbohydrates, and the dish type proportion requirement is the proportion interval of a single dish type in the total number of dish types N. The type analysis module dynamically adjusts the specific proportion of a single dish type in the total number of dish types N according to the environmental temperature.

[0009] As a further description of the above technical solution: In step S3, the dish scoring model calculates the comprehensive score S of the dish based on five dimensions of nutritional density, seasonal coefficient, cost score, repetition score, and leftover expectation. The calculation formula of the comprehensive score S of the dish is as follows: Comprehensive score S=(nutritional density S1*25)+(seasonal coefficient S2*20)+(cost score S3*20)+(repetition score S4*15)+(leftover expectation S5*20).

[0010] As a further description of the above technical solution: In step S3, the dishes in the dish library of the recipe generation system are labeled by type tags when stored, and the dish classification is stored in a plurality of dish pools.

[0011] As a further description of the above technical solution: In step S3, a visual camera is arranged on the tray recycling conveyor belt, and the visual camera is used to collect the image of the remaining dish in the tray. The leftover analysis module of the recipe generation system identifies the dish and the weight of the dish according to the image of the remaining dish and the daily recipe, and calculates the leftover rate of a single dish.

[0012] As a further description of the above technical solution: In step S5, the procurement module calculates the total demand for food materials based on the dish list, combines the food material inventory data of the inventory management module in the recipe generation system, calculates the net demand for food materials, and generates a food material procurement plan, which includes the target food material name, target food material procurement quantity, and target food material procurement price range.

[0013] As a further description of the above technical solution: The food material procurement plan also includes the backup food material name, backup food material procurement quantity, and backup food material procurement price range.

[0014] In summary, due to the adoption of the above technical solution, the beneficial effects of the present application are: 1、In the present application, the total number of dish types N is adjusted according to the proportion of dining personnel and the attendance rate of canteen staff on the day, balancing the dining demand and the support capacity of the staff.

[0015] 2、In the present application, the model introduces the leftover food rate constraint to form a closed-loop management from demand prediction to effect evaluation, which can effectively control food waste and operating costs while ensuring the scientificity of nutrition. The traditional menu planning process relying on experience is transformed into a standardized and replicable scientific decision-making process, significantly improving the professional level and resource utilization efficiency of canteen management, realizing a feasible digital management path, and the recipe scheme generation is flexible, rich in variety, and low in dish repetition rate.

[0016] 3、In the present application, according to the remaining dish image in the tray collected by the visual camera, the leftover food analysis module uses deep learning and image recognition technology to analyze and recognize the food materials, estimates the weight of the leftover food through the area of the image, and combines the supply quantity of the dish in the daily recipe to calculate the leftover food rate of the single dish. BRIEF DESCRIPTION OF DRAWINGS

[0017] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiments. It should be understood that the following drawings only show some embodiments of the present application, and therefore should not be regarded as a limitation on the scope, and for those skilled in the art, other related drawings can also be obtained without creative labor on the basis of these drawings.

[0018] Figure 1 A flowchart of a recipe generation method based on AI analysis. DETAILED DESCRIPTION

[0019] In order to make the purposes, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, rather than all the embodiments. The components of the embodiments of the present application described and shown in the drawings can be arranged and designed in various different configurations.

[0020] Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the claimed present application, but only represents selected embodiments of the present application. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the present application without creative labor are within the scope of protection of the present application. Embodiment 1

[0021] Please refer to Figure 1 The present application provides a technical solution: a recipe generation method based on AI analysis, comprising the following steps: S1, demand confirmation: the demand collection module of the recipe generation system determines the total number N of dish types based on the collected number of dining personnel in the canteen on the day and the number of canteen employees; S2, dish demand list generation: the type analysis module of the recipe generation system generates a dish demand list based on the total number N of dish types, dish type proportion requirements and environmental temperature, the dish demand list including dish types and the number of individual dish types; S3, dish recommendation list generation: input the dish demand list to the dish screening module, the dish screening module sorts the dishes in the dish library based on the constructed dish scoring model, and generates a dish recommendation list after filling the dish demand list; S4, recipe confirmation: the chef confirms the dish recommendation list manually to obtain the recipe for the day; S5, procurement plan generation: the procurement module of the recipe generation system calculates the dish list based on the recipe for the day and the number of dining personnel in the canteen on the day, and generates a food material procurement plan based on the dish list and the food material inventory.

[0022] In step S5, when the procurement module calculates the dish list, if the number of dining personnel in the canteen on the day is 1000, there are 1000 portions of each of the 5 dishes of high-calorie, high-protein, high-fiber, low-fat light food and composite carbohydrate in the recipe for the day. Taking high-calorie dishes as an example, if there are two dishes of high-calorie type in the recipe for the day, there are 500 portions of each dish, and the portion of each dish is calculated according to its calorie, such as requiring a single high-calorie dish to provide 600 kcal of heat, calculating the weight of a single dish providing 600 kcal of heat. The same applies to other types of dishes.

[0023] In step S1, the total number of dish types N is calculated according to the formula: N=Nb+K*(R1*0.7+R2*0.5-1), where Nb is the reference dish number, K is the scaling factor, R1 is the proportion of diners, and R2 is the attendance rate of canteen staff. The result is rounded to the nearest integer to obtain the total number of dish types N.

[0024] The total number of dish types N is based on the reference dish number, and is adjusted according to the proportion of diners and the attendance rate of canteen staff on the day of the canteen. The proportion of diners is given a weight coefficient of 0.7, and the attendance rate of canteen staff is given a weight coefficient of 0.5, to balance the dining demand and the support capacity of the staff.

[0025] For example, if the reference dish number Nb is set to 20, the scaling factor is set to 10, there are 600 diners, and the proportion of diners R1 is 0.6, there are 28 staff members, and the attendance rate of canteen staff R2 is 0.7.

[0026] At this time, N=20+10×(0.6×0.7+0.7×0.5−1)=17.7≈18, the number of diners and the number of staff are both low, and the number of dishes is reduced.

[0027] In step S2, the dish types include high-calorie, high-protein, high-fiber, low-fat light food, and composite carbohydrates, and the dish type proportion requirement is the proportion interval of a single dish type in the total number of dish types N. The type analysis module dynamically adjusts the specific proportion of a single dish type in the total number of dish types N according to the environmental temperature.

[0028] According to the dish type proportion requirement based on nutritional matching, the proportion of high-calorie dishes in the total number of dish types N is 5-10%, the proportion of high-protein dishes is 20-30%, the proportion of high-fiber dishes is 20-25%, the proportion of low-fat light food dishes is 10-20%, and the proportion of composite carbohydrate dishes is 20-25%. If the environment is in low temperature (<10℃), the type analysis module increases the proportion of high-calorie dishes or high-protein dishes. If the environment is in comfortable temperature (10-25℃), the type analysis module balances the proportion of each type. If the environment is in high temperature (>25℃), the proportion of low-fat light food dishes or high-fiber dishes is increased. The specific adjustment range can be dynamically adjusted later according to the leftover rate and staff score.

[0029] In step S3, a visual camera is installed on the tray recycling conveyor belt to capture the image of the remaining dishes in the tray, so that the leftover analysis module of the recipe generation system can identify the dishes and their weights based on the image of the remaining dishes, the daily recipe, and the dish weight, to obtain the leftover rate of a single dish.

[0030] According to the image of the remaining dishes in the tray collected by the visual camera, the remaining dish analysis module analyzes and identifies the food materials by using deep learning and image recognition technology, estimates the weight of the remaining dishes according to the area of the remaining dishes in the image, combines the supply amount of the dish in the daily menu, and then calculates the remaining dish rate of the single dish.

[0031] Working principle: Embodiment 2

[0032] In the step S5, the procurement module calculates the total demand of food materials based on the dish list, combines the food material inventory data of the inventory management module in the menu generation system, calculates the net demand of food materials, and generates a food material procurement plan, which includes the target food material name, target food material procurement quantity, and target food material procurement price range.

[0033] When the procurement module generates the food material procurement plan, the net demand of food materials is calculated by subtracting the food material inventory from the total demand of food materials, and the food material procurement plan including the target food material name, target food material procurement quantity, and target food material procurement price range is generated to ensure the procurement quality of food materials, maximize the utilization of inventory, and control the cost under the premise of ensuring supply.

[0034] Preferably, the food material procurement plan also includes the backup food material name, backup food material procurement quantity, and backup food material procurement price range.

[0035] In order to further control the cost, the procurement plan of the backup food material with similar nutritional components to the target food material is added to the food material procurement plan, and when the procurement price of the target food material changes, the backup food material with lower price is used to realize price replacement, control the procurement cost, and ensure the stability of meal supply. Embodiment 3

[0036] In the step S3, the dish scoring model calculates the comprehensive score S of the dish based on the nutritional density, seasonal coefficient, cost score, repetition score, and remaining dish expectation, and the calculation formula of the comprehensive score S of the dish is as follows: Comprehensive score S = (nutritional density S1 x 25) + (seasonal coefficient S2 x 20) + (cost score S3 x 20) + (repetition score S4 x 15) + (remaining dish expectation S5 x 20).

[0037] The nutritional density S1 is whether the single serving of a certain type of dish meets the standard, such as whether the high-calorie dish has a calorie of 450 kcal / serving or more, whether the high-protein dish has a protein of 20 g / serving or more, whether the high-fiber dish has a dietary fiber of 5 g / serving or more, whether the low-fat light dish has a fat of less than 8 g / serving, and whether the low GI staple of the compound carbohydrate dish accounts for more than 60%. According to the nutritional density of the dish, the nutritional density S1 is given a specific value, such as 1.0 for more than 20% of the standard value, and 0.8 for meeting the standard.

[0038] The seasonal coefficient S2 is given to the main food material of the season 1.0, the out-of-season food material 0.4, and the transitional period food material 0.7.

[0039] The cost score S3 = 1-(actual cost-benchmark cost) / benchmark cost, and the benchmark cost is the historical average purchase price.

[0040] The repetition score S4 = (100-occurrence in the last 3 days x 15) / 100.

[0041] The leftover expectation S5 = 1-predicted leftover rate. The predicted leftover rate is the average of the historical leftover rates of similar dishes.

[0042] The model forms a closed-loop management from demand prediction to effect evaluation by introducing the leftover rate constraint, which can effectively control food waste and operating costs while ensuring the scientificity of nutrition. The traditional menu planning process relying on experience is transformed into a standardized and replicable scientific decision-making process, significantly improving the professional level and resource utilization efficiency of canteen management, and realizing a feasible digital management path. The recipe scheme generation is flexible, rich in variety, and low in dish repetition rate. Embodiment 4

[0043] In this embodiment, the following improved technical solutions are further made on the basis of the above-mentioned embodiments: in step S3, the dishes in the dish library of the recipe generation system are marked by type tags, and the dishes are classified and stored in a plurality of dish pools.

[0044] The dishes of the five types of high-calorie, high-protein, high-fiber, low-fat light food, and compound carbohydrate are classified and stored, so as to quickly generate a recipe.

[0045] The above is only a preferred specific embodiment of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art can make equivalent replacements or changes to the technical solutions and inventive concepts of the present application within the technical scope disclosed by the present application, which should be covered within the protection scope of the present application.

Claims

1. A recipe generation method based on AI analysis, characterized in that, Includes the following steps: S1. Demand Confirmation: The demand collection module of the recipe generation system determines the total number of dish types N based on the number of diners and the number of canteen staff collected that day. S2. Dish Requirement List Generation: The type analysis module of the recipe generation system generates a dish requirement list based on the total number of dish types N, the required proportion of dish types, and the ambient temperature. The dish requirement list includes the dish types and the quantity of each dish type. S3. Generation of recommended dishes list: Input the list of dishes required into the dish selection module. The dish selection module classifies and sorts the dishes in the dish library based on the constructed dish rating model, and generates a recommended dish list after filling the list of dishes required. S4. Recipe Confirmation: The head chef manually confirms the recommended menu to obtain the daily recipe. S5. Procurement Plan Generation: The procurement module of the recipe generation system calculates a menu based on the daily menu and the number of diners in the canteen that day, and generates a food procurement plan based on the menu and food inventory.

2. The recipe generation method based on AI analysis according to claim 1, characterized in that, In step S1, the formula for calculating the total number of dish types N is: N=Nb+ K*(R1*0.7+ R2*0.5-1), where Nb is the base number of dishes, K is the scaling factor, R1 is the proportion of diners, and R2 is the attendance ratio of canteen staff. The calculation result is rounded to the nearest integer to obtain the total number of dish types N.

3. The recipe generation method based on AI analysis according to claim 1, characterized in that, In step S2, the types of dishes include high-calorie, high-protein, high-fiber, low-fat light meals and complex carbohydrates. The proportion of each dish type is required to be a range of proportions of a single dish type in the total number of dish types N. The type analysis module dynamically adjusts the specific proportion of a single dish type in the total number of dish types N according to the ambient temperature.

4. The recipe generation method based on AI analysis according to claim 1, characterized in that, In step S3, the dish scoring model calculates the overall dish score S based on five dimensions: nutritional density, seasonality coefficient, cost score, repetition score, and expected leftovers. The formula for calculating the overall dish score S is as follows: Overall score S = (Nutritional density S1 × 25) + (Seasonal coefficient S2 × 20) + (Cost score S3 × 20) + (Repetition score S4 × 15) + (Expected leftovers S5 × 20).

5. The recipe generation method based on AI analysis according to claim 1, characterized in that, In step S3, when storing dishes in the recipe generation system's dish library, the dishes are marked by type tags, and the dishes are categorized and stored in several dish pools.

6. The recipe generation method based on AI analysis according to claim 1, characterized in that, In step S3, a vision camera is set on the tray recycling conveyor belt to capture images of the remaining food on the trays. The leftover food analysis module of the recipe generation system identifies the dishes and their weights based on the images of the remaining dishes, the daily recipe, and calculates the leftover rate of each dish.

7. The recipe generation method based on AI analysis according to claim 1, characterized in that, In step S5, the procurement module calculates the total demand for ingredients based on the menu list, and calculates the net demand for ingredients by combining the ingredient inventory data from the inventory management module in the recipe generation system. The procurement plan includes the target ingredient name, the target ingredient quantity, and the target ingredient price range.

8. The recipe generation method based on AI analysis according to claim 7, characterized in that, The food procurement plan also includes the names of backup ingredients, the quantity of backup ingredients to be procured, and the price range of backup ingredients.

Images