Production control method based on food demand prediction

Abstract

The invention relates to the technical field of data processing, and provides a production control method based on food demand prediction, and the method comprises the steps: generating the predicted number of food orders at different moments of the day through a gray system theory prediction model constructed according to the number of historical orders; using an improved grey neural network model constructed according to the average production of historical orders to generate the average predicted production of food orders at different moments of the day; determining the demand predicted quantity of the food on the current day; an improved fuzzy comprehensive evaluation model constructed according to the order plan delivery volume, the food production raw material reserve, the food product inventory cost and the food demand importance is utilized to generate a pre-measurement correction value; and according to the food demand predicted quantity and the predicted quantity correction value, generating the scheduling plan quantity of the food production line at different moments on the same day, and adjusting the scheduling of the food production line. According to the invention, accurate prediction of short-term food demand and better dynamic production scheduling control of periodic small orders are realized.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a production control method based on food demand forecasting. Background Art

[0002] The production line control methods of food production enterprises are divided into order-based static production scheduling control and order-based dynamic production scheduling control. Order-based static production scheduling control means that all order data is known and there will be no changes during the production allocation process. Order-based dynamic production scheduling control means that order information arrives in real time, and the enterprise needs to respond as soon as possible and rearrange the current production plan to ensure timely production of the order. Therefore, in the process of order-based dynamic production scheduling control, the enterprise needs to take interference factors into consideration and formulate dynamic production plans to cope with the ever-changing food production needs, especially the food production needs of small orders.

[0003] In the continuous exploration of digital factory systems for food production companies, it was found that the production demand for small orders of similar foods has a periodic change pattern. Adjustments can be made in the dynamic production scheduling control of orders based on this periodic change pattern to better meet production needs.

[0004] At present, for example, the patent document with application number 202410687555.1 proposes a task collaborative planning method and system for smart factories. However, it only constructs a prediction model through historical food production to predict food demand, and this process is relatively simple.

[0005] Therefore, how to determine a more accurate food demand forecast and combine it with multiple factors to generate a production schedule for adjusting the food production line is a technical problem that needs to be solved. Summary of the invention

[0006] To this end, the present invention provides a production control method based on food demand forecasting, which predicts food demand with a small amount of data and unclear rules in a short period of time through a grey system theory forecasting model and a grey neural network model, and combines a variety of scheduling influencing factors in dynamic production scheduling control to adjust the scheduling of the food production line, thereby achieving accurate prediction of food demand in a short period of time for periodic small orders and better dynamic production scheduling control.

[0007] To achieve the above object, the present invention proposes a production control method based on food demand forecasting, comprising:

[0008] The forecast quantity of food orders at different times of the day is generated using the grey system theory forecast model constructed based on the historical order quantities of multiple days;

[0009] The improved grey neural network model constructed based on the average production volume of historical orders over multiple days is used to generate the average predicted production volume of food orders at different times of the day;

[0010] Determining the forecast demand quantity of food for the day according to the forecast quantity of food orders and the average forecast production quantity of food orders;

[0011] The forecast quantity correction value is generated by using an improved fuzzy comprehensive evaluation model constructed according to the order plan delivery quantity, food production raw material reserve, food product inventory cost and food demand importance;

[0012] Generate a food production line scheduling plan at different times of the day based on the food demand forecast and the forecast correction value;

[0013] Adjust the production schedule of the food production line according to the production line scheduling plan.

[0014] Furthermore, the grey system theory prediction model includes a differential equation. The process of generating the predicted quantity of food orders at different times of the day by using the grey system theory prediction model constructed according to the historical order quantities of multiple days is:

[0015] Construct an order quantity time series using the historical order quantities at different times over multiple days;

[0016] The least square method is used to solve the parameter to be estimated, and a differential equation is constructed through the time series of the parameter to be estimated and the order quantity;

[0017] The order quantity at the current moment is input into the differential equation to generate the predicted food order quantity.

[0018] Furthermore, the average production volume of historical orders for multiple days is used to construct an improved grey neural network model, and the process of generating the average predicted production volume of food orders for the day is as follows:

[0019] The average production volume of the historical orders at different times over multiple days is used to construct a variable sequence, the variable sequence is superimposed once to construct a first-order equation, and the first-order equation is subjected to curve fitting using a back propagation neural network to generate the improved grey neural network model;

[0020] The order production volume at the current moment is input into the grey system theoretical prediction model to generate the average predicted production volume of the food orders.

[0021] Furthermore, the process of using back propagation neural network to perform curve fitting on the first-order equation to generate the improved grey neural network model is as follows:

[0022] The coefficient solution of the first-order equation is used as the output variable of the back-propagation neural network, the superposition sequence after the variable sequence is superimposed once is used as the input variable of the back-propagation neural network, and the back-propagation neural network is fitted and trained through the average production volume of historical orders to generate the improved grey neural network model.

[0023] In the above scheme, the good mining ability of data change rules of gray prediction is combined with the neural network model to realize the accurate prediction of food demand with small amount of data and unclear rules in the short term.

[0024] Furthermore, the improved fuzzy comprehensive evaluation model includes a hierarchy analysis method weight generation sub-model and a fuzzy comprehensive evaluation correction value generation sub-model;

[0025] According to the planned delivery volume of orders for the day, the reserve volume of food production raw materials, the inventory cost of food products and the importance of food demand, the fuzzy comprehensive evaluation model is updated and improved to generate the correction value of the forecast volume as follows:

[0026] Generate multiple first-level weights based on the gradient levels of the planned delivery volume of orders for the day, the reserve volume of food production raw materials, the inventory cost of food products, and the importance of food demand, and generate a judgment matrix based on the production tendency corresponding to the food varieties;

[0027] The first-level weight and the judgment matrix are used to generate a first-level weight vector and a second-level weight vector through the weight generation sub-model of the hierarchical analysis method;

[0028] Construct a membership matrix based on the gradient duration level of the sub-factors in the historical data;

[0029] The first-level weight vector, the second-level weight vector and the membership matrix are used to generate the predicted quantity correction value through the fuzzy comprehensive evaluation correction value generation sub-model.

[0030] Furthermore, the analytic hierarchy process weight generation sub-model performs consistency checks on the primary weight vector and the secondary weight vector respectively. If the consistency check fails, the values ​​of the food demand importance and its sub-factors are reduced, and the primary weight vector and the secondary weight vector are regenerated.

[0031] Furthermore, the sub-factors include: idleness and event risk of food production lines subordinate to the planned order delivery volume, raw material supply stability and raw material reserve production volume ratio subordinate to the food production raw material reserve volume, storage management costs and turnover time subordinate to the food product inventory cost, customer importance subordinate to the food demand importance, seasonal demand influencing factor level and consumer population structure change level;

[0032] Evaluation data is generated for the durations of the evaluation levels of the sub-factors within a plurality of set durations, and the evaluation data is normalized to generate the membership matrix.

[0033] Furthermore, the process of normalizing the evaluation data is: after subtracting the evaluation data average value from the evaluation data, dividing the result by the evaluation data standard value to generate the corresponding element of the membership matrix.

[0034] In the above scheme, the hierarchical analysis method and fuzzy comprehensive evaluation method are used to calculate the forecast correction value, so that it fully considers the influence of various factors on the food demand forecast, and thus makes the value of the production line scheduling plan more accurate.

[0035] Further, the product or weighted sum of the predicted quantity of food orders and the average predicted production volume of food orders is used as the predicted quantity of food demand;

[0036] The forecast quantity correction value is converted to generate a forecast quantity correction coefficient, and the product or weighted sum of the forecast quantity correction coefficient and the food demand forecast quantity is used as the production line scheduling plan quantity.

[0037] Furthermore, if the current total output of the food production line is less than the production line scheduling plan after the set time, the adjustment time is determined according to the difference, and the food production line is adjusted to enter the handover period of the next production cycle at the adjustment time.

[0038] Compared with the prior art, the present invention has the following beneficial effects:

[0039] 1. Through the grey system theory prediction model and the grey neural network model, the food demand with a small amount of data and unclear rules in the short term is predicted, and the production schedule of the food production line is adjusted by combining various scheduling influencing factors in dynamic production scheduling control, realizing accurate prediction of food demand in the short term for periodic small orders and better dynamic production scheduling control.

[0040] 2. Through the good mining ability of data change rules of gray prediction, it is combined with the neural network model to realize the accurate prediction of food demand with small amount of data and unclear rules in the short term.

[0041] 3. Through the good mining ability of data change rules of gray prediction, it is combined with the neural network model to realize the accurate prediction of food demand with small amount of data and unclear rules in the short term.

[0042] 4. Apply the analytic hierarchy process and fuzzy comprehensive evaluation method to calculate the forecast correction value, so that it can fully consider the influence of various factors on the food demand forecast, and thus make the value of the production line scheduling plan more accurate. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 A schematic diagram of a general flow chart of a production control method based on food demand forecasting according to an embodiment of the present invention;

[0044] Figure 2 It is a schematic diagram of the construction process of an improved grey neural network model of a production control method based on food demand forecasting according to an embodiment of the present invention;

[0045] Figure 3 It is a detailed flow chart of an improved fuzzy comprehensive evaluation model of a production control method based on food demand forecasting according to an embodiment of the present invention;

[0046] Figure 4 The present invention is a schematic diagram of the process of adjusting production scheduling of the production control method based on food demand forecasting according to an embodiment of the present invention. DETAILED DESCRIPTION

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

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

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

[0050] In addition, it should be noted that in the description of the present invention, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For those skilled in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0051] like Figures 1 to 4As shown, the present invention provides a production control method based on food demand forecasting, which predicts the food demand with a small amount of data and unclear rules in a short period of time through a grey system theory forecasting model and a grey neural network model, and combines a variety of scheduling influencing factors in dynamic production scheduling control to adjust the scheduling of the food production line, thereby achieving accurate prediction of food demand in a short period of time for periodic small orders and better dynamic production scheduling control.

[0052] in, Figure 1 A schematic diagram of a general flow chart of a production control method based on food demand forecasting according to an embodiment of the present invention; Figure 2 It is a schematic diagram of the construction process of an improved grey neural network model of a production control method based on food demand forecasting according to an embodiment of the present invention; Figure 3 It is a detailed flow chart of an improved fuzzy comprehensive evaluation model of a production control method based on food demand forecasting according to an embodiment of the present invention; Figure 4 The present invention is a schematic diagram of the process of adjusting production scheduling of the production control method based on food demand forecasting according to an embodiment of the present invention.

[0053] like Figures 1 to 4 As shown, this embodiment proposes a production control method based on food demand forecasting, including: using a gray system theory forecasting model constructed according to the historical order quantities of multiple days to generate the predicted quantity of food orders at different times of the day; using an improved gray neural network model constructed according to the average production volume of historical orders of multiple days to generate the average predicted production volume of food orders at different times of the day; determining the predicted quantity of food demand for the day based on the predicted quantity of food orders and the average predicted production volume of food orders; using an improved fuzzy comprehensive evaluation model constructed according to the planned delivery quantity of orders, the reserve quantity of food production raw materials, the inventory cost of food products and the importance of food demand to generate a forecast quantity correction value; generating a food production line scheduling plan quantity at different times of the day based on the food demand forecast quantity and the predicted quantity correction value; and adjusting the production schedule of the food production line based on the production line scheduling plan quantity.

[0054] It should be noted that the method described in this embodiment is applied to the control of dynamic production scheduling of a food production line of the same kind of food for one day or less.

[0055] Furthermore, the grey system theoretical prediction model includes a differential equation. The process of generating the predicted quantity of food orders at different times of the day using the grey system theoretical prediction model constructed based on the historical order quantities of multiple days is as follows: constructing an order quantity time series using the historical order quantities of multiple days and different times; constructing a differential equation using the order quantity time series, and using the least squares method to solve the estimated parameters of the differential equation; and inputting the order quantity at the current moment into the differential equation to generate the predicted quantity of food orders.

[0056] Specifically, the grey system theory prediction model essentially processes the original data and establishes a differential equation that reflects the regular changes between data, thereby predicting the future development trend of the data. The specific implementation process is as follows:

[0057] The historical order quantities at different times over multiple days are used to construct an order quantity time series, which is represented by X (0) ={X (0) (1),X (0) (2),...,X (0) (m)};

[0058] The time series are stacked one by one to generate a stacked sequence, which is represented by X (1) ={X (1) (1),X (1) (2),...,X (1) (m)}, where

[0059] The differential equation is constructed by the superposition sequence, and the differential equation is specifically:

[0060]

[0061] In the formula, p and q are two parameters to be estimated, X (1) Represents a superimposed sequence.

[0062] The least square method is used to solve the estimated parameters of the differential equation, specifically:

[0063] (p,q) T =(B T B) -1 B T A

[0064] In the formula, p and q are two parameters to be estimated, A is the superposition sequence, B is the characteristic matrix, and the calculation process of the characteristic matrix B is:

[0065]

[0066] In the formula, B is the feature matrix, X (1) (n) represents the nth superposition sequence.

[0067] The order quantity at the current moment is input into the differential equation to generate the predicted food order quantity, thereby realizing the prediction of the food order quantity at different moments of the day.

[0068] Furthermore, the average production volume of historical orders for multiple days is used to construct an improved grey neural network model, and the process of generating the average predicted production volume of food orders for the day is as follows:

[0069] The average production volume of the historical orders at different times over multiple days is used to construct a variable sequence, the variable sequence is superimposed once to construct a first-order equation, and the first-order equation is subjected to curve fitting using a back propagation neural network to generate the improved grey neural network model;

[0070] The order production volume at the current moment is input into the grey system theoretical prediction model to generate the average predicted production volume of the food orders.

[0071] Furthermore, the process of using back propagation neural network to perform curve fitting on the first-order equation to generate the improved grey neural network model is as follows:

[0072] The coefficient solution of the first-order equation is used as the output variable of the back-propagation neural network, the superposition sequence after the variable sequence is superimposed once is used as the input variable of the back-propagation neural network, and the back-propagation neural network is fitted and trained through the average production volume of historical orders to generate the improved grey neural network model.

[0073] Specifically, the first-order equation is:

[0074]

[0075] In the formula, a and u are two coefficients to be estimated, X (1) It should be noted that although the expression of the first-order equation is consistent with that of the differential equation, the least squares method is suitable for model fitting with relatively large sample sizes, while the back propagation neural network is more suitable for model fitting with very small sample sizes.

[0076] Transform the first-order equation to generate a solution sequence expression:

[0077]

[0078] In the formula, a and u are two coefficients to be estimated, X (1) (n) represents the nth cumulative sequence, and bn-1 is the coefficient to be estimated corresponding to the nth cumulative sequence.

[0079] The expression of the back propagation neural network is:

[0080]

[0081] In the formula, w 11 、w 2i 、w 3i 、w 4i They represent the weights of the neurons in the first to fourth layers respectively, and θ is the threshold of the output node.

[0082] Therefore, the estimated coefficients of the prediction expression are fitted by back-propagation neural network.

[0083] In the above scheme, the good mining ability of data change rules of gray prediction is combined with the neural network model to realize the accurate prediction of food demand with small amount of data and unclear rules in the short term.

[0084] Furthermore, the improved fuzzy comprehensive evaluation model includes a hierarchical analysis method weight generation sub-model and a fuzzy comprehensive evaluation correction value generation sub-model; the process of updating the improved fuzzy comprehensive evaluation model to generate the forecast quantity correction value according to the planned delivery quantity of orders for the day, the reserve quantity of food production raw materials, the inventory cost of food products and the importance of food demand is:

[0085] Generate multiple first-level weights according to the gradient levels of the values ​​of the planned delivery volume of orders for the day, the reserve volume of food production raw materials, the inventory cost of food products and the importance of food demand, and generate a judgment matrix according to the production tendency corresponding to the food varieties; generate a first-level weight vector and a second-level weight vector by the weight generation sub-model of the hierarchical analysis method from the first-level weights and the judgment matrix;

[0086] A membership matrix is ​​constructed according to the gradient duration level of the sub-factors in the historical data; the first-level weight vector, the second-level weight vector and the membership matrix are used to generate the predicted quantity correction value through the fuzzy comprehensive evaluation correction value generation sub-model.

[0087] For example, if the predicted food variety is seasonal food or popular food, the first-level weight of the food demand importance value is set to 70 to 90, which is proportional to the average daily production of food for multiple days. For example, if the planned delivery volume of orders, the reserve volume of food production raw materials, and the inventory cost of food products on the day are less than the corresponding standard volume, the ratio of the difference to the standard volume is set as the first-level weight, otherwise the first-level weight is the set value, and the set value is preferably 50, indicating that it has little impact on the predicted food demand.

[0088] Furthermore, the sub-factors include: idleness and event risks of food production lines belonging to the planned delivery volume of the order, raw material supply stability and raw material reserve production ratio belonging to the food production raw material reserve, storage management costs and turnover time belonging to the food product inventory cost, customer importance, seasonal demand influencing factor level and consumer population structure change level belonging to the food demand importance; generate evaluation data for the duration of the evaluation level of the sub-factors within multiple set durations, and normalize the evaluation data to generate the membership matrix.

[0089] Specifically, secondary weights reflecting the impact of each sub-factor on the primary index are set according to the predicted food type. For example, for seasonal food or popular food, the ratio of event risk to food production line idleness in the judgment matrix is ​​greater than 1.

[0090] For example, the gradient duration levels are divided into low, medium, and high, and their scoring values ​​are 25, 50, and 75, respectively. Construct a membership matrix based on the gradient duration levels of the sub-factors in the historical data. For example, the duration of the food production line idleness less than 20% in the first hour of the food production of the variety on that day is a low gradient duration level, greater than or equal to 20% and less than 60% is a medium gradient duration level, and greater than or equal to 60% is a high gradient duration level. Divide the above low gradient duration level by 1 hour to obtain the elements of the membership matrix. For example, the low, medium, and high gradient duration levels are 10 minutes, 40 minutes, and 10 minutes, respectively. The membership matrix corresponding to the idle food production line is |0.16 0.68 0.16|, so a membership matrix that evaluates the scores of each sub-factor on the first-level indicator can be generated.

[0091] Furthermore, the analytic hierarchy process weight generation sub-model performs consistency checks on the primary weight vector and the secondary weight vector respectively. If the consistency check fails, the values ​​of the food demand importance and its sub-factors are reduced, and the primary weight vector and the secondary weight vector are regenerated.

[0092] Specifically, the consistency test is to calculate the consistency index CI and the test coefficient CR. If the test coefficient CR is less than 0.1, the consistency test is passed, otherwise it fails. If the consistency test of the first-level weight vector fails, the first-level weight of the food demand importance is reduced by 0.05. If the consistency test of the second-level weight vector fails, the second-level weight of the seasonal demand influencing factor level is reduced by 0.05.

[0093] Furthermore, the process of normalizing the evaluation data is: after subtracting the evaluation data average value from the evaluation data, the evaluation data is divided by the evaluation data standard value to generate the corresponding element of the membership matrix. Therefore, by introducing the evaluation data standard value in the evaluation data normalization process, the correction and adjustment of the evaluation data and the membership matrix are achieved.

[0094] Specifically, the evaluation standard value is a parameter of the food production line when the food production line is in static production arrangement, for example, the idleness of the food production line is 10% when the food production line is in static production arrangement.

[0095] It can be understood that in the fuzzy comprehensive evaluation correction value generation sub-model, the evaluation set of each primary indicator is obtained by cross-producting the secondary weight vector and the membership matrix, such as |0.26, 0.50, 0.23|, and the evaluation set is cross-producted by the transpose of the score matrix |25, 50, 75| TA first-level indicator score of 49 is obtained. All evaluation sets are combined to generate a first-level indicator evaluation matrix. The first-level weight vector is dot-multiplied by the first-level indicator evaluation matrix to obtain the evaluation result vector. The evaluation result vector is, for example, |0.296, 0.310, 0.174|. The evaluation result vector is cross-multiplied by the transpose of the evaluation score matrix |25, 50, 75| T The predicted amount correction value 36 is obtained.

[0096] It can be understood that the process of the hierarchical analysis method is a conventional process and will not be described in detail here.

[0097] In the above scheme, the hierarchical analysis method and fuzzy comprehensive evaluation method are used to calculate the forecast correction value, so that it fully considers the influence of various factors on the food demand forecast, and thus makes the value of the production line scheduling plan more accurate.

[0098] Furthermore, the product or weighted sum of the predicted quantity of food orders and the average predicted production quantity of food orders is used as the predicted food demand; the predicted quantity correction value is converted to generate a predicted quantity correction coefficient, and the product or weighted sum of the predicted quantity correction coefficient and the predicted food demand is used as the production line scheduling plan.

[0099] Specifically, the coefficient of the weighted summation is set according to the characteristics of the food variety. For example, if the food demand forecast for fast food is relatively vague, the coefficients of the weighted summation of the food order forecast quantity and the average forecast production quantity of the food orders are 0.3 and 0.7 respectively. If the adjustment sensitivity requirement is higher, the coefficient of the weighted summation of the forecast quantity correction coefficient and the food demand forecast quantity are 0.8 and 0.2 respectively.

[0100] Furthermore, if the total output of the current food production line is less than the production line scheduling plan after the set time, the adjustment time is determined according to the difference, and the food production line is adjusted to enter the handover period of the next production cycle at the adjustment time. The set time is preferably 15 minutes, the difference is preferably proportional to the adjustment time, and the proportional coefficient is preferably 0.4, which can be adjusted according to actual conditions.

[0101] Specifically, Figure 4 As shown, the adjustment time is Figure 4At the adjustment time t2-c in the example, the cloud management platform moves the order allocation time node forward by c time, so that the production enterprise can adjust and prepare for the next production cycle at t2, for example, reduce the production work that is currently in progress and has a lower priority than the inserted order. Since the cloud management platform allocates and prepares the orders in the order pool for a period of (t2, t3) in advance at t2-c, compared with the cloud management platform allocating and preparing the orders in the order pool for a period of (t2, t3) at t2, the idle time caused by the production enterprise starting to prepare for production at t2 is reduced, and the accurate prediction of the period c realizes the accurate prediction of periodic small orders and short-term food demand and better dynamic production scheduling control.

[0102] It can be understood that this embodiment predicts the food demand with a small amount of data and unclear rules in a short period of time through the gray system theory prediction model and the gray neural network model, and adjusts the production schedule of the food production line by combining various scheduling influencing factors in dynamic production scheduling control, so as to achieve accurate prediction of food demand for periodic small orders and short periods of time and better dynamic production scheduling control. Through the good mining ability of the data change law of gray prediction, it is combined with the neural network model to achieve accurate prediction of food demand with a small amount of data and unclear rules in a short period of time. Through the good mining ability of the data change law of gray prediction, it is combined with the neural network model to achieve accurate prediction of food demand with a small amount of data and unclear rules in a short period of time. The hierarchical analysis method and fuzzy comprehensive evaluation method are used to calculate the forecast correction value, so that it fully considers the influence of various factors on the forecast of food demand, and thus makes the numerical value of the production line scheduling plan obtained more accurate.

[0103] So far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it is easy for those skilled in the art to understand that the protection scope of the present invention is obviously not limited to these specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will fall within the protection scope of the present invention.

[0104] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A production control method based on food demand forecasting, characterized in that: include: The forecast quantity of food orders at different times of the day is generated using the grey system theory forecast model constructed based on the historical order quantities of multiple days; The improved grey neural network model constructed based on the average production volume of historical orders over multiple days is used to generate the average predicted production volume of food orders at different times of the day; Determining the forecast demand quantity of food for the day according to the forecast quantity of food orders and the average forecast production quantity of food orders; The forecast quantity correction value is generated by using an improved fuzzy comprehensive evaluation model constructed according to the order plan delivery quantity, food production raw material reserve, food product inventory cost and food demand importance; Generate a food production line scheduling plan at different times of the day based on the food demand forecast and the forecast correction value; Adjust the production schedule of the food production line according to the production line scheduling plan.

2. The production control method based on food demand forecasting according to claim 1, characterized in that: The grey system theory prediction model includes a differential equation. The process of generating the predicted quantity of food orders at different times of the day by using the grey system theory prediction model constructed according to the historical order quantities of multiple days is as follows: Construct an order quantity time series using the historical order quantities at different times over multiple days; The least square method is used to solve the parameter to be estimated, and a differential equation is constructed through the time series of the parameter to be estimated and the order quantity; The order quantity at the current moment is input into the differential equation to generate the predicted food order quantity.

3. The production control method based on food demand forecasting according to claim 1, characterized in that: The process of generating the average predicted production volume of food orders for the day by using the improved grey neural network model constructed by using the average production volume of historical orders for multiple days is as follows: The average production volume of the historical orders at different times over multiple days is used to construct a variable sequence, the variable sequence is superimposed once to construct a first-order equation, and the first-order equation is subjected to curve fitting using a back propagation neural network to generate the improved grey neural network model; The order production volume at the current moment is input into the improved grey neural network model to generate the average predicted production volume of the food orders.

4. The production control method based on food demand forecasting according to claim 3 is characterized in that: The process of generating the improved grey neural network model is: The coefficient solution of the first-order equation is used as the output variable of the back-propagation neural network, the superposition sequence after the variable sequence is superimposed once is used as the input variable of the back-propagation neural network, and the back-propagation neural network is fitted and trained through the average production volume of historical orders to generate the improved grey neural network model.

5. The production control method based on food demand forecasting according to claim 1, characterized in that: The improved fuzzy comprehensive evaluation model includes a hierarchy analysis method weight generation sub-model and a fuzzy comprehensive evaluation correction value generation sub-model; According to the planned delivery volume of orders for the day, the reserve volume of food production raw materials, the inventory cost of food products and the importance of food demand, the fuzzy comprehensive evaluation model is updated and improved to generate the correction value of the forecast volume as follows: Generate multiple first-level weights based on the gradient levels of the planned delivery volume of orders for the day, the reserve volume of food production raw materials, the inventory cost of food products, and the importance of food demand, and generate a judgment matrix based on the production tendency corresponding to the food varieties; The first-level weight and the judgment matrix are used to generate a first-level weight vector and a second-level weight vector through the weight generation sub-model of the hierarchical analysis method; Construct a membership matrix based on the gradient duration level of the sub-factors in the historical data; The first-level weight vector, the second-level weight vector and the membership matrix are used to generate the predicted quantity correction value through the fuzzy comprehensive evaluation correction value generation sub-model.

6. The production control method based on food demand forecasting according to claim 5, characterized in that: The analytic hierarchy process weight generation sub-model performs consistency checks on the primary weight vector and the secondary weight vector respectively. If the consistency check fails, the values ​​of the food demand importance and its sub-factors are reduced, and the primary weight vector and the secondary weight vector are regenerated.

7. The production control method based on food demand forecasting according to claim 5, characterized in that: The sub-factors include: idleness and event risks of food production lines belonging to the planned order delivery volume, raw material supply stability and raw material reserve production volume ratio belonging to the food production raw material reserve volume, storage management costs and turnover time belonging to the food product inventory cost, customer importance, seasonal demand influencing factor level and consumer population structure change level belonging to the food demand importance; Evaluation data is generated for the durations of the evaluation levels of the sub-factors within a plurality of set durations, and the evaluation data is normalized to generate the membership matrix.

8. The production control method based on food demand forecasting according to claim 7, characterized in that: The process of normalizing the evaluation data is as follows: after subtracting the evaluation data average value from the evaluation data, the evaluation data is divided by the evaluation data standard value to generate the corresponding element of the membership matrix.

9. The production control method based on food demand forecasting according to any one of claims 1 to 8, characterized in that: The product or weighted sum of the predicted quantity of food orders and the average predicted production volume of food orders is used as the predicted quantity of food demand; The forecast quantity correction value is converted to generate a forecast quantity correction coefficient, and the product or weighted sum of the forecast quantity correction coefficient and the food demand forecast quantity is used as the production line scheduling plan quantity.

10. The production control method based on food demand forecasting according to any one of claims 1 to 8, characterized in that: If the current total output of the food production line is less than the production line scheduling plan after the set time, the adjustment time is determined according to the difference, and the food production line is adjusted to enter the handover period of the next production cycle at the adjustment time.

Images