Warehouse optimization method, device, computer equipment and storage medium

Abstract

The present application relates to a warehouse optimization method, device, computer equipment and storage medium. The method obtains a warehouse optimization request; searches for the historical shipping records of each inventory item in the warehouse corresponding to the warehouse optimization request, and obtains the shipping characteristic data corresponding to the historical shipping records; determines the long-tail inventory items in the inventory items according to the historical shipping records; obtains the first expected shipping result according to the shipping characteristic data corresponding to the long-tail inventory items; and optimizes the long-tail inventory items in the warehouse corresponding to the warehouse optimization request according to the first expected shipping result. The present application determines the long-tail inventory items in the inventory items through the historical shipping records of the warehouse, and then conducts a special analysis of the long-tail inventory items, and uses this to perform warehouse optimization, which can effectively manage the inventory items in the warehouse and more effectively match the warehouse settings with the shipping requirements.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to a warehouse optimization method, device, computer equipment and storage medium. Background Art

[0002] With the development of industrial Internet technology, C2M (Customer-to-Manufacturer) technology has emerged. C2M is a new business model of industrial Internet e-commerce, also known as the "short-circuit economy". C2M specifically refers to the reverse production model driven by users in modern manufacturing. The C2M model is based on the Internet, big data, artificial intelligence, and through the automation, customization, energy saving and flexibility of the production line, it uses a huge computer system to exchange data at any time, sets suppliers and production processes according to customer product order requirements, and finally produces an industrial customization model for personalized products. The spread of C2M has led to changes in OTD (Order On-time delivery rate) and production methods, and large fluctuations and short-cycle orders have become the norm. However, some commodities with long cycles, small freight volumes and many types (such as some high-value automotive aftermarket parts) occupy a large amount of resources.

[0003] In traditional technology, the historical sales data of these products are generally summarized, and then the historical average method is used to predict the future sales volume of these products, and then the storage settings of local warehouses are carried out.

[0004] However, the current method of setting up warehouses through the historical average method is not applicable to long-tail inventory items with small shipment volumes, and cannot effectively manage the inventory items in the warehouse. Summary of the invention

[0005] Based on this, it is necessary to provide a warehouse optimization method, device, computer equipment and storage medium that can more effectively manage the inventory in the warehouse in response to the above technical problems.

[0006] A storage optimization method, comprising:

[0007] Get storage optimization request;

[0008] Searching for historical shipment records of each inventory item in the warehouse corresponding to the warehouse optimization request, and obtaining shipment feature data corresponding to the historical shipment records;

[0009] Determine the long-tail inventory items in the inventory items according to the historical shipment records;

[0010] Obtaining a first expected shipment result according to the shipment characteristic data corresponding to the long-tail inventory item;

[0011] According to the first expected shipment result, storage optimization is performed on the long-tail inventory items in the warehouse corresponding to the storage optimization request.

[0012] In one of the embodiments, after determining the long-tail inventory items in the inventory items based on the historical shipping records, it also includes: analyzing the shipping characteristic data corresponding to the inventory items other than the long-tail inventory items through the weighted moving average method combined with the LGB prediction to obtain a second expected shipping result; and performing warehousing optimization on the inventory items other than the long-tail inventory items in the warehouse corresponding to the warehousing optimization request based on the second expected shipping result.

[0013] In one of the embodiments, searching for historical shipping records of each inventory item in the warehouse corresponding to the warehouse optimization request and obtaining shipping feature data corresponding to the historical shipping records includes: searching for historical shipping records of each inventory item in the warehouse corresponding to the warehouse optimization request; performing data cleaning and data conversion processing on the shipping data in the historical shipping records; performing feature engineering processing on the shipping data after the data cleaning and data conversion processing to obtain shipping feature data.

[0014] In one of the embodiments, the data cleaning and data conversion processing are subjected to feature engineering processing to obtain shipment feature data, including: the data cleaning and data conversion processing are subjected to feature engineering processing to obtain engineering feature data; and the engineering feature data are subjected to feature screening through time series stability to obtain shipment feature data.

[0015] In one embodiment, the analyzing the shipment characteristic data corresponding to the inventory items other than the long-tail inventory items by the weighted moving average method combined with the LGB forecast to obtain the second shipment expectation result also includes:

[0016] Get the shipping record training data;

[0017] Extracting shipment feature data corresponding to the shipment record training data;

[0018] The shipment feature data corresponding to the shipment record training data is used as a training parameter, and the average absolute percentage error of the second shipment expected result is used as a model optimization target parameter. The initial prediction model is trained through a Bayesian optimized parameter search to obtain a shipment volume prediction model, wherein the initial prediction model is constructed based on a weighted moving average method combined with an LGB prediction;

[0019] The method of analyzing the shipment characteristic data corresponding to the inventory items other than the long-tail inventory items by combining the weighted moving average method with the LGB forecast to obtain the second expected shipment result includes: inputting the shipment characteristic data corresponding to the inventory items other than the long-tail inventory items into the shipment volume forecast model to obtain the second expected shipment result.

[0020] In one of the embodiments, obtaining the first expected shipment result based on the shipment characteristic data corresponding to the long-tail inventory item includes: when the shipment volume data corresponding to the long-tail inventory item is lower than or equal to a preset shipment volume threshold, obtaining the inventory item category corresponding to the long-tail inventory item, and obtaining the first expected shipment result corresponding to the long-tail inventory item based on the inventory item category; when the shipment volume data corresponding to the long-tail inventory item is higher than the preset shipment volume threshold, obtaining the smoothed mean of the shipment volume corresponding to the long-tail inventory item, and obtaining the first expected shipment result corresponding to the long-tail inventory item based on the Poisson distribution corresponding to the smoothed mean.

[0021] A storage optimization device, comprising:

[0022] Request acquisition module, used to obtain storage optimization requests;

[0023] A feature data acquisition module is used to search for the historical shipment records of each inventory item in the warehouse corresponding to the warehouse optimization request, and obtain the shipment feature data corresponding to the historical shipment records;

[0024] A long-tail screening module, used for determining long-tail inventory items among the inventory items according to the historical shipment records;

[0025] A first analysis module, configured to obtain a first expected shipment result according to shipment characteristic data corresponding to the long-tail inventory items;

[0026] The first optimization module is used to perform warehousing optimization on the long-tail inventory items in the warehouse corresponding to the warehousing optimization request according to the first expected shipment result.

[0027] In one of the embodiments, it also includes a second optimization module, which is used to: analyze the shipping characteristic data corresponding to the inventory items other than the long-tail inventory items through the weighted moving average method combined with LGB (LightGBM) prediction to obtain a second expected shipping result; and perform warehousing optimization on the inventory items other than the long-tail inventory items in the warehouse corresponding to the warehousing optimization request according to the second expected shipping result.

[0028] A computer device comprises a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:

[0029] Get storage optimization request;

[0030] Searching for historical shipment records of each inventory item in the warehouse corresponding to the warehouse optimization request, and obtaining shipment feature data corresponding to the historical shipment records;

[0031] Determine the long-tail inventory items in the inventory items according to the historical shipment records;

[0032] Obtaining a first expected shipment result according to the shipment characteristic data corresponding to the long-tail inventory item;

[0033] According to the first expected shipment result, storage optimization is performed on the long-tail inventory items in the warehouse corresponding to the storage optimization request.

[0034] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the following steps:

[0035] Get storage optimization request;

[0036] Searching for historical shipment records of each inventory item in the warehouse corresponding to the warehouse optimization request, and obtaining shipment feature data corresponding to the historical shipment records;

[0037] Determine the long-tail inventory items in the inventory items according to the historical shipment records;

[0038] Obtaining a first expected shipment result according to the shipment characteristic data corresponding to the long-tail inventory item;

[0039] According to the first expected shipment result, storage optimization is performed on the long-tail inventory items in the warehouse corresponding to the storage optimization request.

[0040] The above-mentioned warehouse optimization method, device, computer equipment and storage medium obtain a warehouse optimization request; search for the historical shipping records of each inventory item in the warehouse corresponding to the warehouse optimization request, and obtain the shipping characteristic data corresponding to the historical shipping records; determine the long-tail inventory items in the inventory items according to the historical shipping records; obtain the first expected shipping result according to the shipping characteristic data corresponding to the long-tail inventory items; and perform warehouse optimization on the long-tail inventory items in the warehouse corresponding to the warehouse optimization request according to the first expected shipping result. The present application determines the long-tail inventory items in the inventory items through the historical shipping records of the warehouse, and then conducts a special analysis of the long-tail inventory items, and uses this to perform warehouse optimization, which can effectively manage the inventory items in the warehouse. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 An application environment diagram of a storage optimization method in an embodiment;

[0042] Figure 2 A schematic diagram of a flow chart of a storage optimization method in one embodiment;

[0043] Figure 3 A schematic diagram of a flow chart of a storage optimization method in another embodiment;

[0044] Figure 4 In one embodiment Figure 3 Schematic diagram of the sub-process of step 203;

[0045] Figure 5 It is a structural block diagram of a storage optimization device in one embodiment;

[0046] Figure 6 FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0047] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0048] The storage optimization method provided in this application can be applied to Figure 1 In the application environment shown. Among them, the terminal 102 communicates with the storage optimization server 104 through the network. When a user wants to optimize the inventory structure of a warehouse, a storage optimization request can be sent to the storage optimization server 104 through the terminal 102. The storage optimization request carries the identity information corresponding to the warehouse. The storage optimization server 104 obtains the storage optimization request from the terminal 102; searches for the historical shipment records of each inventory in the warehouse corresponding to the storage optimization request, and obtains the shipment feature data corresponding to the historical shipment record; determines the long-tail inventory in the inventory according to the historical shipment record; analyzes the shipment feature data corresponding to the long-tail inventory through a preset long-tail shipment prediction algorithm to obtain the first expected shipment result; performs storage optimization on the long-tail inventory in the warehouse corresponding to the storage optimization request according to the first expected shipment result. Among them, the terminal 102 can be, but is not limited to, various personal computers, laptops, smart phones, tablet computers and portable wearable devices, and the storage optimization server 104 can be implemented by an independent server or a server cluster composed of multiple servers.

[0049] In one embodiment, Figure 2 As shown, a storage optimization method is provided, which is applied to Figure 1 Taking the warehouse optimization server 104 in the example as an example, the following steps are included:

[0050] Step 201, obtaining a warehouse optimization request.

[0051] The storage optimization request carries target warehouse information corresponding to the storage optimization. The target warehouse may specifically refer to a warehouse or a warehouse cluster responsible for supplying inventory to the same region.

[0052] Specifically, the inventory of the goods in the warehouse is closely related to the sales of the goods in the warehouse location. When the user wants to optimize the structure of the inventory of the goods in a warehouse to ensure that the inventory of the warehouse can meet the actual delivery requirements, the user can send a corresponding warehouse optimization request to the warehouse optimization server 104 through the terminal 102 to request the warehouse optimization server to perform quantitative analysis on each inventory in the target warehouse to ensure that the warehouse setting matches the delivery requirements. In one embodiment, the warehouse optimization request can also include a cycle setting for warehouse optimization, specifically including a long cycle, a medium cycle and a short cycle. In different optimization cycles, the model used to optimize the warehouse storage, the input data of the model and the optimization results obtained are all different.

[0053] Step 203, searching for historical shipment records of each inventory item in the warehouse corresponding to the warehouse optimization request, and obtaining shipment feature data corresponding to the historical shipment records.

[0054] Among them, inventory can be classified in the form of SKU (Stock Keeping Unit), which is a unique identifier for each product and service. By using the value of SKU in data management, companies can track the inventory of systems such as warehouses and retail stores or products. The warehouse corresponding to the warehouse optimization request refers to the target warehouse for warehouse optimization. If the inventory in the warehouse is sold, the inventory will be shipped out of the warehouse and transferred to the hands of the inventory buyer. At this time, the warehouse will generate the corresponding inventory shipment record. For inventory without shipment records, it can be directly filled with 0. The shipment feature data specifically includes the shipment volume, shipment time, target location of shipment transportation, and shipment frequency in different time periods in the historical records.

[0055] Specifically, after receiving a storage optimization request, the storage optimization server 104 can parse the storage optimization request to determine the warehouse corresponding to the storage optimization request, and then directly search for the inventory shipment record corresponding to the warehouse. In one embodiment, the shipment record corresponds to the cycle setting of the storage optimization. For different cycles, the time span of the extracted shipment records is also different. For a longer storage optimization cycle, the corresponding shipment record span is also longer. When the historical shipment record of the inventory is obtained, in order to more effectively set the storage optimization, the corresponding shipment feature data can be extracted from the historical shipment record, and the corresponding inventory optimization setting result can be obtained by analyzing the shipment feature data.

[0056] Step 205, determining the long-tail inventory items in the inventory items according to the historical shipping records.

[0057] Among them, long-tail inventory refers to inventory with a long-tail effect. From the perspective of people's needs, most of the demands will be concentrated at the head, and this part is called popular, while the demands distributed at the tail are personalized, scattered and small-scale demands. And this part of differentiated and small-scale demands will form a long "tail" on the demand curve, and this part of demand is the long tail. Long-tail inventory generally has a low shipment volume and shipment frequency. In one of the embodiments, the inventory in the warehouse can specifically be automotive aftermarket parts. For automotive aftermarket parts of the same brand, the number of SKUs in stock in the warehouse can be as many as tens of thousands, and most of the SKUs have an annual shipment volume of less than 10 pieces and an average monthly shipment volume of less than 1 piece. At this time, these SKUs can be regarded as long-tail inventory.

[0058] Specifically, the warehouse optimization server 104 can determine which inventory items are long-tail inventory items with low shipping frequency and shipping volume based on historical shipping records, and then conduct a special analysis on the long-tail inventory items, estimate the future shipping volume of these long-tail inventory items, and then optimize the storage structure in the target warehouse.

[0059] Step 207, obtaining a first expected shipment result according to the shipment characteristic data corresponding to the long-tail inventory items.

[0060] The first expected shipment result specifically refers to the result obtained by predicting the future shipment of each long-tail inventory item in the warehouse according to the warehouse optimization request. For example, it is expected that the shipment volume of inventory item A in the next month will be 1 piece, and the shipment volume of inventory item B in the next month will be 3 pieces.

[0061] Specifically, the warehouse optimization server can analyze the historical shipment records of the target warehouse to conduct a detailed analysis of the future shipment status of the long-tail inventory items in the target warehouse, so as to optimize the inventory quantity of each long-tail inventory item in the target warehouse.

[0062] Step 209: Performing storage optimization on the long-tail inventory items in the warehouse corresponding to the storage optimization request according to the first expected shipment result.

[0063] Specifically, the first expected shipment result refers to the future shipment volume of each inventory item in the target warehouse, and the storage optimization request is made to the long-tail inventory items in the warehouse according to the first expected shipment result, specifically including first determining the existing inventory of each long-tail inventory item in the warehouse, and then determining the replenishment of the long-tail inventory items in the warehouse or the transfer of the excess long-tail inventory items to other warehouses according to the future shipment volume. Through storage optimization, the amount of goods in transit can be effectively reduced, thereby improving the production efficiency of the long-tail inventory factory.

[0064] The above-mentioned warehouse optimization method obtains a warehouse optimization request; searches for the historical shipping records of each inventory item in the warehouse corresponding to the warehouse optimization request, and obtains the shipping feature data corresponding to the historical shipping records; determines the long-tail inventory items in the inventory items according to the historical shipping records; obtains the first expected shipping result according to the shipping feature data corresponding to the long-tail inventory items; and optimizes the long-tail inventory items in the warehouse corresponding to the warehouse optimization request according to the first expected shipping result. This application determines the long-tail inventory items in the inventory items through the historical shipping records of the warehouse, and then conducts a special analysis of the long-tail inventory items, and uses this to perform warehouse optimization, which can effectively manage the inventory items in the warehouse.

[0065] In one embodiment, Figure 3 As shown, after step 209, the following steps are also included:

[0066] Step 302: By combining the weighted moving average method with the LGB forecast, the shipment characteristic data corresponding to the inventory items other than the long-tail inventory items are analyzed to obtain the second shipment expectation result.

[0067] Step 304: Performing storage optimization on the inventory items in the warehouse corresponding to the storage optimization request except for the long-tail inventory items according to the second expected shipment result.

[0068] Among them, the weighted moving average method is to give different weights according to the degree of influence of data at different times in the same moving segment on the predicted value, and then perform average movement to predict future values. Unlike the simple moving average method, the weighted moving average method treats the data in the moving period equally when calculating the average value, but treats each data in the moving period differently based on the characteristic that the more recent data has a greater impact on the predicted value. Give a larger weight to recent data and a smaller weight to distant data, so as to make up for the shortcomings of the simple moving average method. LGB prediction is predicted by LigthGBM, which is a new member in the boosting ensemble model. Like XGBoost, it is an efficient implementation of GBDT. In principle, it is similar to GBDT and XGBoost, and both use the negative gradient of the loss function as the residual approximation of the current decision tree to fit the new decision tree. The LGB prediction in this application can be specifically a rolling prediction algorithm set according to the expected length of time required for shipment. In one embodiment, the LGB prediction can be rolled for 28 days to estimate the shipment volume of inventory during this period.

[0069] Specifically, since there are a considerable number of non-long-tail inventory items in addition to long-tail inventory items in the warehouse, for this part of inventory items, the weighted moving average method combined with LGB forecasting can be used to estimate its shipment expectations in the future. In one embodiment, a forecasting model can be constructed based on the weighted moving average method combined with LGB forecasting, and then the second expected shipment result can be predicted by inputting the obtained shipment feature data into the forecasting model. The forecasting model is specifically implemented by a linear regression model combined with an LGB model and a prophet model. In this embodiment, by analyzing the shipment feature data corresponding to inventory items other than long-tail inventory items, the second expected shipment result is obtained, and then the inventory items other than long-tail inventory items in the warehouse are optimized, the inventory structure of the inventory items in the warehouse can be further optimized, and the storage setting can be more effectively matched with the shipment demand.

[0070] like Figure 4 As shown, in one embodiment, step 203 includes:

[0071] Step 401, searching for historical shipment records of each inventory item in the warehouse corresponding to the warehouse optimization request.

[0072] Step 403: Perform data cleaning and data conversion processing on the shipment data in the historical shipment records.

[0073] Step 405, performing feature engineering processing on the shipment data after data cleaning and data conversion processing to obtain shipment feature data.

[0074] Among them, data cleaning specifically refers to removing invalid data in historical shipment record data. For example, some data in historical shipment data is irrelevant to future shipment forecasts and can be directly removed through data cleaning. Data conversion refers to data conversion of null values ​​and meaningless data in historical shipment data. When the inventory in the warehouse contains a large number of long-tail inventory items, the shipment quantity of a certain inventory item may be 0 during a certain period of time. At this time, data conversion can be performed to fill the corresponding shipment data with 0. Feature engineering processing refers to extracting the features contained in the shipment data, and then the warehouse optimization server 104 can directly perform subsequent predictions based on the features. Feature engineering can specifically include box-cox transformation, extraction of sliding window time series statistics, feature crossover and other processing processes.

[0075] Specifically, the warehouse optimization server 104 can directly obtain the historical shipping records of each inventory item in the warehouse corresponding to the user's warehouse optimization request, and then perform corresponding data cleaning and data conversion processing on the shipping data in the historical shipping records to ensure the validity of the data, and then further process the shipping data through feature engineering to extract the shipping feature data contained in the shipping features that can be used for future inventory shipment prediction. In this embodiment, by performing data cleaning, data conversion, and feature engineering on the historical shipping records to obtain the shipping feature data, the validity of the shipping feature data can be effectively guaranteed, thereby improving the prediction accuracy of the subsequent shipment prediction process.

[0076] In one embodiment, step 405 includes: performing feature engineering processing on data cleaning and data conversion processing to obtain engineering feature data; performing feature screening on the engineering feature data through time series stability to obtain shipment feature data.

[0077] Time series stationarity requires that the fitting curve of the features obtained through the sample time series can continue "inertially" along the existing form in the future; if the data is non-stationary, it means that the form of the sample fitting curve does not have the characteristic of "inertial" continuation. Screening engineering feature data based on time series stability means retaining engineering feature data with time series stability and using it as shipping feature data. In addition, feature screening can be performed by recursive feature elimination or the importance of the obtained shipping features can be sorted by a method based on the number of splits and split gain sorting of the tree model. In this embodiment, the shipping feature data is obtained by processing feature screening, etc., which can effectively ensure the effectiveness of the shipping feature data used for shipment analysis, thereby improving the prediction accuracy of the subsequent shipment forecasting process.

[0078] In one embodiment, before step 302, the method further includes: obtaining shipping record training data;

[0079] Extract the shipment feature data corresponding to the shipment record training data; use the shipment feature data corresponding to the shipment record training data as training parameters, use the mean absolute percentage error of the second shipment expected result as the model optimization target parameter, train the initial prediction model through Bayesian optimization parameter search, and obtain the shipment volume prediction model, the initial prediction model is constructed based on the weighted moving average method combined with LGB prediction; step 302 includes: inputting the shipment feature data corresponding to the inventory items except the long-tail inventory items into the shipment volume prediction model to obtain the second shipment expected result.

[0080] The shipping record training data can specifically be the historical shipping data within a period of time before the current forecast date. It is used to train the prediction model. This part of the data still needs to be processed by data cleaning and data conversion. The training parameters can specifically include training set data, test set data and validation set data. Specifically, the data after data cleaning and data conversion can be processed by time series stratification and random sampling to obtain available training data, and the data can be divided into training set data, test set data and validation set data. After obtaining the training data, the shipping feature data corresponding to the shipping record training data is used as the training parameter, and the average absolute percentage error of the second expected shipping result is used as the model optimization target parameter. The initial prediction model is trained through the parameter search of Bayesian optimization to obtain the shipment volume prediction model. The initial prediction model is constructed based on the weighted moving average method combined with LGB prediction. When it is necessary to estimate the second expected shipment result, the shipping feature data corresponding to the inventory items other than the long-tail inventory items can be directly input into the shipment volume prediction model to directly obtain the corresponding second expected shipment result. In this embodiment, through the historical shipping records corresponding to the warehouse

[0081] In one embodiment, the shipment characteristic data includes shipment quantity data, and step 207 includes: when the shipment quantity data corresponding to the long-tail inventory item is lower than or equal to a preset shipment quantity threshold, obtaining the inventory item category corresponding to the long-tail inventory item, and obtaining the first expected shipment result corresponding to the long-tail inventory item according to the inventory item category; when the shipment quantity data corresponding to the long-tail inventory item is higher than the preset shipment quantity threshold, obtaining the smoothed mean of the shipment quantity corresponding to the long-tail inventory item, and obtaining the first expected shipment result corresponding to the long-tail inventory item according to the Poisson distribution corresponding to the smoothed mean.

[0082] Among them, the preset shipment threshold can be a very small value, such as 0 or 1. When the shipment data corresponding to the long-tail inventory is lower than or equal to the preset shipment threshold, it can be determined that the long-tail sales data corresponding to the long-tail inventory is too low. At this time, the corresponding inventory index can be directly allocated to it according to the category of the long-tail inventory. When the shipment data corresponding to the long-tail inventory is higher than the preset shipment threshold, it can be determined that the corresponding sales data can be further estimated. At this time, the smoothed mean of the shipment corresponding to the long-tail inventory can be obtained, and the first expected shipment result corresponding to the long-tail inventory can be obtained according to the Poisson distribution corresponding to the smoothed mean. In another embodiment, the subsequent shipment estimation can also be performed based on the smoothed mean of the shipment by the SAA (Sample Average Approximation) method. In this embodiment, different shipment estimates are performed for different types of long-tail inventory according to the actual shipment situation of the long-tail inventory, which can effectively ensure the accuracy of the shipment estimation process.

[0083] It should be understood that although Figure 2-4 The steps in the flowchart are shown in sequence as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified in this document, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders. Moreover, Figure 2-4 At least part of the steps may include multiple steps or multiple stages. These steps or stages are not necessarily performed at the same time, but can be performed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed in turn or alternately with other steps or at least part of the steps or stages in other steps.

[0084] In one embodiment, Figure 5 As shown, a storage optimization device is provided, including: a request acquisition module 501, a feature data acquisition module 503, a long-tail screening module 505, a first analysis module 507 and a first optimization module 509, wherein:

[0085] The request acquisition module 501 is used to acquire a storage optimization request.

[0086] The characteristic data acquisition module 503 is used to search for the historical shipment records of each inventory item in the warehouse corresponding to the warehouse optimization request, and obtain the shipment characteristic data corresponding to the historical shipment records.

[0087] The long-tail screening module 505 is used to determine the long-tail inventory items in the inventory items according to the historical shipment records.

[0088] The first analysis module 507 is used to obtain a first expected shipment result according to the shipment characteristic data corresponding to the long-tail inventory items.

[0089] The first optimization module 509 is used to perform storage optimization on the long-tail inventory items in the warehouse corresponding to the storage optimization request according to the first expected shipment result.

[0090] In one of the embodiments, a second optimization module is also included, which is used to: analyze the shipping characteristic data corresponding to inventory items other than long-tail inventory items through the weighted moving average method combined with LGB prediction to obtain a second expected shipping result; and perform warehousing optimization on the inventory items other than long-tail inventory items in the warehouse corresponding to the warehousing optimization request based on the second expected shipping result.

[0091] In one of the embodiments, the feature data acquisition module 503 is used to: search for historical shipment records of each inventory item in the warehouse corresponding to the warehouse optimization request; perform data cleaning and data conversion processing on the shipment data in the historical shipment records; perform feature engineering processing on the shipment data after data cleaning and data conversion processing to obtain shipment feature data.

[0092] In one of the embodiments, the feature data acquisition module 503 is also used to: perform feature engineering processing on data cleaning and data conversion processing to obtain engineering feature data; perform feature screening on the engineering feature data through time series stability to obtain shipment feature data.

[0093] In one embodiment, a model training module is further included, which is used to: obtain the shipping record training data; extract the shipping feature data corresponding to the shipping record training data; use the shipping feature data corresponding to the shipping record training data as the training parameters, use the average absolute percentage error of the second expected shipping result as the model optimization target parameter, train the initial prediction model through Bayesian optimization parameter search, and obtain the shipment volume prediction model, the initial prediction model is constructed based on the weighted moving average method combined with the LGB prediction. The second optimization module is specifically used to: input the shipping feature data corresponding to the inventory items other than the long-tail inventory items into the shipment volume prediction model to obtain the second expected shipping result.

[0094] In one of the embodiments, the first analysis module is specifically used to: when the shipment volume data corresponding to the long-tail inventory item is lower than or equal to a preset shipment volume threshold, obtain the inventory item category corresponding to the long-tail inventory item, and obtain the first expected shipment result corresponding to the long-tail inventory item based on the inventory item category; when the shipment volume data corresponding to the long-tail inventory item is higher than the preset shipment volume threshold, obtain the smoothed mean of the shipment volume corresponding to the long-tail inventory item, and obtain the first expected shipment result corresponding to the long-tail inventory item based on the Poisson distribution corresponding to the smoothed mean.

[0095] For the specific definition of the storage optimization device, please refer to the definition of the storage optimization method above, which will not be repeated here. Each module in the above storage optimization device can be implemented in whole or in part by software, hardware and their combination. The above modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0096] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Figure 6As shown. The computer device includes a processor, a memory and a network interface connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store warehouse optimization data. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a warehouse optimization method is implemented.

[0097] Those skilled in the art will understand that Figure 6 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0098] In one embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the following steps are implemented:

[0099] Get storage optimization request;

[0100] Find the historical shipment records of each inventory item in the warehouse corresponding to the warehouse optimization request, and obtain the shipment feature data corresponding to the historical shipment records;

[0101] Determine the long-tail inventory items in the inventory based on historical shipment records;

[0102] Obtain the first expected shipment result based on the shipment characteristic data corresponding to the long-tail inventory items;

[0103] According to the expected result of the first shipment, the warehouse optimization request is responded to by performing warehouse optimization on the long-tail inventory items in the warehouse.

[0104] In one embodiment, when the processor executes the computer program, the following steps are also implemented: by combining the weighted moving average method with the LGB prediction, the shipping characteristic data corresponding to the inventory items other than the long-tail inventory items are analyzed to obtain a second expected shipping result; and according to the second expected shipping result, the storage optimization is performed on the inventory items other than the long-tail inventory items in the warehouse corresponding to the storage optimization request.

[0105] In one embodiment, when the processor executes the computer program, the following steps are also implemented: searching for historical shipment records of each inventory item in the warehouse corresponding to the warehouse optimization request; performing data cleaning and data conversion processing on the shipment data in the historical shipment records; performing feature engineering processing on the shipment data after data cleaning and data conversion processing to obtain shipment feature data.

[0106] In one embodiment, when the processor executes the computer program, the following steps are also implemented: performing feature engineering processing on data cleaning and data conversion processing to obtain engineering feature data; performing feature screening on the engineering feature data through time series stability to obtain shipment feature data.

[0107] In one embodiment, when the processor executes the computer program, the following steps are also implemented: obtaining shipment record training data; extracting shipment feature data corresponding to the shipment record training data; using the shipment feature data corresponding to the shipment record training data as training parameters, and using the average absolute percentage error of the second shipment expected result as the model optimization target parameter, training the initial prediction model through Bayesian optimized parameter search, and obtaining a shipment volume prediction model, wherein the initial prediction model is constructed based on the weighted moving average method combined with LGB prediction.

[0108] In one embodiment, when the processor executes the computer program, the following steps are also implemented: when the shipment volume data corresponding to the long-tail inventory item is lower than or equal to a preset shipment volume threshold, the inventory item category corresponding to the long-tail inventory item is obtained, and the first expected shipment result corresponding to the long-tail inventory item is obtained according to the inventory item category; when the shipment volume data corresponding to the long-tail inventory item is higher than the preset shipment volume threshold, the smoothed mean of the shipment volume corresponding to the long-tail inventory item is obtained, and the first expected shipment result corresponding to the long-tail inventory item is obtained according to the Poisson distribution corresponding to the smoothed mean.

[0109] In one embodiment, a computer readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the following steps are implemented:

[0110] Get storage optimization request;

[0111] Find the historical shipment records of each inventory item in the warehouse corresponding to the warehouse optimization request, and obtain the shipment feature data corresponding to the historical shipment records;

[0112] Determine the long-tail inventory items in the inventory based on historical shipment records;

[0113] Obtain the first expected shipment result based on the shipment characteristic data corresponding to the long-tail inventory items;

[0114] According to the expected result of the first shipment, the warehouse optimization request is responded to by performing warehouse optimization on the long-tail inventory items in the warehouse.

[0115] In one embodiment, when the computer program is executed by the processor, the following steps are also implemented: by combining the weighted moving average method with the LGB prediction, the shipping characteristic data corresponding to the inventory items other than the long-tail inventory items are analyzed to obtain a second expected shipping result; and according to the second expected shipping result, the storage optimization is performed on the inventory items other than the long-tail inventory items in the warehouse corresponding to the storage optimization request.

[0116] In one embodiment, when the computer program is executed by the processor, the following steps are also implemented: searching for historical shipment records of each inventory item in the warehouse corresponding to the warehouse optimization request; performing data cleaning and data conversion processing on the shipment data in the historical shipment records; performing feature engineering processing on the shipment data after data cleaning and data conversion processing to obtain shipment feature data.

[0117] In one embodiment, when the computer program is executed by the processor, the following steps are also implemented: performing feature engineering on data cleaning and data conversion processing to obtain engineering feature data; performing feature screening on the engineering feature data through time series stability to obtain shipment feature data.

[0118] In one embodiment, when the computer program is executed by the processor, the following steps are also implemented: obtaining shipment record training data; extracting shipment feature data corresponding to the shipment record training data; using the shipment feature data corresponding to the shipment record training data as training parameters, and using the average absolute percentage error of the second shipment expected result as the model optimization target parameter, training the initial prediction model through Bayesian optimized parameter search, and obtaining a shipment volume prediction model, wherein the initial prediction model is constructed based on the weighted moving average method combined with LGB prediction.

[0119] In one embodiment, when the computer program is executed by the processor, the following steps are also implemented: when the shipment volume data corresponding to the long-tail inventory item is lower than or equal to a preset shipment volume threshold, the inventory item category corresponding to the long-tail inventory item is obtained, and the first expected shipment result corresponding to the long-tail inventory item is obtained according to the inventory item category; when the shipment volume data corresponding to the long-tail inventory item is higher than the preset shipment volume threshold, the smoothed mean of the shipment volume corresponding to the long-tail inventory item is obtained, and the first expected shipment result corresponding to the long-tail inventory item is obtained according to the Poisson distribution corresponding to the smoothed mean.

[0120] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory may include read-only memory (ROM), magnetic tape, floppy disk, flash memory or optical memory, etc. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).

[0121] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0122] The above embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be subject to the attached claims.

Claims

1. A storage optimization method, comprising: Get storage optimization request; Searching for historical shipment records of each inventory item in the warehouse corresponding to the warehouse optimization request, and obtaining shipment characteristic data corresponding to the historical shipment records, wherein the shipment characteristic data includes shipment quantity, shipment time, target location of shipment transportation, and shipment frequency in different time periods in the historical records; Determine the long-tail inventory items in the inventory items according to the historical shipment records; Obtaining a first expected shipment result according to the shipment characteristic data corresponding to the long-tail inventory item; Performing storage optimization on the long-tail inventory items in the warehouse corresponding to the storage optimization request according to the first expected shipment result, so as to determine the quantity of replenishing the long-tail inventory items in the warehouse, or the quantity of transferring the excess long-tail inventory items to other warehouses except the warehouse; Get the shipping record training data; Extracting shipment feature data corresponding to the shipment record training data; The shipment feature data corresponding to the shipment record training data is used as a training parameter, and the average absolute percentage error of the second shipment expectation result is used as a model optimization target parameter. The initial prediction model is trained through a Bayesian optimized parameter search to obtain a shipment volume prediction model. The initial prediction model is constructed based on a weighted moving average method combined with an LGB prediction. The LGB prediction is a rolling prediction algorithm set according to the expected time length of the required shipment. Inputting the shipment characteristic data corresponding to the inventory items other than the long-tail inventory items into the shipment volume prediction model to obtain a second shipment expectation result; According to the second expected shipment result, storage optimization is performed on the inventory items in the warehouse corresponding to the storage optimization request except for the long-tail inventory items.

2. The method according to claim 1, characterized in that The searching for the historical shipment records of each inventory item in the warehouse corresponding to the warehouse optimization request and obtaining the shipment characteristic data corresponding to the historical shipment records includes: Searching for historical shipment records of each inventory item in the warehouse corresponding to the warehouse optimization request; Performing data cleaning and data conversion processing on the shipment data in the historical shipment records; Perform feature engineering processing on the shipment data after the data cleaning and data conversion processing to obtain shipment feature data.

3. The method according to claim 2, characterized in that The performing feature engineering processing on the shipment data after the data cleaning and data conversion processing to obtain shipment feature data includes: Performing feature engineering processing on the shipment data after the data cleaning and data conversion processing to obtain engineering feature data; The engineering feature data is feature screened through time series stability to obtain shipment feature data.

4. The method according to claim 1, characterized in that The shipment characteristic data includes shipment quantity data, and obtaining the first shipment expected result according to the shipment characteristic data corresponding to the long-tail inventory item includes: When the shipment volume data corresponding to the long-tail inventory item is lower than or equal to a preset shipment volume threshold, obtaining the inventory item category corresponding to the long-tail inventory item, and obtaining the first expected shipment result corresponding to the long-tail inventory item according to the inventory item category; When the shipment volume data corresponding to the long-tail inventory item is higher than a preset shipment volume threshold, the smoothed mean shipment volume corresponding to the long-tail inventory item is obtained, and the first expected shipment result corresponding to the long-tail inventory item is obtained according to the Poisson distribution corresponding to the smoothed mean.

5. A storage optimization device, characterized in that: The device comprises: Request acquisition module, used to obtain storage optimization requests; A feature data acquisition module is used to search for the historical shipment records of each inventory item in the warehouse corresponding to the warehousing optimization request, and obtain the shipment feature data corresponding to the historical shipment records, wherein the shipment feature data includes shipment quantity, shipment time, target location of shipment transportation, and shipment frequency in different time periods in the historical records; A long-tail screening module, used for determining long-tail inventory items among the inventory items according to the historical shipment records; A first analysis module, configured to obtain a first expected shipment result according to shipment characteristic data corresponding to the long-tail inventory items; The first optimization module is used to perform storage optimization on the long-tail inventory items in the warehouse corresponding to the storage optimization request according to the first expected shipment result, so as to determine the quantity of the long-tail inventory items in the warehouse to be replenished, or the quantity of the excess long-tail inventory items to be transferred to other warehouses except the warehouse. A model training module is used to obtain shipment record training data; extract shipment feature data corresponding to the shipment record training data; use the shipment feature data corresponding to the shipment record training data as training parameters, use the average absolute percentage error of the second shipment expected result as the model optimization target parameter, train the initial prediction model through Bayesian optimization parameter search, and obtain a shipment volume prediction model, wherein the initial prediction model is constructed based on the weighted moving average method combined with LGB prediction; The second optimization module is used to input the shipping characteristic data corresponding to the inventory items other than the long-tail inventory items into the shipment volume prediction model to obtain a second expected shipment result; and perform warehousing optimization on the inventory items other than the long-tail inventory items in the warehouse corresponding to the warehousing optimization request based on the second expected shipment result.

6. The device according to claim 5, characterized in that The characteristic data acquisition module is specifically used to: search for the historical shipment records of each inventory item in the warehouse corresponding to the warehouse optimization request; perform data cleaning and data conversion processing on the shipment data in the historical shipment records; Perform feature engineering processing on the shipment data after the data cleaning and data conversion processing to obtain shipment feature data.

7. The device according to claim 6, characterized in that The feature data acquisition module is also used to: perform feature engineering processing on the shipment data after the data cleaning and data conversion processing to obtain engineering feature data; and perform feature screening on the engineering feature data through time series stability to obtain shipment feature data.

8. The device according to claim 5, characterized in that The first analysis module is specifically used to: when the shipment volume data corresponding to the long-tail inventory item is lower than or equal to a preset shipment volume threshold, obtain the inventory item category corresponding to the long-tail inventory item, and obtain the first expected shipment result corresponding to the long-tail inventory item according to the inventory item category; When the shipment volume data corresponding to the long-tail inventory item is higher than a preset shipment volume threshold, the smoothed mean shipment volume corresponding to the long-tail inventory item is obtained, and the first expected shipment result corresponding to the long-tail inventory item is obtained according to the Poisson distribution corresponding to the smoothed mean.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 4 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 4 are implemented.

Images