An automatic vending machine goods configuration method based on artificial intelligence

Abstract

The present invention discloses a method for replenishing goods of a vending machine. Through this method, effective goods replenishment operations can be achieved for various vending machines that cannot access the network to provide sales data in real time. The method includes: obtaining the total sales volume STotal of each single-category commodity in each vending machine during the replenishment cycle; integrating the total sales volume data into an M×N-dimensional sales volume prediction matrix S; calculating the maximum storage matrix Q and the remaining commodity matrix Y; constructing a calculation model for the replenishment quantity matrix R involving all vending machines in the vending machine management area, performing actual replenishment operations on the vending machines in the vending machine management area according to the replenishment quantity matrix R, and calculating the replenishment loss RLost according to the actual replenishment situation to optimize the calculation model of the replenishment quantity matrix R.

Description

Technical Field

[0001] The present invention relates to the technical field of vending machines, and particularly to a method for configuring goods in a vending machine based on artificial intelligence. Background Art

[0002] Vending machines have increasingly entered people's lives, and the management of commodity distribution for vending machines has become increasingly important. In the prior art, for example, a vending machine and a vending control system based on real-time intelligent control of the Internet of Things are disclosed in CN204946178U. The vending control system wirelessly communicates with each vending machine through a monitoring center to real-time understand information such as the working status and sales situation of multiple vending machines, realizing intelligent control of the vending machines, which is beneficial to ensuring the stable working performance of the vending machines and avoiding situations such as out-of-stock and internal system failures of the vending machines. Another example is that an automatic replenishment method and device are disclosed in CN105205926A. The automatic replenishment system includes a vending machine, a server, a query terminal, and a replenishment terminal. The vending machine includes a plurality of vending grids for placing goods, and the server stores the sales status of the goods in the vending grids. The query terminal sends a replenishment instruction to the replenishment terminal through the server; the replenishment terminal receives the replenishment instruction and displays it. At the same time, a device for managing the distribution of goods in a vending machine and an automatic vending system are disclosed in CN204463203U. The device for managing the distribution of goods includes a server, and the server includes a data receiving unit for receiving various data transmitted from each vending machine and from a commodity information input device. All the various data at least include replenishment commodity information data of the goods replenished into each vending machine and all goods information data to be distributed to all vending machines. This device and system can effectively track and monitor all goods to be distributed to the vending machines, so that there will be no shortage of goods in the commodity distribution link.

[0003] It can be seen from the above-mentioned prior art that the replenishment of vending machines still adopts the traditional periodic replenishment scheme, that is, the vending machines are replenished after a fixed period. The characteristics of this method are that it is relatively easy and time-saving, and only needs to be replenished on fixed days. However, there are many problems such as uncertain types of goods and uncertain number of goods, which will also affect the sales of goods, making it impossible to achieve both easy and increased sales. Although the sampling method of real-time sales data monitoring of vending machines in the prior art is more accurate, this method relies on real-time communication between the vending machine and the remote service with large amounts of data. When the operator of the vending machine has laid a large number of vending machines, due to the need to generate large amounts of communication, it will cause a significant increase in operating costs. A more prominent problem is that there are still a large number of old vending machines in the existing market that cannot access the network. Especially in the European and American markets, such old vending machines are very common. Such vending machines cannot provide sales data to the server platform in real time, so it is impossible to perform more effective goods replenishment configuration for such vending machines. It can be seen that there is a need in the art for a method of configuring goods for a vending machine that is low-cost, accurate and effective. Summary of the invention

[0004] One of the technical problems to be solved by the present invention is to provide a method for replenishing goods for a vending machine, by which an effective goods replenishment operation can be implemented for various vending machines that cannot access the network to provide sales data in real time.

[0005] In order to solve the above technical problems, the present invention provides a method for configuring goods for a vending machine based on artificial intelligence, the method comprising:

[0006] Step S1, obtaining historical sales data and channel configuration information of each vending machine in the vending machine management area;

[0007] Step S2: predict the daily sales volume s of each type of commodity in each vending machine in the vending machine management area during the replenishment cycle, and obtain the total sales volume S of each type of commodity in each vending machine during the replenishment cycle by summing the daily sales volume s. Total ;

[0008] Step S3, integrating the total sales data of all types of goods of all vending machines in the vending machine management area during the replenishment cycle into an M×N dimensional sales volume prediction matrix S; wherein M is the number of categories of goods sold by all vending machines, and N is the number of all vending machines;

[0009] Step S4: Calculate the maximum storage matrix Q and the remaining quantity matrix Y of all vending machines in the vending machine management area based on the out-of-stock time and historical sales data of each vending machine in the vending machine management area.

[0010] Step S5: Construct a calculation model for the replenishment quantity matrix R of all vending machines in the vending machine management area, that is, the replenishment quantity matrix R = α × sales volume prediction matrix S - β × remaining quantity matrix Y of goods; where α and β are adjustment coefficients.

[0011] Step S6: Perform actual replenishment operations on the vending machines in the vending machine management area according to the replenishment quantity matrix R, and calculate the replenishment loss R according to the actual replenishment situation. Lost ;

[0012] Step S7: Modulate the adjustment coefficients α and β according to the replenishment loss R Lost to optimize the calculation model of the replenishment quantity matrix R. And use the optimized calculation model of the replenishment quantity matrix R to predict the replenishment quantity in the next replenishment cycle.

[0013] In one embodiment, the process of predicting the daily sales volume s of each single-category product in each vending machine in the vending machine management area during the replenishment cycle includes:

[0014] Extract the historical sales volume data sequence L of a single-category product of a single vending machine. If the historical sales volume data sequence L is greater than or equal to 365, that is, it means that the historical sales volume data sequence L records sales data for more than 365 days. Then, extract the sales volume data on the 365th day before, denoted as S 365 ; Further, whether the 365th day before is a holiday. If it is a holiday, then the sales volume s = S 365 ×0.6 + S three ×0.2 + S 7 ×0.2; If the 365th day before is not a holiday, then the sales volume s = S 365 ×0.2 + S three ×0.4 + S 7 ×0.4; where S three is the triple exponential smoothing prediction value, and S 7 is the average value of the sales volume data in the previous 7 days at the current time point.

[0015] In one embodiment, if the historical sales volume data sequence L is less than 365, then further determine whether the historical sales volume data sequence L is greater than or equal to 7, that is, whether the historical sales volume data sequence L records historical sales data for more than or equal to 7 days. If the historical sales volume data sequence L is greater than or equal to 7, then extract all the sales volumes of the historical sales volume data sequence L, and let the sales volume s) = S three ×0.5 + S7 ×0.5。

[0016] In one embodiment, the calculation process of the maximum storage matrix Q includes: calculating the mode of the total sales volume of each type of commodity between two out-of-stock time points of each vending machine, which is the maximum storage volume of this type of commodity for this vending machine; integrating the maximum storage volumes of each type of commodity for each vending machine obtained above to obtain the maximum storage matrix Q of all types of commodities of all vending machines in the management area.

[0017] In one embodiment, the calculation process of the commodity remaining matrix Y includes: finding the time point closest to the current time among the out-of-stock times, starting from the closest out-of-stock time point, and calculating the remaining quantity of each type of commodity for each vending machine at the current time based on the maximum storage volume and the daily sales volume of this commodity; integrating the remaining quantities of each type of commodity for each vending machine obtained above to obtain the commodity remaining matrix Y of all types of commodities of all vending machines in the management area.

[0018] In one embodiment, the replenishment loss R Lost includes the labor loss Lost (人工) , the out-of-stock loss Lost (缺货) , and the transportation loss Lost (运输) ; and the replenishment loss R Lost = a×Lost (人工) + b×Lost (缺货) + c×Lost (运输) ; where a, b, and c are the proportionality coefficients of the labor loss Lost (人工) , the out-of-stock loss Lost (缺货) , and the transportation loss Lost (运输) respectively. The proportionality coefficients a, b, and c are used to adjust the order-of-magnitude differences caused by different units among the three losses.

[0019] In one embodiment, the calculation method of the labor loss Lost (人工) is to sum all the elements of the replenishment matrix R to obtain the total replenishment quantity, and this total replenishment quantity is recorded as the labor loss Lost (人工) .

[0020] In one embodiment, the calculation method of the out-of-stock loss Lost (缺货) is to obtain the out-of-stock quantity matrix X according to the sales volume matrix S, the replenishment quantity matrix R, and the commodity remaining matrix Y, that is, the out-of-stock quantity matrix X = the sales volume matrix S - the replenishment quantity matrix R - the commodity remaining matrix Y; multiply the out-of-stock matrix X by the commodity price matrix P, and sum all the elements of the multiplied matrix to obtain the profit loss caused by out-of-stock. This profit loss is recorded as the out-of-stock loss Lost (缺货) .

[0021] In one embodiment, the transportation loss Lost (运输) is calculated by extracting the vending machine codes that need replenishment from the replenishment matrix R, and further obtaining the geographical locations of the vending machines that need replenishment; path planning is performed on the geographical locations of the vending machines that need replenishment according to the genetic algorithm to obtain the shortest transportation path, and the transportation distance D of the transportation path is the transportation loss Lost (运输) .

[0022] In one embodiment, the process of modulating the adjustment coefficients α and β to optimize the replenishment quantity matrix R calculation model according to the replenishment loss R Lost includes: when the out-of-stock loss Lost (缺货) is greater than 1 / 3 of the replenishment loss R Lost , it means that the out-of-stock loss Lost (缺货) accounts for a relatively large proportion, and it is necessary to increase the replenishment quantity to reduce the total loss, that is, α increases and β decreases (if β is 0, then β remains unchanged); when the out-of-stock loss Lost (缺货) is less than 1 / 3 of the replenishment loss R Lost , it means that the out-of-stock loss Lost (缺货) accounts for a relatively small proportion, and the replenishment quantity can be appropriately reduced to reduce the labor loss Lost (人工) and the transportation loss Lost (运输) to reduce the replenishment loss R Lost , that is, α decreases and β increases (if α is 0, then α remains unchanged); when the out-of-stock loss Lost (缺货) is equal to 1 / 3 of the replenishment loss R Lost , it is considered that the proportion of the out-of-stock loss Lost (缺货) reaches the optimal state, and this state needs to be maintained, that is, α remains unchanged and β remains unchanged.

[0023] The advantages of the present invention are as follows: The planning of the goods configuration of all vending machines in the management area is completed only by extracting the historical data of the vending machines and the physical data of the cargo channels. This planning process does not depend on the real-time sensing ability of the vending machines, that is, it does not require the vending machines to access the Internet and transmit information to the remote management platform. Thus, the optimization and upgrade of the goods configuration planning of the old-fashioned vending machines are realized.

[0024] Other features and advantages of the present invention will be described in the following specification, and, in part, will be obvious from the specification, or will be understood by implementing the present invention. The objectives and other advantages of the present invention can be achieved and obtained by the structures specifically pointed out in the specification, claims, and drawings. Brief Description of the Drawings

[0025] The accompanying drawings are used to provide a further understanding of the present invention, and constitute a part of the specification. They are used together with the embodiments of the present invention to explain the present invention, and do not constitute a limitation to the present invention. In the accompanying drawings:

[0026] Figure 1 is a flowchart of a method for configuring goods in a vending machine according to an embodiment of the present invention;

[0027] Figure 2 is a flowchart of calculating the single-day single-category sales volume of a single machine according to an embodiment of the present invention;

[0028] Figure 3 is to calculate the replenishment loss R according to an embodiment of the present invention Lost of the flowchart. Detailed implementation manners

[0029] To make the objectives, technical solutions and advantages of the present invention clearer, the following further describes the present invention in detail with reference to the accompanying Figures 1-3 drawings.

[0030] Figure 1 is a flowchart of a method according to an embodiment of the present invention. The following describes this method with reference to Figure 1 drawings.

[0031] Step S100, in this embodiment, first, the sales volume of each single-category commodity in each vending machine within the replenishment cycle T in the vending machine management area is predicted. The prediction tool used is the single-day single-category sales volume prediction model of the single machine of the present invention. The sales volume algorithm reflected by the single-day single-category commodity sales volume prediction model of the single machine is as Figure 2 shown. According to the single-day single-category sales volume prediction model of the single machine in this embodiment, for example, the sales volume of the i-th category of commodities on the first day within the replenishment cycle T of a certain vending machine is predicted, and this sales volume is denoted as s(1, i).

[0032] First, the historical sales volume data sequence L of the i-th category of commodities of this vending machine is extracted. The historical sales volume data sequence L is a time-stamped sequence of numbers. For old-fashioned vending machines that cannot access the network, the historical sales volume data sequence L can be obtained by reading the memory information of the vending machine on-site.

[0033] Subsequently, it is judged whether the length of the historical sales volume data sequence L is greater than or equal to 365, that is, whether the historical sales volume data sequence L records historical sales data of more than or equal to 365 days.

[0034] If the historical sales volume data sequence L is greater than or equal to 365, that is, it indicates that the historical sales volume data sequence L records sales data of more than 365 days, then the sales volume data of the 365th day before is extracted at this time, and is denoted as S 365 .

[0035] In the previous step, further determine whether the 365th day before is a holiday. If it is a holiday, then s(1,i) = S 365 ×0.6 + S three ×0.2 + S 7 ×0.2, where S three is the triple exponential smoothing prediction value, and S 7 is the average value of the sales volume data of the previous 7 days at the current time point. The prediction calculation method at this time is to make the sales volume on the corresponding date last year dominant, while the triple exponential smoothing prediction value and the average value of the sales volume data of the previous 7 days at the current time point only play a regulatory role. If the 365th day before is not a holiday, then s(1,i) = S 365 ×0.2 + S three ×0.4 + S 7 ×0.4. The triple exponential smoothing prediction value S three retains seasonal information on the basis of double exponential smoothing, enabling it to predict time series with seasonality. Its calculation formula is:

[0036]

[0037] Predict the value x of the future replenishment cycle T t+T The calculation formula is:

[0038]

[0039] where are the first, second, and third exponential smoothing values of the t-th cycle respectively, are the first, second, and third exponential smoothing values of the (t - 1)-th cycle respectively, and x t is the actual value of the t-th cycle, and μ is the smoothing coefficient.

[0040] If the historical sales volume data sequence L is less than 365, then further determine whether the historical sales volume data sequence L is greater than or equal to 7, that is, whether the historical sales volume data sequence L records historical sales data of greater than or equal to 7 days. If the historical sales volume data sequence L is greater than or equal to 7, then extract all the sales volumes of the historical sales volume data sequence L, and let s(1,i) = S three ×0.5 + S 7 ×0.5. If the historical sales volume data sequence L is less than 7, then temporarily abandon the prediction of s(1,i) for this replenishment prediction, and wait until the next replenishment prediction to re - determine whether to predict it according to the length of its historical sales volume data sequence L.

[0041] Output s(1, i), and at the same time add s(1, i) to the historical sales data sequence L to update the historical sales data sequence L. Predict s(2, i) based on the updated historical sales data sequence L until T rounds of prediction calculations are performed according to the replenishment cycle T to obtain the predicted values s(1, i), s(2, i), ……, s(T, i) for the i-th category of goods.

[0042] Sum up the above s(1, i), s(2, i), ……, s(T, i) to obtain the total predicted sales volume S of the i-th category of goods in the replenishment cycle for this vending machine. Total (i), if the vending machine number is the j-th vending machine, then the total predicted sales volume of the i-th category of goods for this vending machine is denoted as S Total (i, j).

[0043] Step S200, if the total number of vending machines in the management area is N, and the number of categories of goods sold by all these vending machines is M, then a sales volume prediction matrix S of M×N dimensions for all goods of all vending machines in the management area can be constructed. Denote it as:

[0044]

[0045] Step S300, input the sales volume prediction matrix S into the linear replenishment model to obtain the replenishment quantity matrix R within the replenishment cycle T. The linear replenishment model is to make the replenishment quantity matrix R = α×S - β×Y, where Y is an M×N-dimensional commodity remaining quantity matrix. α and β are adjustment coefficients, and their specific values need to be further optimized in the subsequent optimization training of the linear replenishment model.

[0046] Compare the replenishment quantity matrix R with the maximum storage quantity matrix Q of the vending machine. If the element value in the replenishment quantity matrix R is greater than the corresponding element value in the maximum storage quantity matrix Q, then modify the element value in the replenishment quantity matrix R to the element value in the maximum storage quantity matrix Q, so as to achieve the correction of the replenishment quantity matrix R. The element values in the maximum storage quantity matrix Q represent the maximum storage quantity of a certain category of goods in each vending machine, and they are inherent parameters of the vending machine. Under the condition that the vending machine channel configuration information can be obtained, they can be directly obtained from the vending machine channel configuration information. However, for old-fashioned vending machines, their maximum storage quantity data and commodity remaining quantity data cannot be obtained accurately in real time. In this embodiment, the calculation methods for the commodity remaining quantity matrix Y and the maximum storage quantity matrix Q are as follows:

[0047] When finding the maximum storage matrix Q, the data used is the out-of-stock time of a single type of commodity in the vending machine. When solving the maximum storage of a single type of commodity in a single vending machine, the mode of the total sales volume between two out-of-stock time points of the vending machine is the maximum storage. At the same time, the upper limit value of the maximum storage can be set according to experience. For example, in this embodiment, the upper limit of the maximum storage is set to 30. If the mode of the total sales volume between the two out-of-stock time points calculated above is greater than 30, the maximum storage of this type of commodity in the vending machine is set to 30.

[0048] When finding the remaining quantity of the commodity, first find the time point closest to the current time among the out-of-stock times according to the current time. Starting from the closest out-of-stock time point, calculate the remaining quantity of the commodity at the current time based on the maximum storage and the daily sales volume of the commodity.

[0049] Integrate the maximum storage of each type of commodity for each vending machine obtained above to obtain the maximum storage matrix Q of all types of commodities of all vending machines in the management area, that is:

[0050]

[0051] Integrate the remaining commodity quantities of each type of commodity for each vending machine obtained above to obtain the commodity remaining quantity matrix Y of each type of commodity of all vending machines in the management area, that is:

[0052]

[0053] Step S400, perform a replenishment operation on the vending machines in the management area according to the replenishment quantity matrix R obtained in the previous step, and calculate the replenishment loss R according to the actual replenishment operation structure Lost . The replenishment loss R Lost is calculated as Figure 3 shown. In this embodiment, the replenishment loss R Lost consists of three parts, namely the labor loss Lost (人工) , the out-of-stock loss Lost (缺货) and the transportation loss Lost (运输) .

[0054] The calculation method of the labor loss Lost (人工) is to sum all the elements of the replenishment matrix R to obtain the total replenishment quantity. Since the total replenishment quantity and the cost of manual real-time replenishment are in a linear positive proportional relationship, this total replenishment quantity is recorded as the labor loss Lost (人工) .

[0055] The calculation method of the out-of-stock loss Lost (缺货) is:

[0056] First, obtain the out-of-stock quantity matrix X based on the sales volume matrix S, replenishment matrix R, and remaining product matrix Y, i.e., X = S - R - Y.

[0057] Subsequently, multiply the out-of-stock matrix X by the product price matrix P, and sum all elements of the multiplied matrix to obtain the profit loss caused by out-of-stock. This profit loss is denoted as the out-of-stock loss Lost. (缺货) 。

[0058] The calculation method of the transportation loss Lost (运输) is as follows:

[0059] First, extract the vending machine codes that need to be replenished according to the replenishment matrix R, and further obtain the geographical locations of the vending machines that need to be replenished. That is, all non-zero element values in the replenishment matrix R represent the vending machines that generate product transportation costs.

[0060] Subsequently, perform path planning on the geographical locations of the above-mentioned vending machines that need to be replenished according to the genetic algorithm to obtain the shortest transportation path. The transportation distance D of the transportation path is the transportation loss Lost. (运输) 。

[0061] After obtaining the manual loss Lost (人工) , out-of-stock loss Lost (缺货) , and transportation loss Lost (运输) , let the replenishment loss R Lost = a × Lost (人工) + b × Lost (缺货) + c × Lost (运输) . a, b, and c are the proportionality coefficients of the manual loss Lost (人工) , out-of-stock loss Lost (缺货) , and transportation loss Lost (运输) respectively. The proportionality coefficients a, b, and c are used to adjust the problem of large differences in magnitude caused by different units among the three losses, so as to ensure that the proportions of the three losses in the total loss are relatively close.

[0062] Step S500. In step S300, the replenishment quantity matrix R = α × S - β × Y has been defined, that is, the replenishment quantity depends on the relationship between the product sales volume and the remaining product quantity within the replenishment cycle T. This relationship is determined by the two parameters α and β. Therefore, in each round of optimization operation of the replenishment quantity matrix R, it is necessary to adjust the two parameters α and β according to the proportion relationship of the out-of-stock loss Lost (缺货) in the replenishment loss R Lost . When the out-of-stock loss Lost (缺货) is greater than 1 / 3 of the replenishment loss R Lost , it indicates that the out-of-stock loss Lost (缺货)If the proportion of the shortage loss is relatively large, it is necessary to increase the replenishment quantity to reduce the total loss, that is, α increases and β decreases (if β is 0, then β remains unchanged); when the shortage loss Lost (缺货) is less than 1 / 3 of the replenishment loss R Lost , it means that the proportion of the shortage loss Lost (缺货) is relatively small, and the replenishment quantity can be appropriately reduced to reduce the labor loss Lost (人工) and the transportation loss Lost (运输) to reduce the replenishment loss R Lost , that is, α decreases and β increases (if α is 0, then α remains unchanged); when the shortage loss Lost (缺货) is equal to 1 / 3 of the replenishment loss R Lost , it is considered that the proportion of the shortage loss Lost (缺货) reaches the optimal state, and this state needs to be maintained, that is, α remains unchanged and β remains unchanged.

[0063] Step S600: Use the optimized linear replenishment quantity model to replace the original linear replenishment quantity model, and predict the replenishment quantity for the next replenishment cycle T.

[0064] As described above, only specific implementation cases of the present invention are provided. The protection scope of the present invention is not limited thereto. Any modification or replacement of the present invention by those skilled in the art within the technical specifications described in the present invention shall fall within the protection scope of the present invention.

Claims

1. An automatic vending machine goods configuration method based on artificial intelligence, characterized in that, the method includes: Step S1, obtaining the historical sales data and channel configuration information of each vending machine in the vending machine management area; Step S2, predict the daily sales volume s of each single-category product in each vending machine within the replenishment cycle in the vending machine management area, and sum up the daily sales volume s to obtain the total sales volume S of each single-category product in each vending machine within the replenishment cycle Total ; Step S3, integrating the total sales volume data of all types of goods of all vending machines in the vending machine management area during the replenishment cycle into an M×N-dimensional sales volume prediction matrix S; where M is the number of commodity categories involved in the sales of all vending machines, and N is the number of all vending machines; Step S4, calculating the maximum storage matrix Q and the commodity remaining matrix Y of all vending machines in the vending machine management area according to the commodity emptying time and historical sales data of each vending machine in the vending machine management area; Step S5, constructing a replenishment quantity matrix R calculation model for all vending machines involved in the vending machine management area, that is, replenishment quantity matrix R = α×sales volume prediction matrix S - β×commodity remaining matrix Y; where α and β are adjustment coefficients; Step S6: Conduct actual replenishment operations on the vending machines within the vending machine management area according to the replenishment quantity matrix R, and calculate the replenishment loss R based on the actual replenishment situation Lost ; Step S7, according to the replenishment loss R Lost Modulate the adjustment coefficients α and β to optimize the replenishment quantity matrix R calculation model; and use the optimized replenishment quantity matrix R calculation model to predict the replenishment quantity for the next replenishment cycle; The process of predicting the daily sales volume s of each single-category product in each vending machine within the replenishment cycle in the vending machine management area includes: extracting the historical sales volume data sequence L of the single-category product of a single vending machine. If the historical sales volume data sequence L is greater than or equal to 365, it means that the historical sales volume data sequence L records sales data for more than 365 days. At this time, the sales volume data on the 365th day before is extracted and denoted as S 365 ; Further, whether the 365th day before is a holiday. If it is a holiday, then the sales volume s = S 365 ×0.6 + S three ×0.2 + S 7 ×0.2; If the 365th day before is not a holiday, then the sales volume s = S 365 ×0.2 + S three ×0.4 + S 7 ×0.4; Where S three is the triple exponential smoothing prediction value, and S 7 is the average value of the sales volume data in the 7 days before the current time point; Replenishment loss R Lost including labor loss Lost (人工) , out-of-stock loss Lost (缺货) and transportation loss Lost (运输) ; and the replenishment loss R Lost = a × Lost (人工) + b × Lost (缺货)+ c × Lost (运输) ; where a, b, and c are the proportionality coefficients of labor loss Lost (人工) , out-of-stock loss Lost (缺货) and transportation loss Lost (运输) ; the proportionality coefficients a, b, and c are used to adjust the order-of-magnitude differences caused by different units among the three losses.

2. The automatic vending machine goods configuration method according to claim 1, characterized in that, If the historical sales volume data sequence L is less than 365, further determine whether the historical sales volume data sequence L is greater than or equal to 7, that is, whether the historical sales volume data sequence L records historical sales data for 7 days or more; if the historical sales volume data sequence L is greater than or equal to 7, extract all the sales volumes of the historical sales volume data sequence L, and let the sales volume s = S three ×0.5 + S 7 ×0.5。 3. The automatic vending machine goods configuration method according to claim 1, characterized in that, The calculation process of the maximum storage matrix Q includes: calculating the mode of the total sales volume of each type of commodity between two emptying time points of each vending machine as the maximum storage volume of this type of commodity of this vending machine; integrating the maximum storage volume of each type of commodity of each vending machine obtained to obtain the maximum storage matrix Q of all types of commodities of all vending machines in the management area.

4. The automatic vending machine goods configuration method according to claim 3, characterized in that, The calculation process of the commodity remaining matrix Y includes: finding the time point closest to the current time among the emptying times according to the current time, and starting from the closest emptying time point, calculating the remaining quantity of each type of commodity of each vending machine at the current time based on the maximum storage volume and the daily sales volume of this commodity; integrating the remaining quantity of each type of commodity of each vending machine obtained above to obtain the commodity remaining matrix Y of all types of commodities of all vending machines in the management area.

5. The automatic vending machine goods configuration method according to claim 1, characterized in that, The artificial loss Lost (人工) is calculated by summing all elements of the replenishment matrix R to obtain the total replenishment quantity, which is denoted as the artificial loss Lost (人工) .

6. The automatic vending machine goods configuration method according to claim 1, characterized in that, The out-of-stock loss Lost (缺货) is calculated by obtaining the out-of-stock quantity matrix X based on the sales volume matrix S, the replenishment quantity matrix R, and the remaining quantity matrix Y of goods, that is, the out-of-stock quantity matrix X = the sales volume matrix S - the replenishment quantity matrix R - the remaining quantity matrix Y of goods; multiplying the out-of-stock matrix X by the commodity price matrix P, and summing all the elements of the multiplied matrix to obtain the profit loss caused by out-of-stock; this profit loss is recorded as the out-of-stock loss Lost (缺货) .

7. The automatic vending machine goods configuration method according to claim 1, characterized in that, The transportation loss Lost (运输) is calculated by extracting the vending machine codes that need replenishment from the replenishment matrix R, and further obtaining the geographical locations of the vending machines that need replenishment; performing path planning on the geographical locations of the vending machines that need replenishment according to the genetic algorithm to obtain the shortest transportation path, and the transportation distance D of the transportation path is the transportation loss Lost (运输) .

8. The automatic vending machine goods configuration method according to claim 1, characterized in that, Said according to the replenishment loss R Lost The process of modulating the adjustment coefficients α and β to optimize the replenishment quantity matrix R calculation model includes: when the out-of-stock loss Lost (缺货) is greater than 1 / 3 of the replenishment loss R Lost , it means that the out-of-stock loss Lost (缺货) accounts for a relatively large proportion, and it is necessary to increase the replenishment quantity to reduce the total loss, that is, α increases and β decreases. If β is 0, then β remains unchanged; when the out-of-stock loss Lost (缺货) is less than 1 / 3 of the replenishment loss R Lost , it means that the out-of-stock loss Lost (缺货) accounts for a relatively small proportion, and the replenishment quantity can be appropriately reduced to reduce the labor loss Lost (人工) and the transportation loss Lost (运输) to reduce the replenishment loss RLost, that is, α decreases and β increases. If α is 0, then α remains unchanged; when the out-of-stock loss Lost (缺货) is equal to 1 / 3 of the replenishment loss R Lost , it is considered that the out-of-stock loss Lost (缺货) accounts for the best state, and this state needs to be maintained, that is, α remains unchanged and β remains unchanged.

Images