Group meal purchase optimization method based on order data

Abstract

The invention discloses a group meal purchase optimization method based on order data, and belongs to the technical field of purchase, a neural network is adopted to construct a group meal sales volume prediction model, historical order data is learned through the group meal sales volume prediction model, and the group meal sales volume prediction model with prediction capability is determined. And determining a group meal sales volume prediction result through the group meal sales volume prediction model with prediction capability, and obtaining a group meal purchase plan on the basis of the group meal consumable stock quantity and the group meal sales volume prediction result, thereby realizing reduction of purchase cost and reduction of food material waste.

Description

Technical Field

[0001] The present invention belongs to the technical field of procurement, and in particular relates to a group meal procurement optimization method based on order data. Background Art

[0002] Group meals, whose full name is group catering services, refer to catering services provided collectively for enterprises, institutions, schools, factories, conferences, events, etc. Group meals usually have the following characteristics: Group meal services are usually aimed at a large number of diners, so a large amount of food needs to be prepared and distributed in a short period of time. Group meal services require advance planning, including menu design, ingredient procurement, food preparation, and delivery arrangements. In the process of group meal procurement, due to inaccurate forecasts of ingredient demand, it is easy to purchase too much or too little, resulting in food waste or insufficient supply. In the prior art, group meal procurement mainly relies on manual experience, which lacks scientificity and accuracy. In order to improve the efficiency of group meal procurement and reduce costs, it is necessary to study a group meal procurement optimization method based on order data. Summary of the invention

[0003] The present invention provides a group meal purchasing optimization method based on order data, which is used to solve the problems existing in the prior art.

[0004] A group meal purchasing optimization method based on order data, comprising:

[0005] Acquire historical order data, and preprocess the historical order data to obtain training samples for training a neural network and a label corresponding to each training sample;

[0006] A group meal sales prediction model is constructed using a neural network, and the group meal sales prediction model is trained using training samples and labels corresponding to each training sample to determine a group meal sales prediction model with prediction capabilities;

[0007] Based on the order data of the day, a group meal sales feature sequence is constructed, and a group meal sales prediction model with prediction ability is used to analyze the group meal sales feature sequence to obtain a group meal sales prediction result;

[0008] Based on the inventory quantity of group meal consumables and the forecast results of group meal sales, the group meal procurement plan is obtained to complete the group meal procurement optimization based on order data.

[0009] Furthermore, historical order data is obtained and preprocessed to obtain training samples for training the neural network and labels corresponding to each training sample, including:

[0010] Obtain historical order data, analyze the daily historical order data, and determine the corresponding data on the factors affecting group meal sales each day;

[0011] The data of factors affecting group meal sales for any N consecutive days are constructed as vectors to obtain training samples for training the neural network; and the sales volume corresponding to the N+1th day is used as the label corresponding to the training sample.

[0012] Furthermore, the data on factors affecting group meal sales include at least the type of group meal, the number of collections, the number of positive reviews, the unit price of the product, the sales volume on the day, whether there are coupons, and the amount of the coupons.

[0013] Furthermore, a neural network is used to construct a group meal sales prediction model, including: constructing a GRU-BP neural network, and using the GRU-BP neural network as a group meal sales prediction model.

[0014] Furthermore, the group meal sales prediction model is trained using training samples and labels corresponding to each training sample to determine a group meal sales prediction model with prediction capability, including:

[0015] Initializing a population for optimizing hyperparameters of the group meal sales prediction model;

[0016] For each individual in the population, based on the training samples and the labels corresponding to each training sample, the fitness of each individual is obtained, and the optimal individual is determined according to the fitness of each individual;

[0017] Generate the first random number rand between (0,1) 1 and the second random number rand 2 , and determine the first random number rand 1 With the second random number rand 2 The size relationship between them;

[0018] When the first random number rand 1 Greater than the second random number rand 2 , then enter the first exploration and development process; when the first random number rand 1 Less than the second random number rand 2 When the time comes, it enters the second exploration and development process;

[0019] Repeat the above exploration and development until the current number of training times reaches the maximum number of training times, obtain the optimal individual, and apply the hyperparameters contained in the optimal individual to the group meal sales prediction model to obtain a group meal sales prediction model with prediction capabilities;

[0020] The first exploration and development process includes:

[0021] Generate the first decision coefficient P between (0,1) 1 , and determine the first decision coefficient P 1Is it greater than the first decision threshold φ 1 If yes, then the global random position exploration strategy is used to update the individuals in the population to obtain the updated individuals. Otherwise, the global fast guidance strategy is used to update the individuals in the population according to the optimal individuals to obtain the updated individuals.

[0022] The second exploration and development process includes:

[0023] For any individual in the population, determine a first reference position and a second reference position corresponding to the individual;

[0024] Generate the second decision coefficient P between (0,1) 2 , and determine the second decision coefficient P 2 Is it greater than the second decision threshold φ 2 If so, then according to the first reference position, the second reference position and the optimal individual corresponding to the individual, the multi-guided greedy strategy is used to update the individual to obtain the updated individual. Otherwise, the nearest development strategy is used to update the individual to obtain the updated individual.

[0025] Furthermore, the global random position exploration strategy is used to update the individuals in the population to obtain the updated individuals, including:

[0026] Use Levy flight to generate the third random number rand 3 And use the standard normal distribution to generate the fourth random number rand 4 ;

[0027] According to the third random number rand 3 And use the standard normal distribution to generate the fourth random number rand 4 , update the individuals in the population, and the updated individuals are:

[0028]

[0029] in, represents the d-dimensional parameter of the i-th individual in the t-th training process, i = 1, 2, ..., K, K represents the total number of individuals in the population, d = 1, 2, ..., D, D represents the total dimension of the parameters contained in the individual, represents the d-th dimension parameter of the individual after the i-th update, represents the d-th dimension parameter of the population center, represents the d-th dimension parameter of the first random individual except the i-th individual, represents the d-th dimension parameter of the second random individual other than the i-th individual, rand 5 Represents the fifth random number between (0,1), U drepresents the upper limit of the d-th dimension parameter, L d represents the lower limit of the d-th dimension parameter, and T represents the maximum number of training times.

[0030] Furthermore, the individuals in the population are updated using a global fast bootstrap strategy based on the optimal individual, and the updated individuals are:

[0031]

[0032] in, represents the d-th dimension parameter of the optimal individual, rand 6 Represents the sixth random number between (0,1).

[0033] Furthermore, for any individual in the population, the first reference position and the second reference position corresponding to the individual are determined as:

[0034]

[0035] in, represents the first reference position corresponding to the i-th individual in the t-th training process, represents the first reference position corresponding to the i-th individual in the t+1-th training process, α represents the first exploration coefficient, and θ represents the random degree between (0,2π). represents the first random individual other than the i-th individual, represents the second random individual other than the i-th individual, π represents the circumference of a circle, represents the second reference position corresponding to the i-th individual during the t-th training process, represents the second reference position corresponding to the i-th individual in the t+1th training process, U represents the upper limit vector corresponding to the individual, L represents the lower limit vector corresponding to the individual, and rand 7 represents the seventh random number between (0,1), β represents the second exploration coefficient, and β is randomly 1 or 0;

[0036] According to the first reference position, the second reference position and the optimal individual corresponding to the individual, the multi-guide greedy strategy is used to update the individual to obtain the updated individual, including:

[0037] According to the first reference position and the second reference position corresponding to the individual, the search position is determined as:

[0038]

[0039] in, It represents the first search position corresponding to the ith individual in the tth training process, Indicates the second search position corresponding to the ith individual in the tth training process, rand8 Represents the eighth random number between (0,1), rand 9 Represents the ninth random number between (0,1), represents a random individual, represents the fitness of the first search position, Represents the fitness of the second search position.

[0040] Furthermore, the nearest development strategy is adopted to update the individuals, and the updated individuals are obtained, including:

[0041] According to the fitness corresponding to the first reference position corresponding to the individual, the fitness corresponding to the second reference position and the fitness corresponding to the individual itself, the position with the maximum fitness is determined as the updated individual.

[0042] Furthermore, based on the inventory quantity of group meal consumables and the group meal sales forecast results, the group meal procurement plan is obtained, including:

[0043] Determine the total amount of consumables based on the group meal sales forecast results; the total amount of consumables is obtained by multiplying the sales volume by the consumables for a single group meal;

[0044] The total amount of consumables is multiplied by the preset minimum inventory supply days to obtain the predicted consumables amount;

[0045] Determine whether the predicted amount of consumables is greater than the group meal consumables inventory quantity. If so, subtract the group meal consumables inventory quantity from the predicted amount to get the purchase quantity, and merge the purchase quantities corresponding to each consumable into the same file to get the group meal purchase plan. Otherwise, it is determined that the consumables are sufficient and no purchase is needed for the time being.

[0046] The present invention provides a group meal procurement optimization method based on order data, which adopts a neural network to construct a group meal sales prediction model, and learns historical order data through the group meal sales prediction model to determine a group meal sales prediction model with prediction ability, and determines the group meal sales prediction result through the group meal sales prediction model with prediction ability, and obtains the group meal procurement plan based on the inventory quantity of group meal consumables and the group meal sales prediction result, thereby achieving the reduction of procurement costs and the reduction of food waste. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.

[0048] Figure 1 A flowchart of a group meal purchasing optimization method based on order data provided in an embodiment of the present invention.

[0049] Figure 2A flowchart for determining a group meal sales prediction model with prediction capabilities provided in an embodiment of the present invention.

[0050] The above drawings have shown clear embodiments of the present invention, which will be described in more detail below. These drawings and text descriptions are not intended to limit the scope of the present invention in any way, but to illustrate the concept of the present invention for those skilled in the art by referring to specific embodiments. DETAILED DESCRIPTION

[0051] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present invention. Instead, they are merely examples of devices and methods consistent with some aspects of the present invention as detailed in the appended claims.

[0052] The embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0053] like Figure 1 As shown, an embodiment of the present invention provides a group meal purchase optimization method based on order data, including:

[0054] S11, obtaining historical order data, and preprocessing the historical order data to obtain training samples for training a neural network and a label corresponding to each training sample;

[0055] In an embodiment of the present invention, historical order data is obtained, and the historical order data is preprocessed to obtain training samples for training a neural network and labels corresponding to each training sample, including:

[0056] Obtain historical order data, analyze the daily historical order data, and determine the corresponding data on the factors affecting group meal sales each day;

[0057] The data of factors affecting group meal sales for any N consecutive days are constructed as vectors to obtain training samples for training the neural network; and the sales volume corresponding to the N+1th day is used as the label corresponding to the training sample.

[0058] In an embodiment of the present invention, the group meal sales influencing factor data includes at least the group meal type, the number of collections, the number of positive reviews, the unit price of the product, the sales volume on the day, whether there is a coupon, and the coupon amount.

[0059] It is worth mentioning that the data on factors affecting group meal sales can also include date, day of the week, holiday information, weather conditions, special events, menu types, prices, etc., to make the prediction more accurate.

[0060] S12. Use a neural network to build a group meal sales prediction model, and use training samples and labels corresponding to each training sample to train the group meal sales prediction model to determine a group meal sales prediction model with prediction capabilities;

[0061] The use of neural networks to construct a group meal sales prediction model can include: using BP (Back Propagation) neural network, convolutional neural network, LSTM (Long Short-Term Memory), RBF (Radial Basis Function) neural network, etc. to construct a group meal sales prediction model.

[0062] Using training samples and labels corresponding to each training sample to train the group meal sales prediction model can include: using the training samples as input data, using the labels corresponding to the training samples as expected output data, and using optimization algorithms such as gradient descent algorithm, particle swarm algorithm, genetic algorithm, etc. to train the group meal sales prediction model.

[0063] S13. Based on the order data of the day, a group meal sales feature sequence is constructed, and a group meal sales prediction model with prediction capability is used to analyze the group meal sales feature sequence to obtain a group meal sales prediction result;

[0064] For example, you can take the current order data as a basis, then obtain the order data of the previous N days (including the current order data), and construct a group meal sales feature sequence. The form of the group meal sales feature sequence is actually consistent with the training sample, so that the group meal sales prediction model can accurately identify the data and finally obtain the group meal sales prediction result.

[0065] S14. Based on the inventory quantity of group meal consumables and the forecast results of group meal sales, obtain the group meal procurement plan and complete the group meal procurement optimization based on order data.

[0066] Based on the group meal sales forecast results, the total amount of consumables for the next few days can be determined (the specific number of days can be set by the management staff). Then, when the inventory quantity of group meal consumables is greater than the total amount of consumables for the next few days, there is no need to purchase, otherwise purchase is required.

[0067] In the implementation of the present invention, a forecast is required for each type of group meal consumables, and transportation loss, human loss, and other losses can be added to make procurement more accurate.

[0068] The present invention provides a group meal procurement optimization method based on order data, which adopts a neural network to construct a group meal sales prediction model, and learns historical order data through the group meal sales prediction model to determine a group meal sales prediction model with prediction ability, and determines the group meal sales prediction result through the group meal sales prediction model with prediction ability, and obtains the group meal procurement plan based on the inventory quantity of group meal consumables and the group meal sales prediction result, thereby achieving the reduction of procurement costs and the reduction of food waste.

[0069] In an embodiment of the present invention, a neural network is used to construct a group meal sales prediction model, including: constructing a GRU-BP (Gated Recurrent Unit-BackPropagation) neural network, and using the GRU-BP neural network as a group meal sales prediction model.

[0070] It is worth noting that the embodiment of the present invention uses the GRU-BP neural network as a group meal sales prediction model is only a preferred method, and other neural networks can also be used to construct a group meal sales prediction model.

[0071] like Figure 2 As shown, the group meal sales prediction model is trained using training samples and labels corresponding to each training sample to determine a group meal sales prediction model with prediction capability, including:

[0072] S21. Initializing a population for optimizing hyperparameters of the group meal sales prediction model;

[0073] S22, for each individual in the population, based on the training samples and the labels corresponding to each training sample, obtain the fitness corresponding to each individual, and determine the optimal individual according to the fitness corresponding to each individual;

[0074] S23, generate a first random number rand between (0,1) 1 and the second random number rand 2 , and determine the first random number rand 1 With the second random number rand 2 The size relationship between them;

[0075] S24, when the first random number rand 1 Greater than the second random number rand 2 , then enter the first exploration and development process; when the first random number rand 1 Less than the second random number rand 2 When the time comes, it enters the second exploration and development process;

[0076] S25. Repeat the above exploration and development until the current number of training times reaches the maximum number of training times, obtain the optimal individual, and apply the hyperparameters contained in the optimal individual to the group meal sales prediction model to obtain a group meal sales prediction model with prediction capabilities;

[0077] The first exploration and development process includes:

[0078] Generate the first decision coefficient P between (0,1) 1 , and determine the first decision coefficient P 1 Is it greater than the first decision threshold φ 1 If yes, then the global random position exploration strategy is used to update the individuals in the population to obtain the updated individuals. Otherwise, the global fast guidance strategy is used to update the individuals in the population according to the optimal individuals to obtain the updated individuals.

[0079] The second exploration and development process includes:

[0080] For any individual in the population, determine a first reference position and a second reference position corresponding to the individual;

[0081] Generate the second decision coefficient P between (0,1) 2 , and determine the second decision coefficient P 2 Is it greater than the second decision threshold φ 2 If so, then according to the first reference position, the second reference position and the optimal individual corresponding to the individual, the multi-guided greedy strategy is used to update the individual to obtain the updated individual. Otherwise, the nearest development strategy is used to update the individual to obtain the updated individual.

[0082] Since neural networks cannot be used directly, their hyperparameters (such as weights) need to be optimized before they can effectively identify data. Therefore, the prior art often uses gradient descent and back propagation to optimize hyperparameters, which often falls into local optimality in the early stages of the algorithm, resulting in pseudo-convergence and ultimately making it difficult to achieve prediction. Therefore, an embodiment of the present invention provides an algorithm to solve the problems in the prior art and improve the accuracy of sales forecasting.

[0083] In an embodiment of the present invention, a global random position exploration strategy is used to update individuals in a population to obtain updated individuals, including:

[0084] Use Levy flight to generate the third random number rand 3 And use the standard normal distribution to generate the fourth random number rand 4 ;

[0085] According to the third random number rand 3And use the standard normal distribution to generate the fourth random number rand 4 , update the individuals in the population, and the updated individuals are:

[0086]

[0087] in, represents the d-dimensional parameter of the i-th individual in the t-th training process, i = 1, 2, ..., K, K represents the total number of individuals in the population, d = 1, 2, ..., D, D represents the total dimension of the parameters contained in the individual, represents the d-th dimension parameter of the individual after the i-th update, represents the d-th dimension parameter of the population center, represents the d-th dimension parameter of the first random individual except the i-th individual, represents the d-th dimension parameter of the second random individual other than the i-th individual, rand 5 Represents the fifth random number between (0,1), U d represents the upper limit of the d-th dimension parameter, L d represents the lower limit of the d-th dimension parameter, and T represents the maximum number of training times.

[0088] Optionally, an annealing simulation algorithm may be used to control the above-mentioned global random position exploration strategy, thereby improving the convergence speed of the algorithm and ensuring global search.

[0089] In the embodiment of the present invention, the individuals in the population are updated using a global fast bootstrap strategy according to the optimal individual, and the updated individuals are:

[0090]

[0091] in, represents the d-th dimension parameter of the optimal individual, rand 6 Represents the sixth random number between (0,1).

[0092] Optionally, since the individuals in the population are updated according to the optimal individual using a global fast bootstrap strategy, not only the convergence speed but also the convergence accuracy can be improved, this step can be performed once in each training process when the algorithm reaches 3T / 4, thereby improving the final convergence accuracy.

[0093] In the embodiment of the present invention, for any individual in the population, the first reference position and the second reference position corresponding to the individual are determined as follows:

[0094]

[0095] in, represents the first reference position corresponding to the i-th individual in the t-th training process, represents the first reference position corresponding to the i-th individual in the t+1-th training process, α represents the first exploration coefficient, and θ represents the random degree between (0,2π). represents the first random individual other than the i-th individual, represents the second random individual other than the i-th individual, π represents the circumference of a circle, represents the second reference position corresponding to the i-th individual during the t-th training process, represents the second reference position corresponding to the i-th individual in the t+1th training process, U represents the upper limit vector corresponding to the individual, L represents the lower limit vector corresponding to the individual, and rand 7 represents the seventh random number between (0,1), β represents the second exploration coefficient, and β is randomly 1 or 0;

[0096] According to the first reference position, the second reference position and the optimal individual corresponding to the individual, the multi-guide greedy strategy is used to update the individual to obtain the updated individual, including:

[0097] According to the first reference position and the second reference position corresponding to the individual, the search position is determined as:

[0098]

[0099] in, It represents the first search position corresponding to the ith individual in the tth training process, Indicates the second search position corresponding to the ith individual in the tth training process, rand 8 Represents the eighth random number between (0,1), rand 9 Represents the ninth random number between (0,1), represents a random individual, represents the fitness of the first search position, Represents the fitness of the second search position.

[0100] By searching the reference position, the traversal of the solution space can be enhanced, and more new areas can be searched, thereby increasing the possibility of searching for the global optimal solution.

[0101] In the embodiment of the present invention, the individual is updated by adopting the nearby development strategy to obtain the updated individual, including:

[0102] According to the fitness corresponding to the first reference position corresponding to the individual, the fitness corresponding to the second reference position and the fitness corresponding to the individual itself, the position with the maximum fitness is determined as the updated individual.

[0103] In the embodiment of the present invention, based on the inventory quantity of group meal consumables and the group meal sales forecast result, a group meal procurement plan is obtained, including:

[0104] Determine the total amount of consumables based on the group meal sales forecast results; the total amount of consumables is obtained by multiplying the sales volume by the consumables for a single group meal;

[0105] The total amount of consumables is multiplied by the preset minimum inventory supply days to obtain the predicted consumables amount;

[0106] Determine whether the predicted amount of consumables is greater than the group meal consumables inventory quantity. If so, subtract the group meal consumables inventory quantity from the predicted amount to get the purchase quantity, and merge the purchase quantities corresponding to each consumable into the same file to get the group meal purchase plan. Otherwise, it is determined that the consumables are sufficient and no purchase is needed for the time being.

[0107] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0108] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0109] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0110] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0111] A person of ordinary skill in the art can understand that all or part of the steps in realizing the above-mentioned facts and methods can be completed by instructing the relevant hardware through a program, and the program involved or the program described can be stored in a computer-readable storage medium. When the program is executed, it includes the following steps: At this time, the corresponding method steps are derived, and the storage medium can be ROM / RAM, a disk, an optical disk, etc.

[0112] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A group meal purchasing optimization method based on order data, characterized in that: include: Acquire historical order data, and preprocess the historical order data to obtain training samples for training a neural network and a label corresponding to each training sample; A group meal sales prediction model is constructed using a neural network, and the group meal sales prediction model is trained using training samples and labels corresponding to each training sample to determine a group meal sales prediction model with prediction capabilities; Based on the order data of the day, a group meal sales feature sequence is constructed, and a group meal sales prediction model with prediction ability is used to analyze the group meal sales feature sequence to obtain a group meal sales prediction result; Based on the inventory quantity of group meal consumables and the forecast results of group meal sales, the group meal procurement plan is obtained to complete the group meal procurement optimization based on order data.

2. The group meal purchasing optimization method based on order data according to claim 1 is characterized in that: Obtain historical order data, and preprocess the historical order data to obtain training samples for training a neural network and labels corresponding to each training sample, including: Obtain historical order data, analyze the daily historical order data, and determine the corresponding data on the factors affecting group meal sales each day; The data of factors affecting group meal sales for any N consecutive days are constructed as vectors to obtain training samples for training the neural network; and the sales volume corresponding to the N+1th day is used as the label corresponding to the training sample.

3. The group meal purchasing optimization method based on order data according to claim 2 is characterized in that: The data on factors affecting group meal sales at least include group meal type, number of collections, number of positive reviews, unit price of goods, sales volume on the day, whether there are coupons, and the amount of coupons.

4. The group meal purchasing optimization method based on order data according to claim 1 is characterized in that: A neural network is used to construct a group meal sales prediction model, including: constructing a GRU-BP neural network, and using the GRU-BP neural network as a group meal sales prediction model.

5. The group meal purchasing optimization method based on order data according to claim 1 is characterized in that: The group meal sales prediction model is trained using training samples and labels corresponding to each training sample to determine a group meal sales prediction model with prediction capability, including: Initializing a population for optimizing hyperparameters of the group meal sales prediction model; For each individual in the population, based on the training samples and the labels corresponding to each training sample, the fitness of each individual is obtained, and the optimal individual is determined according to the fitness of each individual; Generate a first random number rand1 and a second random number rand2 between (0,1), and determine the magnitude relationship between the first random number rand1 and the second random number rand2; When the first random number rand1 is greater than the second random number rand2, the first exploration and development process is entered; when the first random number rand1 is less than the second random number rand2, the second exploration and development process is entered; Repeat the above exploration and development until the current number of training times reaches the maximum number of training times, obtain the optimal individual, and apply the hyperparameters contained in the optimal individual to the group meal sales prediction model to obtain a group meal sales prediction model with prediction capabilities; The first exploration and development process includes: Generate a first decision coefficient P1 between (0,1), and determine whether the first decision coefficient P1 is greater than a first decision threshold φ1. If so, use a global random position exploration strategy to update the individuals in the population to obtain the updated individuals. Otherwise, use a global fast guidance strategy to update the individuals in the population according to the optimal individuals to obtain the updated individuals. The second exploration and development process includes: For any individual in the population, determine a first reference position and a second reference position corresponding to the individual; Generate a second decision coefficient P2 between (0,1), and determine whether the second decision coefficient P2 is greater than the second decision threshold φ2. If so, the individual is updated by a multi-guided greedy strategy according to the first reference position, the second reference position and the optimal individual corresponding to the individual to obtain the updated individual. Otherwise, the individual is updated by a nearby development strategy to obtain the updated individual.

6. The group meal purchasing optimization method based on order data according to claim 5 is characterized in that: The global random position exploration strategy is used to update the individuals in the population to obtain the updated individuals, including: A third random number rand3 is generated by using levy flight, and a fourth random number rand4 is generated by using standard normal distribution; According to the third random number rand3 and the fourth random number rand4 generated by the standard normal distribution, the individuals in the population are updated, and the updated individuals are: in, represents the d-dimensional parameter of the i-th individual in the t-th training process, i = 1, 2, ..., K, K represents the total number of individuals in the population, d = 1, 2, ..., D, D represents the total dimension of the parameters contained in the individual, represents the d-th dimension parameter of the individual after the i-th update, represents the d-th dimension parameter of the population center, represents the d-th dimension parameter of the first random individual except the i-th individual, represents the d-th dimension parameter of the second random individual other than the i-th individual, rand5 represents the fifth random number between (0,1), and U d represents the upper limit of the d-th dimension parameter, L d represents the lower limit of the d-th dimension parameter, and T represents the maximum number of training times.

7. The group meal purchasing optimization method based on order data according to claim 6 is characterized in that: According to the optimal individual, the global fast bootstrap strategy is used to update the individuals in the population, and the updated individuals are: in, represents the d-th dimension parameter of the optimal individual, and rand6 represents the sixth random number between (0,1).

8. The group meal purchasing optimization method based on order data according to claim 7 is characterized in that: For any individual in the population, the first reference position and the second reference position corresponding to the individual are determined as: in, represents the first reference position corresponding to the i-th individual in the t-th training process, represents the first reference position corresponding to the i-th individual in the t+1-th training process, α represents the first exploration coefficient, and θ represents the random degree between (0,2π). represents the first random individual other than the i-th individual, represents the second random individual other than the i-th individual, π represents the circumference of a circle, represents the second reference position corresponding to the i-th individual during the t-th training process, represents the second reference position corresponding to the i-th individual in the t+1-th training process, U represents the upper limit vector corresponding to the individual, L represents the lower limit vector corresponding to the individual, rand7 represents the seventh random number between (0,1), β represents the second exploration coefficient, and β is randomly 1 or 0; According to the first reference position, the second reference position and the optimal individual corresponding to the individual, the multi-guide greedy strategy is used to update the individual to obtain the updated individual, including: According to the first reference position and the second reference position corresponding to the individual, the search position is determined as: in, It represents the first search position corresponding to the ith individual in the tth training process, represents the second search position corresponding to the ith individual in the tth training process, rand8 represents the eighth random number between (0,1), and rand9 represents the ninth random number between (0,1). represents a random individual, represents the fitness of the first search position, Represents the fitness of the second search position.

9. The group meal purchasing optimization method based on order data according to claim 8 is characterized in that: The nearest development strategy is used to update individuals, and the updated individuals are as follows: According to the fitness corresponding to the first reference position corresponding to the individual, the fitness corresponding to the second reference position and the fitness corresponding to the individual itself, the position with the maximum fitness is determined as the updated individual.

10. The group meal purchasing optimization method based on order data according to claim 1 is characterized in that: Based on the inventory quantity of group meal consumables and the group meal sales forecast results, obtain the group meal procurement plan, including: Determine the total amount of consumables based on the group meal sales forecast results; the total amount of consumables is obtained by multiplying the sales volume by the consumables for a single group meal; The total amount of consumables is multiplied by the preset minimum inventory supply days to obtain the predicted consumables amount; Determine whether the predicted amount of consumables is greater than the group meal consumables inventory quantity. If so, subtract the group meal consumables inventory quantity from the predicted amount to get the purchase quantity, and merge the purchase quantities corresponding to each consumable into the same file to get the group meal purchase plan. Otherwise, it is determined that the consumables are sufficient and no purchase is needed for the time being.

Images