Method and device for determining buffer inventory quantity data in a logistics supply chain

Abstract

The application provides a method and device for determining buffer inventory quantity data in a logistics supply chain, which can improve the prediction accuracy of cargo demand quantity prediction results to a certain extent. The method comprises the following steps: obtaining historical cargo demand quantity data of a target node, wherein the target node is a node in the logistics supply chain, and the historical cargo demand quantity data is used to indicate the demand quantity of the target node for a target cargo in a historical time period; inputting the historical cargo demand quantity data into a cargo demand quantity prediction model to obtain cargo demand quantity prediction data, wherein the cargo demand quantity prediction model is obtained by training an initial neural network model using cargo demand quantity data in different time periods, and the cargo demand quantity data in different time periods is labeled with distribution types of different cargos in different time periods and different nodes; and determining buffer inventory quantity data of the target node for the target cargo according to the cargo demand quantity prediction data and current cargo inventory quantity data.

Description

Technical Field

[0001] This application relates to the field of logistics, specifically to a method and apparatus for determining the quantity data of buffer inventory in a logistics supply chain. Background Technology

[0002] In the logistics sector, with the continuous improvement and expansion of logistics services, logistics companies can not only provide traditional logistics services such as transporting goods, but also play an upstream role in the supply chain. Logistics companies can provide corresponding goods to downstream merchants through their own logistics services, such as providing the retail goods needed by logistics outlets, supermarkets, convenience stores and other stores.

[0003] In terms of the supply chain, inventory can be considered from the demand side to facilitate stock preparation. In the process, in addition to traditional methods that rely on manual forecasting, demand and inventory can also be predicted through big data analysis. Historical demand for goods can be used as time-series data for forecasting via time-series models.

[0004] In existing research on related technologies, the inventors found that the demand forecasts given by time series models are unstable and often do not match the actual situation. The inventory deployed based on the demand forecasts given by time series models is prone to two situations: too much inventory or supply falling short of demand. Obviously, the accuracy of its demand forecasts is low. Summary of the Invention

[0005] This application provides a method and apparatus for determining the quantity data of buffer inventory in a logistics supply chain, which can improve the prediction accuracy of the quantity forecast of goods demand to a certain extent, thereby improving the accuracy of buffer inventory.

[0006] Firstly, this application provides a method for determining the quantity data of buffer inventory in a logistics supply chain, the method comprising:

[0007] Obtain historical cargo demand data for the target node, where the target node is a node in the logistics supply chain, and the historical cargo demand data is used to indicate the demand for the target cargo at the target node during a historical period.

[0008] Historical cargo demand data is input into the cargo demand prediction model to predict the demand for target cargo at the target node in the current time period and obtain cargo demand prediction data. The cargo demand prediction model is trained on the initial neural network model using cargo demand data from different time periods. The cargo demand data from different time periods are labeled with the distribution type of different cargo at different nodes in different time periods.

[0009] Based on the forecast data of the quantity of goods demand and the current quantity of goods inventory, the buffer inventory quantity data for the target node for the target goods is determined. The current quantity of goods inventory indicates the quantity of goods in stock at the target node in the current time period, and the buffer inventory quantity data indicates the quantity of reserve inventory that is higher than the quantity of inventory corresponding to the current quantity of goods inventory.

[0010] In conjunction with the first aspect of this application, in a first possible implementation of the first aspect of this application, obtaining the historical cargo demand quantity data of the target node includes:

[0011] Acquire multiple cargo images of the target node over a historical time period. The cargo images contain images of shelf layers where goods are placed, and the cargo images are taken on-site by the target node.

[0012] The goods in the image are identified to obtain goods data, which includes the category and quantity of the identified goods.

[0013] Confirm historical demand data for goods based on commodity data.

[0014] In conjunction with the first possible implementation of the first aspect of this application, in the second possible implementation of the first aspect of this application, the goods image is marked with a date, and the commodity data includes multiple sets of sub-commodity data labeled with days as the date unit. The historical goods demand quantity data confirmed based on the commodity data includes:

[0015] Extract target sub-product data labeled with target dates from the product data;

[0016] Based on the target sub-product data, determine the daily inventory quantity and daily sales volume of the target goods at the target node;

[0017] Based on the daily inventory and sales data, determine the historical demand data for goods.

[0018] In conjunction with the second possible implementation of the first aspect of this application, in the third possible implementation of the first aspect of this application, determining the historical demand quantity data based on the daily inventory quantity data and the daily sales data includes:

[0019] Based on the node identifier of the target node, determine the location type of the target node in the node network of the logistics supply chain, where the location type includes end node type or non-end node type;

[0020] Based on location type, daily inventory data, and daily sales data, determine historical demand data for goods.

[0021] In conjunction with the third possible implementation of the first aspect of this application, in the fourth possible implementation of the first aspect of this application, determining the historical goods demand quantity data based on location type, daily inventory quantity data, and daily sales data includes:

[0022] When the location type is end node type, determine whether the amount of the daily inventory quantity data is greater than zero;

[0023] If the value is greater than zero, the sales data for that day will be determined as the historical demand data for goods at the target node on the target day.

[0024] If the value is zero, then based on the daily inventory quantity data and the daily sales data, extract the calibration day that has the same preset characteristics as the target node and the target day, where the daily inventory quantity data of the calibration day is greater than zero.

[0025] The average sales volume of historical goods demand data on the calibration date is used as the historical goods demand data for the target node on the target date.

[0026] In conjunction with the third possible implementation of the first aspect of this application, in the fifth possible implementation of the first aspect of this application, determining the historical goods demand quantity data based on location type, daily inventory quantity data, and daily sales data includes:

[0027] When the location type is a non-end node type, determine the downstream node with a downstream identifier, where the downstream identifier is used to identify the node that is downstream of the target node in the architecture of the logistics supply chain.

[0028] Collect order quantity data from downstream nodes on the target day, and determine the historical demand data for goods at the target node on the target day based on the order quantity data, daily inventory data, and daily sales data.

[0029] In conjunction with the fourth or fifth possible implementation of the first aspect of this application, in the sixth possible implementation of the first aspect of this application, the distribution type includes the normal distribution type, gamma distribution type, Poisson distribution type and negative binomial distribution type that respectively satisfy the preset fluctuation type. The preset fluctuation type is obtained by removing three fluctuation types from the fluctuation type set, namely, extreme low-frequency fluctuation type, extreme high-frequency fluctuation type and extreme small fluctuation type, based on the proportion of zero values ​​and the amplitude of fluctuation. The preset fluctuation type includes high-frequency stable fluctuation type, low-frequency stable fluctuation type, high-frequency fluctuation type and low-frequency fluctuation type.

[0030] Secondly, this application provides an apparatus for determining the quantity data of buffer inventory in a logistics supply chain, the apparatus comprising:

[0031] The receiving and sending unit is used to obtain historical cargo demand quantity data of the target node, where the target node is a node in the logistics supply chain, and the historical cargo demand quantity data is used to indicate the demand quantity of the target node for the target cargo in a historical period.

[0032] The processing unit is used to input historical goods demand quantity data into the goods demand quantity prediction model to predict the demand quantity of the target node for the target goods in the current time period and obtain goods demand quantity prediction data. The goods demand quantity prediction model is obtained by training an initial neural network model with goods demand quantity data from different time periods. The goods demand quantity data from different time periods are labeled with the distribution type of different goods at different nodes in different time periods. Based on the goods demand quantity prediction data and the current goods inventory quantity data, the buffer inventory quantity data for the target node for the target goods is determined. The current goods inventory quantity data is used to indicate the inventory quantity of the target node for the target goods in the current time period, and the buffer inventory data is used to indicate the reserve inventory quantity that is higher than the inventory amount corresponding to the current goods inventory quantity data.

[0033] In conjunction with the second aspect of this application, in a first possible implementation of the second aspect of this application, the obtaining unit is specifically used for:

[0034] Acquire multiple cargo images of the target node over a historical time period. The cargo images contain images of shelf layers where goods are placed, and the cargo images are taken on-site by the target node.

[0035] The processing unit is specifically used to identify the goods contained in the goods image to obtain goods data, wherein the goods data includes the goods category and quantity of the identified goods; and to confirm the historical goods demand quantity data based on the goods data.

[0036] In conjunction with the first possible implementation of the second aspect of this application, in the second possible implementation of the second aspect of this application, the goods image is identified by a date, the commodity data includes multiple sets of sub-commodity data labeled with days as the date unit, and the processing unit is specifically used for:

[0037] Extract target sub-product data labeled with target dates from the product data;

[0038] Based on the target sub-product data, determine the daily inventory quantity and daily sales volume of the target goods at the target node;

[0039] Based on the daily inventory and sales data, determine the historical demand data for goods.

[0040] In conjunction with the second possible implementation of the second aspect of this application, in the third possible implementation of the second aspect of this application, the processing unit is specifically used for:

[0041] Based on the node identifier of the target node, determine the location type of the target node in the node network of the logistics supply chain, where the location type includes end node type or non-end node type;

[0042] Based on location type, daily inventory data, and daily sales data, determine historical demand data for goods.

[0043] In conjunction with the third possible implementation of the second aspect of this application, in the fourth possible implementation of the second aspect of this application, the processing unit is specifically used for:

[0044] When the location type is end node type, determine whether the amount of the daily inventory quantity data is greater than zero;

[0045] If the value is greater than zero, the sales data for that day will be determined as the historical demand data for goods at the target node on the target day.

[0046] If the value is zero, then based on the daily inventory quantity data and the daily sales data, extract the calibration day that has the same preset characteristics as the target node and the target day, where the daily inventory quantity data of the calibration day is greater than zero.

[0047] The average sales volume of historical goods demand data on the calibration date is used as the historical goods demand data for the target node on the target date.

[0048] In conjunction with the third possible implementation of the second aspect of this application, in the fifth possible implementation of the second aspect of this application, the processing unit is specifically used for:

[0049] When the location type is a non-end node type, determine the downstream node with a downstream identifier, where the downstream identifier is used to identify the node that is downstream of the target node in the architecture of the logistics supply chain.

[0050] Collect order quantity data from downstream nodes on the target day, and determine the historical demand data for goods at the target node on the target day based on the order quantity data, daily inventory data, and daily sales data.

[0051] In conjunction with the fourth or fifth possible implementation of the second aspect of this application, in the sixth possible implementation of the second aspect of this application, the distribution type includes the normal distribution type, gamma distribution type, Poisson distribution type and negative binomial distribution type that respectively satisfy the preset fluctuation type. The preset fluctuation type is obtained by removing three fluctuation types from the fluctuation type set, namely, extreme low-frequency fluctuation type, extreme high-frequency fluctuation type and extreme small fluctuation type, based on the proportion of zero values ​​and the amplitude of fluctuation. The preset fluctuation type includes high-frequency stable fluctuation type, low-frequency stable fluctuation type, high-frequency fluctuation type and low-frequency fluctuation type.

[0052] Thirdly, this application also provides a device for determining buffer inventory quantity data in a logistics supply chain, including a processor and a memory, wherein the memory stores a computer program, and the processor executes the steps of any of the methods provided in the first aspect of this application when it calls the computer program in the memory.

[0053] Fourthly, this application also provides a computer-readable storage medium storing a plurality of instructions adapted for loading by a processor to perform the steps of any of the methods provided in the first aspect of this application.

[0054] From the above, it can be concluded that this application has the following beneficial effects:

[0055] This application trains a goods demand quantity prediction model. This model is obtained by training an initial neural network model with goods demand quantity data from different time periods. Since these goods demand quantity data are labeled with the distribution type of different goods at different time periods and nodes, the trained model can focus on the distribution type of the goods demand quantity of the target node in different time periods when inputting the historical goods demand quantity data of the target node and making goods demand quantity prediction. It can also determine the specific distribution type corresponding to the target node, thereby more accurately predicting the demand quantity of the target node for the target goods in the current time period based on the distribution type.

[0056] Secondly, since more accurate cargo demand quantity forecast data is obtained, this application can determine the corresponding buffer inventory quantity data based on the cargo demand quantity forecast data and the current cargo inventory quantity data of the target node. The current cargo inventory quantity data is used to indicate the inventory quantity of the target goods at the target node in the current time period, and the buffer inventory quantity data is used to indicate the reserve inventory quantity that is higher than the inventory amount corresponding to the current cargo inventory quantity data. Thus, when the target node prepares goods for the current time period, it can obtain more accurate and stable preparation results. As a supply chain node in the logistics supply chain, the target node can work and play a more stable role in the supply chain operation. Attached Figure Description

[0057] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0058] Figure 1 This is a flowchart illustrating a method for determining the quantity data of buffer inventory in the logistics supply chain of this application.

[0059] Figure 2 A flowchart illustrating the process of obtaining historical cargo demand quantity data for this application;

[0060] Figure 3 Another flowchart illustrating the process for determining historical cargo demand data for this application;

[0061] Figure 4 Another flowchart illustrating the process for determining historical cargo demand data for this application;

[0062] Figure 5 Another flowchart illustrating the process for determining historical cargo demand data for this application;

[0063] Figure 6 This is a schematic diagram of a scenario for the fluctuation classification processing in this application;

[0064] Figure 7 This is a schematic diagram of a device for determining the quantity data of buffer inventory in the logistics supply chain of this application.

[0065] Figure 8 This is a schematic diagram of a device for determining the quantity of buffer inventory in the logistics supply chain in this application. Detailed Implementation

[0066] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0067] In the following description, specific embodiments of this application will be illustrated with reference to steps and symbols performed by one or more computers, unless otherwise stated. Therefore, these steps and operations will be referred to several times as being performed by a computer, and computer execution as referred to herein includes operations by a computer processing unit representing electronic signals of data in a structured format. This operation transforms the data or maintains it at a location in the computer's memory system, which can be reconfigured or otherwise alter the operation of the computer in a manner well known to those skilled in the art. The data structure maintained by the data is the physical location of the memory, which has specific characteristics defined by the data format. However, the principles of this application are described in the foregoing text, which is not intended to be limiting, and those skilled in the art will understand that many of the steps and operations described below can also be implemented in hardware.

[0068] The principles of this application are based on many other general-purpose or purpose-specific computing, communication environments, or configurations. Examples of computing systems, environments, and configurations well known to be suitable for use in this application may include (but are not limited to) handheld phones, personal computers, servers, multiprocessor systems, microcomputer-based systems, mainframe computers, and distributed computing environments, including any of the aforementioned systems or devices.

[0069] The terms “first,” “second,” and “third,” etc., used in this application are used to distinguish different objects, not to describe a specific order. Furthermore, the terms “comprising” and “having,” and any variations thereof, are intended to cover non-exclusive inclusion.

[0070] First, before introducing this application, let me first introduce the relevant content regarding the application background.

[0071] The method, apparatus, and computer-readable storage medium for determining buffer inventory quantity data in the logistics supply chain provided in this application can be applied to devices for determining buffer inventory quantity data, thereby improving the prediction accuracy of goods demand quantity forecasting results to a certain extent, and thus improving the accuracy of buffer inventory.

[0072] In this application, the device for determining the quantity of buffer inventory in the logistics supply chain can be understood as a hardware device with data processing capabilities, such as a server device, physical host, or user equipment (UE). Specifically, the UE can be a terminal device such as a smartphone, tablet computer, laptop computer, handheld computer, desktop computer, or personal digital assistant (PDA).

[0073] Logistics services can specifically include express delivery services. A logistics supply chain is a supply chain that a logistics company implements based on its logistics services. A logistics supply chain consists of different supply chain nodes, such as logistics nodes of the logistics company itself used to transport goods in the logistics supply chain, third-party nodes that join the logistics supply chain used to transport goods in the logistics supply chain, logistics nodes of the logistics company itself used to sell goods in the logistics supply chain, and third-party nodes that join the logistics supply chain used to sell goods in the logistics supply chain.

[0074] The following section introduces the method for determining the quantity of buffer inventory in the logistics supply chain provided in this application.

[0075] First, refer to Figure 1 , Figure 1 This application illustrates a flowchart of a method for determining the quantity of buffer inventory in the logistics supply chain. Specifically, the method for determining the quantity of buffer inventory in the logistics supply chain may include:

[0076] Step S101: Obtain historical cargo demand quantity data for the target node, where the target node is a node in the logistics supply chain, and the historical cargo demand quantity data is used to indicate the demand quantity of the target node for the target cargo during a historical period.

[0077] Step S102: Input historical cargo demand quantity data into the cargo demand quantity prediction model to predict the demand quantity of the target node for the target cargo in the current time period and obtain cargo demand quantity prediction data. The cargo demand quantity prediction model is obtained by training the initial neural network model with cargo demand quantity data of different time periods. The cargo demand quantity data of different time periods are labeled with the distribution type of different cargoes at different nodes in different time periods.

[0078] Step S103: Based on the forecast data of the quantity of goods demand and the current quantity of goods inventory, determine the buffer inventory quantity data of the target node for the target goods. The current quantity of goods inventory is used to indicate the inventory quantity of the target node for the target goods in the current time period, and the buffer inventory quantity data is the reserve inventory quantity that is higher than the inventory amount corresponding to the current quantity of goods inventory.

[0079] From the above Figure 1As can be seen from the embodiments shown, this application trains a goods demand quantity prediction model. This model is obtained by training an initial neural network model with goods demand quantity data from different time periods. Since these goods demand quantity data are labeled with the distribution type of different goods at different time periods and nodes, the trained model can focus on the distribution type of the goods demand quantity of the target node in different time periods when inputting the historical goods demand quantity data of the target node and making goods demand quantity predictions. It can also determine the specific distribution type corresponding to the target node, thereby more accurately predicting the demand quantity of the target node for the target goods in the current time period based on the distribution type.

[0080] Secondly, since more accurate cargo demand quantity forecast data is obtained, this application can determine the corresponding buffer inventory quantity data based on the cargo demand quantity forecast data and the current cargo inventory quantity data of the target node. The current cargo inventory quantity data is used to indicate the inventory quantity of the target goods at the target node in the current time period, and the buffer inventory quantity data is used to indicate the reserve inventory quantity that is higher than the inventory amount corresponding to the current cargo inventory quantity data. Thus, when the target node prepares goods for the current time period, it can obtain more accurate and stable preparation results. As a supply chain node in the logistics supply chain, the target node can work and play a more stable role in the supply chain operation.

[0081] The following continues with... Figure 1 The specific implementation methods of each step in the illustrated embodiment are described in detail below:

[0082] In this application, the prediction of the quantity of goods demand for the target node can be triggered by staff on the device that determines the quantity of buffer inventory in the logistics supply chain, or it can be triggered by accessing the device that determines the quantity of buffer inventory in the logistics supply chain through devices such as UE, or it can be automatically triggered by the device that determines the quantity of buffer inventory in the logistics supply chain according to a preset strategy. The specific triggering conditions can be adjusted as needed and are not limited here.

[0083] For example, an employee of courier company A can access the server on their smartphone through the X mini-program in application B and initiate a request to predict the quantity of goods demand for a target node. The server then triggers a prediction of the quantity of goods demand for the target node based on the request. Another example is that the server automatically predicts the quantity of goods demand for each node in the logistics supply chain on a daily basis, based on the nodes identified in the node list of the logistics supply chain.

[0084] Historical demand data for goods is used to indicate the demand for corresponding goods by a given node during a historical period. The historical period and the corresponding goods can be determined according to actual needs. For example, the historical period can be a time span of the past year, the past quarter, the past month, or the past week. The goods can be all the goods involved by the node, the goods forwarded by the node, or the hot-selling products of the node in the current season.

[0085] This historical data on the quantity of goods demand can be obtained primarily through the following methods:

[0086] The first type is configured by staff.

[0087] It is understood that this historical goods demand quantity data can be manually configured by relevant personnel. Specifically, personnel can manually configure the required historical goods demand quantity data for the method of determining the buffer inventory quantity data in the logistics supply chain of this application; or, personnel can also configure relevant interfaces on the devices in this application that need to store historical goods demand quantity data. When the device for determining the buffer inventory quantity data in the logistics supply chain of this application triggers the method of determining the buffer inventory quantity data in the logistics supply chain, it can retrieve the historical goods demand quantity data configured by the personnel through these devices.

[0088] The second method involves processing buffer inventory quantity data using equipment.

[0089] It is understandable that the demand for goods can be obtained through paper questionnaires or online questionnaires sent to the UE, with staff or customers at the node filling in the demand information; alternatively, relevant supply chain order data, such as relevant purchase records, shipment records, sales records, and production records, can be retrieved to determine the quantity of goods required based on the actual order data.

[0090] In one exemplary implementation, see [reference] Figure 2 The diagram shown illustrates a process for obtaining historical cargo demand quantity data according to this application. In this application, the process for obtaining historical cargo demand quantity data may include the following steps S201 to S203:

[0091] Step S201: Obtain multiple cargo images of the target node during a historical time period. The cargo images contain images of shelf layers where goods are placed. The cargo images are taken on-site by the target node.

[0092] In this application, the acquisition of historical cargo demand data can be achieved by combining image recognition methods.

[0093] At the node site, fixed cameras or handheld shooting devices can be deployed to photograph the shelf layers at the node site, thereby monitoring the entry and exit of goods at the node site. Furthermore, due to the use of image recognition, richer and more direct data can be left behind, providing a better reconstruction of the node site.

[0094] Step S202: Identify the goods contained in the goods image to obtain goods data, wherein the goods data includes the goods category and quantity of the identified goods;

[0095] In practical applications, artificial intelligence (AI) technology can be used to identify the image features of products to achieve product recognition.

[0096] Images containing goods, such as product images and images of shelves with goods, can be used as the training set for the model. The images in the training set are labeled with the products contained in their content. After obtaining the training set, the images in the training set are sequentially input into the initial neural network model for forward propagation. Then, the loss function is calculated based on the product recognition results output by the neural network model. Backpropagation is then performed based on the loss function to adjust the parameters of the neural network model. Through multiple forward and backpropagation operations, the goal of training the neural network model is achieved. When the requirements for the number of training iterations and product recognition accuracy are met, the trained neural network model is the product recognition model. This model can be used in practical applications to identify the products contained in images of goods.

[0097] The product identification results, i.e., product data, may include the product category and the corresponding quantity of the product identified from the product image.

[0098] Different products can be distinguished according to preset product identifiers. For example, product A is identified by identifier 1, and the quantity of product A is represented by 1*5. Product C is identified by identifier 6.

[0099] Furthermore, product data can also be presented as a grid diagram. As we understand it, the placement of products on shelves is usually quite regular. Based on the image position of each identified product and the space it occupies, a one-to-one grid can be generated, with each grid labeled with a corresponding product identifier. This yields product data with richer spatial data, facilitating related data processing.

[0100] Step S203: Confirm the historical demand quantity data for goods based on the commodity data.

[0101] Once the identified product data is obtained, it can be confirmed as the quantity of goods required in this application. By using image recognition, the quantity of goods required can be determined based on the actual order data. Since the actual order data is determined by combining image monitoring, it not only provides rich image data support, but also accurately restores the actual goods entry and exit situation at the target node. Using this product data as historical quantity of goods required, it can fully reflect the true quantity of demand when there is no supply shortage.

[0102] Of course, considering the possibility of supply falling short of demand in practical applications, the commodity data can be appropriately amplified by combining empirical adjustment coefficients after obtaining the commodity data, so that the historical commodity demand data can include the unmet commodity demand when there is a supply shortage.

[0103] Taking the image of goods as actual order data as an example, the image of goods can be marked with a date, and the processed product data can include multiple sets of sub-product data labeled with days as the date unit.

[0104] In another exemplary implementation, the process of determining historical goods demand quantity data after obtaining goods data from goods images through image recognition is as follows: Figure 3 The illustrated flowchart of this application for determining historical cargo demand quantity data may specifically include:

[0105] Step S301: Extract target sub-product data marked with the target date from the product data;

[0106] First, the product data identified earlier contains a preset field with a corresponding date. Based on the date identified by this preset field, sub-product data for a specific day can be filtered out from the product data.

[0107] The specific date, or target date, can be adjusted as needed. For example, individual dates can be manually specified, or each day can be selected as the target date by default, or individual dates can be selected according to other preset selection strategies. No specific restrictions are imposed here.

[0108] Step S302: Based on the target sub-product data, determine the daily inventory quantity data and daily sales data of the target node for the target goods;

[0109] The product data can carry the daily inventory and sales volume in preset fields. Therefore, it is also possible to extract the daily inventory quantity data and daily sales volume data corresponding to the sub-product data of a specific day.

[0110] Step S303: Determine the historical demand data for goods based on the daily inventory quantity data and the daily sales data.

[0111] Once the daily inventory and sales volume are obtained, they can be processed to obtain historical demand data for goods.

[0112] In practical applications, the specific processing strategies for historical cargo demand data can also be configured in conjunction with the network location of nodes in the logistics supply chain.

[0113] For example, in yet another exemplary implementation, determining historical goods demand data based on daily inventory and daily sales includes:

[0114] Based on the node identifier of the target node, determine the location type of the target node in the node network of the logistics supply chain, where the location type includes end node type or non-end node type;

[0115] Based on location type, daily inventory data, and daily sales data, determine historical demand data for goods.

[0116] It is understandable that each node in the logistics supply chain can be configured with a corresponding node identifier to identify its location type. Preferably, nodes can be divided into two types: end nodes and non-end nodes. End nodes are nodes that sell goods to end consumers, while non-end nodes are important carriers of the channel function of goods. By classifying nodes in the logistics supply chain into these two types, the two demand methods of goods, namely "procurement-sales" and "procurement-storage-distribution", can be distinguished, and the corresponding demand quantity data of goods can be determined.

[0117] For example, regarding the processing of historical cargo demand data for end nodes, see [link to relevant documentation]. Figure 4 The illustrated flowchart of another method for determining historical cargo demand data in this application may include:

[0118] Step S401: When the location type is end node type, determine whether the amount of the daily inventory quantity data is greater than zero.

[0119] First, determine whether the daily inventory of the end node is greater than zero.

[0120] It should be understood that, in this application, the daily inventory, or the amount of daily inventory quantity data, may refer to the final inventory of the day, or the inventory within a specified time period of the day.

[0121] Step S402: If the value is greater than zero, the sales data of the day is determined as the historical demand data of goods at the target node on the target day.

[0122] The end node of the supply chain is the node that directly faces customers and end consumers. When there is no stockout at the end node and no order postponement mechanism, the sales volume of the day can be used as the quantity of goods required for the day.

[0123] Step S403: If the value is zero, then based on the daily inventory quantity data and the daily sales data, extract the calibration day that has the same preset characteristics as the target node and the target day, wherein the amount of the daily inventory quantity data of the calibration day is greater than zero.

[0124] If sales fall short of demand, this application may consider a customized supply chain solution to capitalize on the lost sales opportunity.

[0125] In this application, for cases where the demand for goods is not met, the actual quantity of goods required can be restored. The actual quantity of goods required on the target date can be restored by combining a calibration date that has the same preset characteristics as the target date and has not experienced a sales volume less than the demand.

[0126] This equivalent preset feature is used to indicate common characteristics of the target date and other dates in terms of demand for goods, such as similar time points and similar trends in demand for goods based on the daily inventory data and the daily sales data.

[0127] Step S404: Determine the average sales volume of the historical goods demand quantity data of the calibration date as the historical goods demand quantity data of the target node on the target date.

[0128] After identifying benchmark days with the same pre-defined characteristics as the target day, the average sales data of these benchmark days can be used as the historical demand data for goods on the target day.

[0129] For example, for any day d, if the inventory is 0, a stockout may occur, and sales ≤ demand. Demand restoration is then performed.

[0130] a. Find the last sales transaction on date d and record the time t when that sales occurred;

[0131] b. Find the dates within a certain period before date d (which can be adjusted according to the industry to January, March, half a year, etc.) where no stockouts occurred. The set of these dates is S.

[0132] c. From the never-out-of-stock dates S, select dates of the same type as date d (the same type classification rule can be adjusted according to industry characteristics, such as: both being weekdays / Mondays), to obtain the set of never-out-of-stock dates S- that are of the same type as date d;

[0133] d. Calculate the average sales m that occurred after time t in the set of non-out-of-stock dates S;

[0134] e. Demand for date d = Sales + m.

[0135] On the other hand, for the determination and processing of historical cargo demand quantity data for non-end nodes, see, for example, the following... Figure 5 The illustrated flowchart of another method for determining historical cargo demand data in this application may include:

[0136] Step S501: When the location type is a non-end node type, determine the downstream node with a downstream identifier, wherein the downstream identifier is used to identify the node that is downstream of the target node in the architecture of the logistics supply chain.

[0137] It is understandable that non-end-point nodes do not directly face customers; their demand originates from orders from downstream nodes. Downstream nodes typically consider future sales fluctuations when placing orders, leading to over- or under-ordering, resulting in order volume ≠ sales volume. Since non-end-point nodes lack accurate information on downstream sales volume, they also tend to over- or under-order when order volume fluctuates. This artificially adjusted order volume is passed up the chain, leading to large inventories or severe stockouts. This application proposes a supply chain solution to determine the true demand quantity for non-end-point nodes. When the target node is a non-end-point node, downstream nodes can be used to help determine the target node's actual goods demand quantity.

[0138] In this application, in the logistics supply chain, corresponding upstream and downstream identifiers can be configured for each node or for related nodes with upstream and downstream relationships. These identifiers can identify related nodes with upstream and downstream relationships, thereby allowing the determination of the downstream node corresponding to the target node based on the upstream and downstream identifiers.

[0139] Step S502: Collect the order quantity data of downstream nodes on the target day, and determine the historical goods demand data of the target nodes on the target day based on the order quantity data, the inventory quantity data of the day and the sales data of the day.

[0140] After identifying the downstream nodes corresponding to the target node, the historical cargo demand data of the target node can be determined by combining the historical cargo demand data of these downstream nodes.

[0141] For example:

[0142] 1) Divide all nodes into a set S of end nodes based on network relationships and whether they directly face customers. E Set of non-terminal nodes S P ;

[0143] 2)S E The nodes in the S are processed by restoring the actual number of end nodes as described above, and the result is used as S. EThe actual number of nodes in the middle, all S E The node is added to the set S of restored nodes. F middle;

[0144] 3) Take S P A node S P (i), whose downstream node set is Sc p＝i When Sc p＝i All of them already exist in S F When, it means S P (i) All downstream nodes have been restored and can begin processing S. P (i) Restore, the process is as follows:

[0145] a. Regarding S P (i) Downstream node set Sc p＝i For each node in the array, determine its respective direction to S. P (i) the ordering cycle and ordering time;

[0146] b. Sum the demand for downstream nodes within their ordering cycle and consolidate it to the ordering date, as the downstream node's demand for S. P (i) The quantity and timing of demand.

[0147] c. S P (i) Remove S P Join S F middle.

[0148] Repeat step 3) until S F Includes all nodes.

[0149] Once the order quantity data for the target node is determined, the daily inventory quantity data and daily sales data can be compared and corrected to restore the actual demand quantity of goods for the target node, which is not the final node.

[0150] In this application, the predictive processing of goods demand quantity data is achieved through a goods demand quantity prediction model. This model allows for the configuration of relevant historical goods demand quantity data and the labeling of corresponding goods demand quantity data for future time periods. Furthermore, this application also labels the distribution types of different goods at different time periods and nodes for these historical goods demand quantity data. This enables the analysis of goods demand quantity based on distribution types, as described in this application. It can be understood that different distribution types of goods demand quantity are distinguished based on different data characteristics and distribution features. The model is trained using this historical goods demand quantity data, allowing it to focus on the distribution types of goods demand quantity at different time periods during training. Thus, the model can accurately fit the development trend of goods demand quantity by combining distribution types.

[0151] During training, historical cargo demand data with configuration annotations is used as the training set for the model. This data is then sequentially input into the initial neural network model for forward propagation. The loss function is then calculated based on the cargo demand quantity prediction results output by the neural network model. Backpropagation is performed based on this loss function to adjust the parameters of the neural network model. Through multiple forward and backward propagations, the goal of training the neural network model is achieved. When the requirements for the number of training iterations and cargo demand quantity prediction accuracy are met, the trained neural network model becomes the cargo demand quantity prediction model. This model can be used in practical applications to predict the cargo demand quantity for a relevant time period based on the input historical cargo demand quantity data.

[0152] As another exemplary implementation, in this application, the distribution type of the demand trend of goods may include a normal distribution type, a gamma distribution type, a Poisson distribution type, and a negative binomial distribution type that respectively satisfy a preset fluctuation type.

[0153] The preset fluctuation type refers to the fluctuation type obtained by removing three types of fluctuation types from the fluctuation type set, namely, extreme low-frequency fluctuation type, extreme high-frequency fluctuation type, and extreme small fluctuation type, based on the proportion of zero value and the amplitude of fluctuation. The preset fluctuation type includes high-frequency stable fluctuation type, low-frequency stable fluctuation type, high-frequency fluctuation type, and low-frequency fluctuation type.

[0154] For example, in this application, demand characteristic parameters can be extracted from the commodity demand quantity data. These demand characteristic parameters can characterize the fluctuations and frequency of occurrence of commodity demand quantity trends. For example, demand characteristic parameters may include:

[0155] n: The number of non-zero values, that is, the number of times non-zero demand occurs;

[0156] p: Average interval, number of data points / number of non-zero data points;

[0157] m: the mean of non-zero demand;

[0158] cv: Coefficient of variation, standard deviation / mean, divided into cv which includes non-zero values ​​and cv_nz which does not include non-zero values. cv is used to judge extreme fluctuations, while cv_nz is used in other cases.

[0159] Based on the obtained demand feature parameters, configure the corresponding classification threshold to classify the fluctuation type:

[0160] Extremely low-frequency fluctuation type: During the analysis period, the number of non-zero values ​​in the data is very small, no more than 3, and it has no statistical significance for the calculation of its required parameters;

[0161] Extremely small fluctuation type: Non-zero demand is very small, with a mean of less than or equal to 1;

[0162] Extreme volatility type: The data fluctuates abnormally, with a coefficient of variation (cv) greater than or equal to 5;

[0163] High-frequency stable type: The proportion of zero-value data is small, and the data is relatively stable;

[0164] Low-frequency stable type: Zero values ​​account for a large proportion, and the data is relatively stable;

[0165] High-frequency fluctuation type: Zero-value data accounts for a small proportion, and data fluctuations are large;

[0166] Low-frequency fluctuation type: Zero-value data accounts for a large proportion, and the data fluctuation is relatively large.

[0167] For example, refer to Figure 6 The diagram shown illustrates a scenario for the fluctuation classification processing described in this application. Figure 6 The specific classification thresholds are n = 3, m = 1, cv = 5, and p = 1.32, and the fluctuation type is classified accordingly. The cv is used to judge the extreme fluctuation type, and cv_nz is used for other cases.

[0168] After determining the fluctuation type, three fluctuation types can be eliminated: extremely low frequency fluctuation type, extremely high frequency fluctuation type, and extremely small fluctuation type. This will exclude the commodity demand quantity data with these three fluctuation characteristics from the subsequent screening of distribution types.

[0169] In the subsequent process of determining the classification type, the optimal fitting distribution and its distribution parameters of the goods demand quantity data can be determined through methods such as maximum likelihood estimation. In this application, the main distribution types include the normal distribution N(μ,σ). 2 The four distribution types are: α, λ, Gamma distribution G(α,λ), Poisson distribution P(λ), and negative binomial distribution NB(r,p).

[0170] After obtaining the predicted quantity of goods demand through the goods demand quantity forecasting model, the corresponding buffer inventory quantity can be determined based on the predicted quantity of goods demand and the current inventory quantity of the target node. The buffer inventory quantity is used to indicate the reserve inventory quantity that is higher than the current inventory quantity. In practical applications, this is to avoid supply shortages in the event of increased demand from new customers or other customers who need to meet basic goods needs.

[0171] Based on the current inventory data and the forecast data of demand, the difference between the corresponding quantities of goods can be determined. If the difference is positive, it means that the current inventory is sufficient to meet the forecast of demand. If the difference is negative, it means that the inventory is insufficient and goods need to be replenished.

[0172] For example, an amplification factor can be configured to appropriately amplify the amount corresponding to the forecast data of the quantity of goods demand, such as amplification factor k = 1.1 or k = 1.2. Then, the amount is subtracted from the amount of the current inventory data of the goods, and the resulting amount difference D can be used as the amount corresponding to the buffer inventory data. If the amount difference D is positive, replenishment processing can be performed based on the value difference D; if it is negative, no processing is required, or other processing can be performed, such as transferring goods in the supply chain network to other logistics nodes.

[0173] Alternatively, a buffer amount can be configured to indicate the amount of goods that is higher than the amount corresponding to the forecast data of goods demand. After determining the difference between the amount corresponding to the forecast data of goods demand and the amount corresponding to the current inventory data of goods, the difference is added to the buffer amount, and the sum can be used as the amount corresponding to the buffer inventory data.

[0174] Of course, different methods can be used to determine the corresponding buffer inventory quantity based on the forecast data of the quantity of goods demand and the current inventory quantity data of goods. The inventory can be processed accordingly based on the buffer inventory quantity data, and can be adjusted as needed. No limit is set here.

[0175] To facilitate better implementation of the method for determining the quantity of buffer inventory in the logistics supply chain provided in this application, this application also provides an apparatus for determining the quantity of buffer inventory in the logistics supply chain.

[0176] See Figure 7 , Figure 7 This is a schematic diagram of a device for determining the quantity of buffer inventory in the logistics supply chain, as described in this application. Specifically, the device 700 for determining the quantity of buffer inventory in the logistics supply chain may include the following structure:

[0177] The transceiver unit 701 is used to acquire historical cargo demand quantity data of the target node, wherein the target node is a node in the logistics supply chain, and the historical cargo demand quantity data is used to indicate the demand quantity of the target node for the target cargo in a historical period.

[0178] Processing unit 702 is used to input historical goods demand quantity data into a goods demand quantity prediction model to predict the demand quantity of the target node for the target goods in the current time period and obtain goods demand quantity prediction data. The goods demand quantity prediction model is obtained by training an initial neural network model with goods demand quantity data from different time periods. The goods demand quantity data from different time periods are labeled with the distribution type of different goods at different nodes in different time periods. Based on the goods demand quantity prediction data and the current goods inventory quantity data, the buffer inventory quantity data for the target node for the target goods is determined. The current goods inventory quantity data is used to indicate the inventory quantity of the target node for the target goods in the current time period, and the buffer inventory quantity data is used to indicate the reserve inventory quantity that is higher than the inventory amount corresponding to the current goods inventory quantity data.

[0179] In one exemplary implementation, the acquisition unit 701 is specifically used for:

[0180] Acquire multiple cargo images of the target node over a historical time period. The cargo images contain images of shelf layers where goods are placed, and the cargo images are taken on-site by the target node.

[0181] Processing unit 702 is specifically used for:

[0182] The goods in the cargo image are identified to obtain cargo data, which includes the category and quantity of the identified goods; historical cargo demand data are then confirmed based on the cargo data.

[0183] In another exemplary implementation, the goods image is labeled with a date, and the commodity data includes multiple sets of sub-commodity data labeled with days as the date unit. The processing unit 702 is specifically used for:

[0184] Extract target sub-product data labeled with target dates from the product data;

[0185] Based on the target sub-product data, determine the daily inventory quantity and daily sales volume of the target goods at the target node;

[0186] Based on the daily inventory and sales data, determine the historical demand data for goods.

[0187] In yet another exemplary implementation, processing unit 702 is specifically used for:

[0188] Based on the node identifier of the target node, determine the location type of the target node in the node network of the logistics supply chain, where the location type includes end node type or non-end node type;

[0189] Based on location type, daily inventory data, and daily sales data, determine historical demand data for goods.

[0190] In conjunction with the third possible implementation of the second aspect of this application, in the fourth possible implementation of the second aspect of this application, the processing unit 702 is specifically used for:

[0191] When the location type is end node type, determine whether the amount of the daily inventory quantity data is greater than zero;

[0192] If the value is greater than zero, the sales data for that day will be determined as the historical demand data for goods at the target node on the target day.

[0193] If the value is zero, then based on the daily inventory quantity data and the daily sales data, extract the calibration day that has the same preset characteristics as the target node and the target day, where the daily inventory quantity data of the calibration day is greater than zero.

[0194] The average sales volume of historical goods demand data on the calibration date is used as the historical goods demand data for the target node on the target date.

[0195] In yet another exemplary implementation, processing unit 702 is specifically used for:

[0196] When the location type is a non-end node type, determine the downstream node with a downstream identifier, where the downstream identifier is used to identify the node that is downstream of the target node in the architecture of the logistics supply chain.

[0197] Collect order quantity data from downstream nodes on the target day, and determine the historical goods quantity demand data for the target node on the target day based on the order quantity data, daily inventory data, and daily sales data.

[0198] In another exemplary implementation, the distribution types include normal distribution, gamma distribution, Poisson distribution and negative binomial distribution, which respectively satisfy the preset fluctuation type. The preset fluctuation type is obtained by removing three fluctuation types from the fluctuation type set: extreme low frequency fluctuation type, extreme high frequency fluctuation type and extreme small fluctuation type, based on the proportion of zero value and the amplitude of fluctuation. The preset fluctuation type includes high frequency stable fluctuation type, low frequency stable fluctuation type, high frequency fluctuation type and low frequency fluctuation type.

[0199] This application also provides equipment for determining buffer inventory quantity data in the logistics supply chain, see reference. Figure 8 , Figure 8This diagram illustrates a structural schematic of a device for determining buffer inventory quantity data in the logistics supply chain according to this application. Specifically, the device for determining buffer inventory quantity data in the logistics supply chain includes a processor 801, a memory 802, and an input / output device 803. The processor 801 executes the computer program stored in the memory 802 to implement, for example... Figures 1 to 6 The steps of the method for determining the quantity of buffer inventory in the logistics supply chain in any embodiment correspond to the following; or, when the processor 801 executes the computer program stored in the memory 802, it implements the following: Figure 7 The functions of each unit in the corresponding embodiment, for example Figure 7 The hardware structure corresponding to the transceiver unit 701 is the input / output device 803, the hardware structure corresponding to the processing unit 702 is the processor 801, and the memory 802 is used to store the data executed by the processor 801. Figures 1 to 6 The computer program required for determining the quantity of buffer inventory in the logistics supply chain in any embodiment.

[0200] For example, a computer program may be divided into one or more modules / units, one or more of which are stored in memory 802 and executed by processor 801 to complete this application. One or more modules / units may be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of the computer program in a computer device.

[0201] The device for determining buffer inventory quantity data in the logistics supply chain may include, but is not limited to, processor 801, memory 802, and input / output device 803. Those skilled in the art will understand that the illustration is merely an example of a device for determining buffer inventory quantity data in the logistics supply chain and does not constitute a limitation on the device. It may include more or fewer components than illustrated, or combine certain components, or use different components. For example, the device for determining buffer inventory quantity data in the logistics supply chain may also include network access devices, buses, etc., and processor 801, memory 802, input / output device 803, and network access devices are connected via a bus.

[0202] The processor 801 can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor. The processor is the control center of the equipment used to determine inventory quantity data in the logistics supply chain, connecting various parts of the equipment through various interfaces and lines.

[0203] The memory 802 can be used to store computer programs and / or modules. The processor 801 implements various functions of the computer device by running or executing the computer programs and / or modules stored in the memory 802 and by calling the data stored in the memory 802. The memory 802 may mainly include a program storage area and a data storage area. The program storage area may store the operating system, application programs required for at least one function (such as sound playback function, image playback function, etc.), etc.; the data storage area may store data created based on the usage of devices determined by buffer inventory quantity data in the logistics supply chain (such as audio data, video data, etc.). In addition, the memory may include high-speed random access memory, and may also include non-volatile memory, such as hard disk, RAM, plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0204] When processor 801 executes a computer program stored in memory 802, it can specifically perform the following functions:

[0205] Obtain historical cargo demand data for the target node, where the target node is a node in the logistics supply chain, and the historical cargo demand data is used to indicate the demand for the target cargo at the target node during a historical period.

[0206] Historical cargo demand data is input into the cargo demand prediction model to predict the demand for target cargo at the target node in the current time period and obtain cargo demand prediction data. The cargo demand prediction model is trained on the initial neural network model using cargo demand data from different time periods. The cargo demand data from different time periods are labeled with the distribution type of different cargo at different nodes in different time periods.

[0207] Based on the forecast data of the quantity of goods demand and the current quantity of goods inventory, the buffer inventory quantity data for the target node for the target goods is determined. The current quantity of goods inventory indicates the quantity of goods in stock at the target node in the current time period, and the buffer inventory quantity data indicates the quantity of reserve inventory that is higher than the quantity of inventory corresponding to the current quantity of goods inventory.

[0208] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the device, equipment, and corresponding units for determining the quantity of buffer inventory in the logistics supply chain described above can be found in, for example... Figures 1 to 6 The method for determining the quantity of buffer inventory in the logistics supply chain in any embodiment is explained below, and will not be repeated here.

[0209] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be performed by instructions, or by instructions controlling related hardware. These instructions can be stored in a computer-readable storage medium and loaded and executed by a processor.

[0210] Therefore, this application provides a computer-readable storage medium storing a plurality of instructions that can be loaded by a processor to execute the present application. Figures 1 to 6 For the steps in the method for determining the quantity of buffer inventory in the logistics supply chain in any embodiment, please refer to the following for specific operations: Figures 1 to 6 The method for determining the quantity of buffer inventory in the logistics supply chain in any embodiment will not be repeated here.

[0211] The computer-readable storage medium may include: read-only memory (ROM), random access memory (RAM), disk or optical disk, etc.

[0212] Because of the instructions stored in the computer-readable storage medium, the present application can be executed as described above. Figures 1 to 6 Corresponding to the steps in the method for determining the quantity of buffer inventory in the logistics supply chain in any embodiment, this application can achieve the following: Figures 1 to 6The beneficial effects that can be achieved by the method for determining the quantity of buffer inventory in the logistics supply chain in any embodiment are detailed in the preceding description and will not be repeated here.

[0213] The above provides a detailed description of the method, apparatus, equipment, and computer-readable storage medium for determining buffer inventory quantity data in the logistics supply chain provided by this application. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for determining the quantity of buffer inventory in a logistics supply chain, characterized in that, The method includes: Obtain historical cargo demand quantity data for a target node, wherein the target node is a node in the logistics supply chain, and the historical cargo demand quantity data is used to indicate the demand quantity of the target cargo for the target node during a historical period. Demand feature parameters are extracted from the historical goods demand quantity data; based on the demand feature parameters, corresponding classification thresholds are configured to classify the fluctuation types and obtain a set of fluctuation types; based on the proportion of zero values ​​and the amplitude of fluctuations, preset fluctuation types are selected from the set of fluctuation types. The historical cargo demand quantity data is input into the cargo demand quantity prediction model to predict the demand quantity of the target cargo at the target node in the current time period, and cargo demand quantity prediction data is obtained. The cargo demand quantity prediction model is obtained by training an initial neural network model with cargo demand quantity data from different time periods. The cargo demand quantity data from different time periods are labeled with the distribution type of different cargo at different nodes in different time periods. The distribution type satisfies the preset fluctuation type. Based on the predicted demand data and the current inventory data, the buffer inventory data for the target node for the target goods is determined. The current inventory data indicates the inventory quantity of the target node for the target goods in the current time period, and the buffer inventory data indicates a reserve inventory quantity that is higher than the inventory amount corresponding to the current inventory data.

2. The method according to claim 1, characterized in that, The acquisition of historical cargo demand quantity data for the target node includes: Multiple cargo images of the target node during the historical time period are acquired, wherein the image content of the cargo images includes shelf layers on which goods are placed, and the cargo images are taken on-site by the target node; The goods in the image are identified to obtain goods data, which includes the goods category and quantity of the identified goods. The historical demand quantity data for goods is confirmed based on the commodity data.

3. The method according to claim 2, characterized in that, The goods image is labeled with a date, and the goods data includes multiple sets of sub-goods data labeled with days as the date unit. Confirming the historical goods demand quantity data based on the goods data includes: Extract target sub-product data labeled with the target date from the product data; Based on the target sub-product data, determine the daily inventory quantity data and daily sales data of the target node for the target goods; Based on the daily inventory data and the daily sales data, the historical demand data for goods is determined.

4. The method according to claim 3, characterized in that, The step of determining the historical demand data for goods based on the daily inventory data and the daily sales data includes: Based on the node identifier of the target node, the location type of the target node in the node network of the logistics supply chain is determined, wherein the location type includes end node type or non-end node type; The historical demand data for goods is determined based on the location type, the daily inventory quantity data, and the daily sales data.

5. The method according to claim 4, characterized in that, The step of determining the historical goods demand quantity data based on the location type, the daily inventory quantity data, and the daily sales data includes: When the location type is an end node type, it is determined whether the amount of the daily inventory quantity data is greater than zero; If the value is greater than zero, the sales data for that day will be determined as the historical demand data for goods at the target node on that target day. If the value is zero, then based on the daily inventory quantity data and the daily sales data, extract the calibration day that has the same preset characteristics as the target node and the target day, wherein the amount of the daily inventory quantity data of the calibration day is greater than zero. The average sales volume of the historical goods demand data on the calibration date is determined as the historical goods demand data of the target node on the target date.

6. The method according to claim 4, characterized in that, The step of determining the historical goods demand quantity data based on the location type, the daily inventory quantity data, and the daily sales data includes: When the location type is a non-end node type, a downstream node with a downstream identifier is determined, wherein the downstream identifier is used to identify a node that is downstream of the target node in the architecture of the logistics supply chain; The order quantity data of the downstream node on the target day is statistically analyzed, and the historical demand data of the target node on the target day is determined based on the order quantity data, the inventory quantity data of the day, and the sales volume data of the day.

7. The method according to claim 5 or 6, characterized in that, The distribution types include normal distribution, gamma distribution, Poisson distribution, and negative binomial distribution types that satisfy the preset fluctuation types respectively. The preset fluctuation types are obtained by removing three types of fluctuation types from the fluctuation type set, namely, extreme low-frequency fluctuation type, extreme high-frequency fluctuation type, and extreme small fluctuation type, based on the proportion of zero values ​​and the amplitude of fluctuations. The preset fluctuation types include high-frequency stable fluctuation type, low-frequency stable fluctuation type, high-frequency fluctuation type, and low-frequency fluctuation type.

8. A device for determining the quantity of buffer inventory in a logistics supply chain, characterized in that, The device includes: The transceiver unit is used to acquire historical cargo demand quantity data of a target node, wherein the target node is a node in the logistics supply chain, and the historical cargo demand quantity data is used to indicate the demand quantity of the target cargo for the target node in a historical time period. The processing unit is configured to extract demand feature parameters from the historical goods demand quantity data; configure corresponding classification thresholds based on the demand feature parameters to classify fluctuation types and obtain a set of fluctuation types; filter the set of fluctuation types based on the proportion of zero values ​​and the amplitude of fluctuations to obtain a preset fluctuation type; input the historical goods demand quantity data into a goods demand quantity prediction model to predict the demand quantity of the target node for the target goods in the current time period and obtain goods demand quantity prediction data, wherein the goods demand quantity prediction model is obtained by training an initial neural network model with goods demand data from different time periods, and the goods demand data from different time periods are labeled with the distribution type of different goods at different nodes in different time periods; the distribution type satisfies the preset fluctuation type; and determine the buffer inventory quantity data of the target node for the target goods based on the goods demand quantity prediction data and the current goods inventory quantity data, wherein the current goods inventory quantity data is used to indicate the inventory quantity of the target node for the target goods in the current time period, and the buffer inventory quantity data is used to indicate the reserve inventory quantity that is higher than the inventory amount corresponding to the current goods inventory data.

9. A device for determining buffer inventory quantity data in a logistics supply chain, characterized in that, The method includes a processor and a memory, wherein the memory stores a computer program, and the processor executes the method as described in any one of claims 1 to 7 when it invokes the computer program in the memory.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a plurality of instructions adapted for loading by a processor to perform the method of any one of claims 1 to 7.

Images