Intelligent storage optimal scheduling method and system based on Internet of Things and AI

Abstract

The embodiment of the invention provides an intelligent storage optimal scheduling method and system based on the Internet of Things and AI, and the method comprises the steps: firstly, achieving the precise description of a complex relation between storage goods locations through the deep analysis of the layout and scheduling behavior data of a storage region; through difference tradeoff calculation and analysis, not only is the relevance between the storage goods locations quantified, but also the difference between the storage goods locations in the aspects of goods storage, carrying efficiency and the like is accurately evaluated, and a scientific basis is provided for subsequent optimization. Furthermore, a targeted storage optimization strategy can be automatically formulated based on the first storage scheduling association degree generated based on the analysis result, reasonable allocation and storage layout of the goods are effectively guided, the goods carrying distance and time are remarkably reduced, and the storage operation efficiency is improved. Therefore, the warehouse management is more flexible and efficient, the market demand change can be dynamically adapted, the operation cost is reduced, the resource utilization is optimized, and the competitiveness of the whole supply chain is improved.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, and more specifically, to an intelligent warehousing optimization scheduling method and system based on the Internet of Things and AI. Background Art

[0002] With the rapid development of e-commerce and logistics industries, warehouse management is facing unprecedented challenges. The traditional manual management model can no longer meet the requirements of modern warehouse systems for efficiency, accuracy, and flexibility. Especially in large-scale warehouse environments, the storage, sorting, and outbound operations of goods are complex and require high coordination and precise control. Therefore, how to optimize warehouse scheduling and improve warehouse operation efficiency and accuracy through intelligent means has become an urgent problem to be solved in the current warehouse management field.

[0003] In the existing technology, although some warehouse management systems based on information technology have emerged, most of these systems focus on the automation of a single link and lack comprehensive analysis and optimization of the overall layout of the warehouse area, the characteristics of cargo flow and scheduling behavior. In addition, these systems often lack flexibility and are difficult to adapt to rapidly changing market demands and business scenarios. Summary of the invention

[0004] In order to at least overcome the above-mentioned deficiencies in the prior art, the purpose of this application is to provide an intelligent warehousing optimization scheduling method and system based on the Internet of Things and AI.

[0005] In a first aspect, the present application provides an intelligent warehousing optimization scheduling method based on the Internet of Things and AI, which is applied to an intelligent warehousing optimization scheduling system based on the Internet of Things and AI, and the method includes: Acquire first warehouse knowledge network data to be optimized, and acquire second warehouse knowledge network data; the first warehouse knowledge network data and the second warehouse knowledge network data are generated by constructing knowledge networks of different IoT warehouse management depths for the same warehouse area, and the knowledge vector data in the second warehouse knowledge network data includes warehouse scheduling behavior data; Determine multiple first associated storage locations of storage location a and knowledge vector data of each first associated storage location based on the first storage knowledge network data, and determine the first knowledge vector data of the storage location a based on the second storage knowledge network data; a∈[1, A], A is the scale of the storage location in the second storage knowledge network data; Based on the knowledge vector data of the plurality of first associated storage locations, performing a difference tradeoff calculation and analysis on the first knowledge vector data of the storage location a to generate a target difference tradeoff calculation and analysis result; Based on the target difference trade-off calculation and analysis results, the first storage scheduling correlation between the storage location a and each first associated storage location is estimated, and the storage optimization strategy corresponding to the storage location a is generated by using the estimated first storage scheduling correlations; According to the storage optimization strategy corresponding to the storage location a, the first storage knowledge network data is optimized.

[0006] In the second aspect, an embodiment of the present application also provides an intelligent warehousing optimization and scheduling system based on the Internet of Things and AI, wherein the intelligent warehousing optimization and scheduling system based on the Internet of Things and AI comprises a processor and a machine-readable storage medium, wherein the machine-readable storage medium stores a computer program, and the computer program is loaded and executed in conjunction with the processor to implement the intelligent warehousing optimization and scheduling method based on the Internet of Things and AI of the first aspect above.

[0007] In combination with the above aspects, the embodiment of the present application significantly improves the intelligent level of warehouse management by integrating warehouse knowledge network data of different management depths. First, by deeply analyzing the layout and scheduling behavior data of the storage area, an accurate characterization of the complex relationship between storage locations is achieved. Through differential trade-off calculation and analysis, not only the correlation between each storage location is quantified, but also the differences between the storage locations in terms of cargo storage and handling efficiency are accurately evaluated, providing a scientific basis for subsequent optimization. Furthermore, based on the first warehouse scheduling correlation generated by the above analysis results, targeted warehouse optimization strategies can be automatically formulated to effectively guide the reasonable allocation and storage layout of goods, significantly reduce the distance and time of cargo handling, and improve warehouse operation efficiency. As a result, warehouse management is made more flexible and efficient, and can dynamically adapt to changes in market demand, reduce operating costs, and optimize resource utilization, thereby enhancing the competitiveness of the overall supply chain. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required to be used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.

[0009] Figure 1 A flow chart of an intelligent warehousing optimization scheduling method based on the Internet of Things and AI provided in an embodiment of the present application; Figure 2 A structural schematic block diagram of an intelligent warehousing optimization scheduling system based on the Internet of Things and AI for implementing the above-mentioned intelligent warehousing optimization scheduling method based on the Internet of Things and AI provided in an embodiment of the present application. DETAILED DESCRIPTION

[0010] The following description is intended to enable one of ordinary skill in the art to implement and utilize the present application, and the description is provided in the context of a specific application scenario and its requirements. It will be apparent to one of ordinary skill in the art that various changes may be made to the disclosed embodiments, and that the general principles defined in the present application may be applied to other embodiments and application scenarios without departing from the principles and scope of the present application. Therefore, the present application is not limited to the described embodiments, but should be given the broadest scope equal to the claims.

[0011] Step S110, obtaining first warehouse knowledge network data to be optimized and obtaining second warehouse knowledge network data. The first warehouse knowledge network data and the second warehouse knowledge network data are generated by constructing knowledge networks of different IoT warehouse management depths for the same warehouse area, and the knowledge vector data in the second warehouse knowledge network data includes warehouse scheduling behavior data.

[0012] In detail, the first warehousing knowledge network data is a knowledge network data with a shallow depth of basic construction and management of the warehousing area, which mainly contains the basic layout information of the warehousing cargo space, such as the location and size of the cargo space, but may not contain detailed operation data or scheduling behavior information. The second warehousing knowledge network data is a knowledge network data for deeper management of the same warehousing area, which contains not only basic layout information but also detailed warehousing scheduling behavior data, such as the entry and exit time of goods, transportation path, equipment scheduling, etc.

[0013] The knowledge vector data refers to representing each entity (such as storage location) as a multi-dimensional vector in the knowledge network, and each dimension represents a specific knowledge attribute (such as storage capacity, frequency of entry and exit, type of goods, etc.).

[0014] The storage dispatching behavior data is detailed record data about dispatching behaviors such as cargo handling, storage, and distribution in the storage area.

[0015] For example, in this embodiment, for a large e-commerce storage area, the e-commerce storage area stores a variety of commodities, ranging from electronic products to daily necessities. The server is responsible for data processing and optimization of the entire storage management system.

[0016] To obtain the first warehouse knowledge network data, the server extracts information from the basic data storage module of the warehouse management system. This basic data storage module contains the basic layout information of the storage locations, such as the coordinate position of each location, the size of the location, etc. The first warehouse knowledge network data constructed from this information is relatively basic, and its management depth is relatively shallow, only describing basic features such as the physical layout of the storage area. For example, the information obtained by the server shows that location 1 is located in the northwest corner of the warehouse, with an area of ​​10 square meters; location 2 is located in the northeast corner of the warehouse, with an area of ​​8 square meters, etc.

[0017] For the acquisition of the second warehouse knowledge network data, the server obtains data from a module that contains more detailed operational information. This module records warehouse scheduling behavior data, such as the time of goods entering and leaving the warehouse, the scheduling path of handling equipment, and the work efficiency of operators. Taking the time of goods entering and leaving the warehouse as an example, the server obtained that the time of goods entering the warehouse at location 3 every day in the past month was concentrated between 9:00 and 11:00 in the morning, and the time of goods leaving the warehouse was concentrated between 3:00 and 5:00 in the afternoon. The scheduling path data of the handling equipment shows that when the forklift moves the goods from location 4 to location 5, it needs to pass through specific channels and turning points. These detailed information constructs the second warehouse knowledge network data, which has a deeper management depth for the warehouse area because it involves the actual scheduling behavior in the warehouse operation process.

[0018] Both types of data are knowledge network data built for the same warehouse area, but they differ in management depth and information richness. The first warehouse knowledge network data is like a simple map with only locations marked, while the second warehouse knowledge network data is like a map with traffic flow and driving rules, containing more operational dynamic information.

[0019] Step S120, determining a plurality of first associated storage locations of storage location a and knowledge vector data of each first associated storage location based on the first storage knowledge network data, and determining first knowledge vector data of the storage location a based on the second storage knowledge network data. a∈[1, A], A is the size of the storage location in the second storage knowledge network data.

[0020] The associated storage locations refer to other storage locations in the knowledge network that are associated with a specific storage location (such as location a) in terms of physical location, goods flow or feature similarity.

[0021] For example, assuming A=100, that is, the number of storage locations in the second storage knowledge network data is 100. The server needs to determine multiple first associated storage locations of storage location a (assuming a=5) and their knowledge vector data, and determine the first knowledge vector data of storage location a based on the second storage knowledge network data.

[0022] First, the server processes the first warehouse knowledge network data. Since the feature detail granularity of the first warehouse knowledge network data is smaller than that of the second warehouse knowledge network data and is in the knowledge representation domain, the server needs to perform feature enhancement prediction processing on the first warehouse knowledge network data. For example, the server may use a machine learning algorithm, such as a feature enhancement model based on deep learning. This model supplements the cargo location features in the first warehouse knowledge network data by analyzing patterns in historical warehouse data. For example, originally only the area and location of the cargo location were known, but after feature enhancement prediction processing, the potential storage capacity of the cargo location (based on factors such as the usage of surrounding cargo locations) can be predicted, thereby generating the third warehouse knowledge network data, so that its feature detail granularity is equal to the feature detail granularity of the second warehouse knowledge network data.

[0023] The server obtains the knowledge mapping toolset, which can be a collection of pre-built data mapping rules and algorithms. For example, it contains algorithms for converting data in different formats into a unified format, and rules for mapping different semantic concepts. At the same time, the server obtains the third-party warehouse knowledge network data after feature enhancement and prediction processing.

[0024] Then, the server generates a source knowledge graph in the knowledge mapping tool set based on the second warehouse knowledge network data. This source knowledge graph is constructed with warehouse storage locations as nodes and warehouse scheduling behaviors (such as cargo handling relationships) as edges. For example, in the source knowledge graph, there is an edge between storage locations 5 and 6, indicating that cargo is often moved from storage location 5 to storage location 6. At the same time, the target knowledge framework in the knowledge mapping tool set is generated based on the third warehouse knowledge network data. This target knowledge framework may be constructed based on the enhanced storage location features.

[0025] Next, the server uses the knowledge mapping toolset to map the source knowledge graph to the knowledge representation domain of the target knowledge framework to generate the fourth warehouse knowledge network data. For example, the warehouse scheduling behavior information in the second warehouse knowledge network data is matched and integrated with the enhanced cargo location features in the third warehouse knowledge network data.

[0026] In the knowledge representation domain, the server performs knowledge vector adaptation on the fourth warehouse knowledge network data and the third warehouse knowledge network data. Assuming that in the third warehouse knowledge network data, an associated cargo location of cargo location 5 is cargo location 7, they should also be adapted to each other in the fourth warehouse knowledge network data. The server performs adaptation by comparing the characteristics of the two, such as the storage capacity of cargo location 7 in the third warehouse knowledge network data and the convenience of transportation in the fourth warehouse knowledge network data.

[0027] After knowledge vector adaptation, the server determines multiple first associated storage locations of storage location a (location 5) in the third storage knowledge network data. For example, by analyzing the flow of goods around location 5 and the characteristics of the location, it is determined that locations 3, 7, and 8 are its first associated storage locations. Then the knowledge vector data of each first associated storage location is obtained. For example, the knowledge vector data of location 3 may include its stored goods type (electronic products), current storage volume (50 pieces), frequency of goods entering and leaving the warehouse (3 times a week), etc.; the knowledge vector data of location 7 may include the type of stored goods (daily necessities), current storage volume (80 pieces), frequency of goods entering and leaving the warehouse (5 times a week), etc. At the same time, the knowledge vector data of storage location a (location 5) is obtained from the fourth storage knowledge network data as its first knowledge vector data. It is assumed that the first knowledge vector data of location 5 includes the type of stored goods (clothing), current storage volume (60 pieces), frequency of goods entering and leaving the warehouse (4 times a week), etc.

[0028] Step S130, performing a difference tradeoff calculation and analysis on the first knowledge vector data of the storage location a based on the knowledge vector data of the plurality of first associated storage locations, and generating a target difference tradeoff calculation and analysis result.

[0029] Difference weight calculation and analysis is to compare and analyze the knowledge vector data of two or more storage locations, calculate the differences between them and assign different weights to reflect the importance and impact of these differences.

[0030] For example, in this embodiment, it is assumed that the server has determined the three first associated storage locations (location 3, location 7, and location 8) of storage location a (location 5) and their knowledge vector data, and obtained the first knowledge vector data of location 5. Now, based on the knowledge vector data of these three first associated storage locations, a difference tradeoff calculation analysis is performed on the first knowledge vector data of location 5 to generate a target difference tradeoff calculation analysis result.

[0031] Assume that Y=5 differential trade-off calculations and analyses are performed. First, cluster each knowledge dimension in the knowledge vector data. Assume that each knowledge vector data includes vector values ​​under Z=6 knowledge dimensions, namely, the type of stored goods, the current storage volume, the frequency of goods entering and leaving the warehouse, the average residence time of goods, the distance between the storage location and the entrance and exit, and the utilization rate of equipment surrounding the storage location. Cluster these 6 knowledge dimensions into Q=3 knowledge dimension clusters. The first knowledge dimension cluster includes the type of stored goods and the average residence time of goods. The second knowledge dimension cluster includes the current storage volume and the distance between the storage location and the entrance and exit. The third knowledge dimension cluster includes the frequency of goods entering and leaving the warehouse and the utilization rate of equipment surrounding the storage location.

[0032] For the first difference trade-off calculation and analysis (d=1), the server determines the first knowledge dimension cluster from the three knowledge dimension clusters as the target knowledge dimension cluster. Then, each knowledge dimension in this target knowledge dimension cluster is polled. First, the knowledge dimension of the type of stored goods is polled. Based on the vector value of each first associated storage location (location 3, location 7, and location 8) under this knowledge dimension, the server performs a difference trade-off calculation and analysis on the vector value of storage location a (location 5) under this knowledge dimension. Assuming that location 3 stores electronic products and location 5 stores clothing, the server calculates a difference value based on the pre-set difference trade-off rules for the types of goods (for example, electronic products and clothing have different priorities in warehouse management, and electronic products may have a higher priority due to reasons such as fragility). Then, the knowledge dimension of the average residence time of goods is polled. Assuming that the average residence time of goods in location 3 is 2 days and the average residence time of goods in location 5 is 3 days, the difference value is calculated according to the set calculation method (such as the absolute value of the difference, etc.), and then the difference values ​​under these two knowledge dimensions are accumulated to generate the first difference trade-off calculation parameter.

[0033] In the same way, the second (d=2) to fifth (d=5) difference tradeoff calculation analysis is performed, and each time a different target knowledge dimension cluster is selected for operation. For example, in the second difference tradeoff calculation analysis, the second knowledge dimension cluster is selected to calculate the difference tradeoff calculation parameters between the cargo location 5 and each first associated storage cargo location under the two knowledge dimensions of current storage capacity and distance from the cargo location to the entrance and exit.

[0034] Finally, based on these five difference trade-off calculation parameters, the target difference trade-off calculation analysis result is constructed. This result may be an array containing five values, each of which represents the result of a difference trade-off calculation analysis, which reflects the difference between storage location a and the first associated storage location in different knowledge dimension clusters.

[0035] Step S140, based on the target difference trade-off calculation and analysis results, estimate the first storage scheduling correlation between the storage location a and each first associated storage location, and use the estimated first storage scheduling correlations to generate a storage optimization strategy corresponding to the storage location a.

[0036] The storage scheduling correlation is estimated by methods such as difference trade-off calculation and analysis, and represents the closeness and mutual influence relationship between two storage locations during the scheduling process.

[0037] The warehouse optimization strategy is generated based on data such as difference trade-off calculation analysis and warehouse scheduling correlation, and is designed to improve warehouse operation efficiency, reduce costs, and optimize resource utilization.

[0038] For example, in this embodiment, it is assumed that the matrix of the storage optimization strategy is X×X (assuming X=5), and the number of first associated storage locations is M (M=X²=25). Based on the previously generated target difference trade-off calculation and analysis results, the server estimates the first storage scheduling association between storage location a (location 5) and each first associated storage location.

[0039] The server calls the knowledge dimension optimizer, which can be an algorithm module based on rules or machine learning. For example, it can be a module using the principal component analysis (PCA) algorithm to optimize the knowledge dimension of the target difference trade-off calculation analysis results. Assume that the target difference trade-off calculation analysis results include the difference trade-off calculation parameters of 5 knowledge dimensions, and after being processed by the knowledge dimension optimizer, 25-dimensional data (M-dimensional data) is generated.

[0040] The server uses the i-th dimension data (i∈[1,25]) of the 25-dimensional data as the first storage scheduling correlation between storage location a (location 5) and the i-th first associated storage location. For example, if the third dimension data is 0.6, it means that the first storage scheduling correlation between location 5 and the third first associated storage location (assuming it is location 7) is 0.6. This means that there is a certain degree of correlation between location 5 and location 7 in terms of storage scheduling, and the higher the value, the stronger the correlation.

[0041] Then, the server uses the estimated correlation degree of each first storage scheduling to generate the storage optimization strategy corresponding to storage location a (location 5). First, the server performs smoothing and filtering on the second storage knowledge network data. For example, the moving average method is used to smooth the data with large fluctuations such as the time of goods entering and leaving the warehouse in the second storage knowledge network data to generate candidate storage knowledge network data. Determine M (25) second associated storage locations of storage location a (location 5) in the candidate storage knowledge network data. Assume that location 6, location 9, etc. are the second associated storage locations of location 5 in the candidate storage knowledge network data.

[0042] The server obtains the second storage scheduling association between storage location a (location 5) and each second associated storage location. For example, for location 6, the server determines the second knowledge vector data of location 5 (such as the storage volume after smoothing, the frequency of goods entering and leaving the warehouse, etc.) based on the second storage knowledge network data, and determines the knowledge vector data of location 6, and then performs a difference trade-off calculation and analysis on the second knowledge vector data of location 5 based on the knowledge vector data of location 6, generates a candidate difference trade-off calculation and analysis result, and then estimates the second storage scheduling association between location 5 and location 6 based on this result, assuming it is 0.5.

[0043] The server integrates the i-th second storage scheduling correlation with the estimated i-th first storage scheduling correlation. For example, for storage location 6, its second storage scheduling correlation of 0.5 is integrated with the corresponding first storage scheduling correlation (assuming it is 0.4), such as using a weighted average method (the weight can be set according to the actual situation, such as the weight of the first storage scheduling correlation is 0.6, and the weight of the second storage scheduling correlation is 0.4), to generate the i-th global storage scheduling correlation corresponding to storage location a (storage location 5). According to this method, all 25 associated storage locations are calculated, and finally, based on these 25 global storage scheduling correlations, a storage optimization strategy corresponding to storage location a (storage location 5) is generated. This storage optimization strategy may be in the form of a matrix, in which each element represents the optimization relationship between storage location 5 and other associated storage locations, such as which storage locations should be prioritized for goods allocation, etc.

[0044] Step S150: performing storage optimization on the first storage knowledge network data according to the storage optimization strategy corresponding to the storage location a.

[0045] In this embodiment, it is assumed that the server has generated a storage optimization strategy corresponding to storage location a (storage location 5). Now, based on this storage optimization strategy, storage optimization is performed on the first storage knowledge network data.

[0046] The server first obtains the allocation migration information a corresponding to the storage location a (location 5), assuming that the allocation migration information a includes M (25) allocation migration parameters. For example, these allocation migration parameters may represent the allocation quantity and allocation direction of goods between location 5 and other related locations.

[0047] The server maps storage location a (location 5) to the first storage knowledge network data, and generates a corresponding position corresponding to storage location a. For example, the basic layout information corresponding to location 5 is found in the first storage knowledge network data to determine its actual physical location in the warehouse.

[0048] According to the corresponding position and the 25 allocation migration parameters in the allocation migration information a, the 25 third associated storage locations of storage location a (location 5) in the first storage knowledge network data and the knowledge vector data of each third associated storage location are determined. Assuming that a allocation migration parameter in the allocation migration information a indicates that location 5 should allocate 10 pieces of goods to a location in a certain direction, the server will find the location in this direction (assuming it is location 10) as the third associated storage location in the first storage knowledge network data based on this parameter, and obtain the knowledge vector data of location 10, including its original storage capacity, cargo type and other information.

[0049] Finally, the server optimizes the knowledge vector data of the 25 third associated storage locations based on the storage optimization strategy corresponding to storage location a (location 5). For example, if the storage optimization strategy indicates that the frequency of goods allocation should be increased between location 5 and location 10, the server updates the knowledge vector data of location 10 in the first storage knowledge network data and adjusts the value of its goods in and out frequency (for example, from 3 times a week to 5 times a week), thereby achieving storage optimization of the first storage knowledge network data. This helps to improve the operational efficiency of the entire storage area, such as reducing the distance of goods transportation and improving the rationality of goods storage.

[0050] Based on the above steps, the embodiment of the present application significantly improves the intelligent level of warehouse management by integrating warehouse knowledge network data of different management depths. First, by deeply analyzing the layout and scheduling behavior data of the storage area, an accurate characterization of the complex relationship between storage locations is achieved. Through differential trade-off calculation and analysis, not only the correlation between each storage location is quantified, but also the differences between the storage locations in terms of cargo storage and handling efficiency are accurately evaluated, providing a scientific basis for subsequent optimization. Furthermore, based on the first warehouse scheduling correlation generated by the above analysis results, targeted warehouse optimization strategies can be automatically formulated to effectively guide the reasonable allocation and storage layout of goods, significantly reduce the distance and time of cargo handling, and improve warehouse operation efficiency. As a result, warehouse management is made more flexible and efficient, and can dynamically adapt to changes in market demand, reduce operating costs, and optimize resource utilization to enhance the competitiveness of the overall supply chain.

[0051] In a possible implementation, the feature detail granularity of the first warehouse knowledge network data is smaller than the feature detail granularity of the second warehouse knowledge network data, and the first warehouse knowledge network data is in the knowledge representation domain.

[0052] Step S120 includes: Step S121, performing feature enhancement prediction processing on the first warehouse knowledge network data to generate third warehouse knowledge network data, wherein the feature detail granularity of the third warehouse knowledge network data is equal to the feature detail granularity of the second warehouse knowledge network data.

[0053] In this embodiment, in the e-commerce warehousing scenario mentioned above, the server is responsible for processing the warehousing knowledge network data. The first warehousing knowledge network data has a small feature detail granularity. For example, it simply records the basic information of each storage location, such as the location number, the approximate area range (such as small location, medium location, large location) and the approximate area (such as area A, area B, etc.).

[0054] To perform feature enhancement prediction processing, the server uses a method based on historical data and machine learning algorithms. It collects warehouse operation data from the past years, including information such as cargo type, cargo flow, and the frequency of cargo in and out of the warehouse in different time periods. Then, the server uses an integrated learning algorithm, such as the random forest algorithm, to build a feature enhancement prediction model.

[0055] Take cargo location 1 as an example. The original data of the first warehouse knowledge network only shows that it is a medium-sized cargo location located in area A. Through the feature enhancement prediction model, the server analyzes the storage of goods in other cargo locations similar to cargo location 1 (based on similarity judgment based on features such as region and area) in historical data. The model found that in the past, cargo locations of similar location and size often stored a certain type of goods (such as small accessories in electronic products) in specific seasons (such as e-commerce promotion seasons), and the frequency of goods entering and leaving the warehouse would have a specific change pattern. Based on these analyses, the server enhances the features of cargo location 1, adding detailed information such as the types of goods that may be stored (such as small accessories in electronic products) and the expected frequency of goods entering and leaving the warehouse in different seasons.

[0056] After such processing is performed on each cargo location in the first warehousing knowledge network data, the third warehousing knowledge network data is generated. At this time, the feature detail granularity of the third warehousing knowledge network data is equal to the feature detail granularity of the second warehousing knowledge network data. For example, cargo location 1 in the third warehousing knowledge network data now contains detailed information such as cargo type tendencies and cargo flow forecasts in different time periods, which has a similar level of detail as the detailed warehousing scheduling behavior data contained in the second warehousing knowledge network data.

[0057] Step S122, obtaining a knowledge mapping toolset, and obtaining third warehouse knowledge network data generated by performing feature enhancement prediction processing on the first warehouse knowledge network data.

[0058] In this embodiment, the server has a dedicated module in the storage system for storing the knowledge mapping toolset. This knowledge mapping toolset is pre-built and maintained by data engineers. It includes a series of data conversions, semantic mappings, and rule definitions. For example, there are tools for converting cargo type codes in different formats into a unified format, and there are rules for uniformly mapping different naming methods of storage areas (such as some areas are called "North Zone 1" and some are called "N1 Zone").

[0059] At the same time, the server obtains the generated third-party warehouse knowledge network data from the previous processing results. This acquisition process is achieved through the index and query mechanism in the data storage system. The server accurately locates and reads the third-party warehouse knowledge network data according to the pre-set identification and index structure. For example, the data storage system may use the cargo location number as part of the index. The server obtains the corresponding third-party warehouse knowledge network data containing detailed features by querying the index entry related to the cargo location number.

[0060] Step S123, generating a source knowledge graph in the knowledge mapping tool set based on the second warehouse knowledge network data, and generating a target knowledge framework in the knowledge mapping tool set based on the third warehouse knowledge network data.

[0061] For the construction of the source knowledge graph, the server is based on the second warehouse knowledge network data. The second warehouse knowledge network data contains rich warehouse scheduling behavior data. For example, for each warehouse storage location, there is information such as which supplier the goods entered the location from, when they entered, which handling equipment was used to transport them, and which processing area or shipping area they were transported to.

[0062] The server uses each storage location as a node and the goods transportation path, the supply relationship between suppliers and the storage location, etc. as edges to construct the source knowledge graph. Taking storage location 3 as an example, the goods are transported from supplier A to storage location 3 by forklift at 10 am, and then transported to processing area B at 2 pm. In the source knowledge graph, there will be supplier A nodes, storage location 3 nodes, processing area B nodes, and edges from supplier A to storage location 3 and from storage location 3 to processing area B. The attributes of the edges may include information such as transportation time and transportation equipment.

[0063] For the construction of the target knowledge framework, the server uses the third-party warehouse knowledge network data as the basis. Since the third-party warehouse knowledge network data contains enhanced cargo location features, such as predicted cargo type, cargo flow in different time periods, etc., the server takes each cargo location as an element in the framework and the characteristics of the cargo location as the attributes of the framework element to construct the target knowledge framework. For example, the attributes of cargo location 5 in the target knowledge framework may include the predicted cargo type is daily necessities, the cargo flow in the promotion season is 100-200 pieces per day, etc.

[0064] Step S124, using the knowledge mapping toolset, mapping the source knowledge graph to the knowledge representation domain where the target knowledge framework is located, and generating fourth warehouse knowledge network data.

[0065] The server starts to operate using the knowledge mapping toolset. For example, there is a rule in the knowledge mapping toolset that unifies the different names of cargo handling equipment. In the source knowledge graph, handling equipment may have different names such as "Forklift No. 1" and "Forklift-A", while in the target knowledge framework, forklifts are uniformly named "Standard Forklift". Through this mapping rule, the server converts the names of handling equipment in the source knowledge graph to make them consistent with the naming method in the target knowledge framework.

[0066] Another rule may be about the semantic mapping of goods types. The goods types in the source knowledge graph may use some industry abbreviations, while the goods types in the target knowledge framework use more detailed and standardized names. The server converts the goods types in the source knowledge graph into the corresponding names in the target knowledge framework through the semantic mapping tool in the knowledge mapping tool set.

[0067] By performing a series of such conversion and mapping operations on the nodes, edges and their attributes in the source knowledge graph, the server maps the source knowledge graph to the knowledge representation domain where the target knowledge framework is located, thereby generating the fourth warehouse knowledge network data. This fourth warehouse knowledge network data integrates the warehouse scheduling behavior relationship in the source knowledge graph and the cargo location feature information in the target knowledge framework, forming a more comprehensive warehouse knowledge network data under the same knowledge representation domain.

[0068] Step S125, in the knowledge representation domain, performing knowledge vector adaptation on the fourth warehousing knowledge network data and the third warehousing knowledge network data, so that two storage locations in the same warehousing node in the fourth warehousing knowledge network data and the third warehousing knowledge network data are adapted to each other.

[0069] In the knowledge representation domain, the server performs knowledge vector adaptation on the fourth warehouse knowledge network data and the third warehouse knowledge network data. For example, for cargo location 7, in the third warehouse knowledge network data, its knowledge vector may include information such as the predicted storage volume of 80 pieces and the average residence time of the goods is 3 days; in the fourth warehouse knowledge network data, its knowledge vector may include information such as the source supplier of the goods and the estimated time to transport to the next area.

[0070] The server makes the cargo location 7 in the two data adapt to each other by establishing an adaptation rule. For example, an association rule is established based on the relationship between the cargo storage volume and the cargo source supplier. If the cargo source supplier is a large and stable supplier, the cargo storage volume may be relatively large and stable. According to this rule, the server adjusts the knowledge vectors of cargo location 7 in the fourth warehouse knowledge network data and the third warehouse knowledge network data so that they logically match each other.

[0071] For other storage locations, the server also uses a similar method to establish adaptation rules based on different business logics and data relationships, and adjusts the knowledge vector to ensure that the two storage locations in the same storage node in the fourth storage knowledge network data and the third storage knowledge network data can adapt to each other in all aspects.

[0072] Step S126, after performing knowledge vector adaptation, determine the knowledge vector data of multiple first associated storage locations of storage location a in the third storage knowledge network data and each first associated storage location, and obtain the knowledge vector data of storage location a from the fourth storage knowledge network data as the first knowledge vector data of storage location a.

[0073] In this embodiment, it is assumed that storage location a is location 10. In the third storage knowledge network data after knowledge vector adaptation, the server determines multiple first associated storage locations of location 10. The server determines the associated locations based on factors such as the flow relationship of goods and the characteristic similarity of the locations. For example, from the perspective of the flow relationship of goods, if goods are often moved from location 8 to location 10, and then from location 10 to location 12, then locations 8 and 12 may be determined as the first associated storage locations of location 10. From the perspective of the characteristic similarity of the locations, if location 10 is a medium-sized location for storing daily necessities, and location 9 is also a medium-sized location with a tendency to store daily necessities, then location 9 may also be determined as the first associated storage location.

[0074] For each first associated storage location, the server obtains its knowledge vector data. Taking location 8 as an example, its knowledge vector data may include information such as the current storage capacity is 50 pieces, the frequency of goods entering and leaving the warehouse is 3 times a week, and the average residence time of goods is 2 days.

[0075] At the same time, the server obtains the knowledge vector data of the cargo location 10 (storage cargo location a) from the fourth warehouse knowledge network data as its first knowledge vector data. For example, the knowledge vector data of the cargo location 10 in the fourth warehouse knowledge network data may include the transportation path of the goods (from which cargo location the goods are transported and to which cargo location the goods are transported), the source supplier information of the goods (such as supplier B), etc., and this knowledge vector data is determined as the first knowledge vector data of the cargo location 10.

[0076] In a possible implementation manner, after performing knowledge vector adaptation, the method further includes: Step A110, performing standardization conversion on the third warehouse knowledge network data and the fourth warehouse knowledge network data respectively to generate the third warehouse knowledge network data after standardization conversion and the fourth warehouse knowledge network data after standardization conversion.

[0077] In the e-commerce warehousing scenario, the server starts to perform standardized conversion operations for the third-party warehousing knowledge network data. Taking the storage location as an example, it is assumed that the location characteristics in the third-party warehousing knowledge network data include information such as the storage volume of goods, the frequency of goods entering and leaving the warehouse, and the average residence time of goods.

[0078] For the cargo storage capacity, the server first obtains the cargo storage capacity data of all cargo locations. Assume that there are 100 cargo locations in the entire third warehousing knowledge network data, and the storage capacity ranges from 0 to 1000 pieces. The server calculates the mean and standard deviation of these storage capacity data. For example, after calculation, the mean is 300 pieces and the standard deviation is 150 pieces. Then, for the cargo storage capacity of each cargo location, the formula: standardized storage capacity = (original storage capacity-mean) / standard deviation is used for conversion. For example, the original storage capacity of a cargo location is 450 pieces. After calculation, the standardized storage capacity = (450-300) / 150=1.

[0079] For the frequency of goods in and out of the warehouse, assuming that the original data ranges from 0 to 10 times per week, after similar calculation of the mean and standard deviation (assuming the mean is 3 times and the standard deviation is 2 times), the frequency of goods in and out of the warehouse for each cargo location is standardized.

[0080] For the average residence time of goods, the same method is used. Assuming that the original data ranges from 1 to 10 days, the mean and standard deviation are calculated and then standardized. In this way, the entire third-tier warehouse knowledge network data is standardized and the standardized third-tier warehouse knowledge network data is generated.

[0081] For the fourth warehousing knowledge network data, it contains information related to the cargo handling path (such as handling distance, handling time, etc.) and information related to the cargo source supplier (such as the supplier's supply stability score, etc.). Taking the handling distance as an example, if the handling distance in the original data ranges from 10 meters to 100 meters, the server calculates the mean and standard deviation of all handling distance data (assuming the mean is 50 meters and the standard deviation is 20 meters), and then converts it according to the formula: standardized handling distance = (original handling distance-mean) / standard deviation. For the supplier's supply stability score (assuming it ranges from 1 to 5 points), the mean and standard deviation are also calculated and then standardized. By analogy, the standardized conversion of the fourth warehousing knowledge network data is completed, and the fourth warehousing knowledge network data after standardized conversion is generated.

[0082] Step A120, obtaining a conversion coefficient shared between the third warehouse knowledge network data and the fourth warehouse knowledge network data.

[0083] When the server obtains the conversion coefficient, it needs to consider the association between the third warehouse knowledge network data and the fourth warehouse knowledge network data. Since both data are related to the warehouse and have undergone operations such as knowledge vector adaptation before, there are some common factors between them.

[0084] The server first analyzes the common features in the two data sets. For example, the storage volume of goods is involved in both data sets, although in the third warehouse knowledge network data it is from the perspective of its own cargo location characteristics, in the fourth warehouse knowledge network data it may be related to the transportation path (such as determining the transportation distance based on the storage volume). The server calculates the conversion coefficient based on this common feature.

[0085] Assume that the server calculates the conversion coefficient through a method based on linear regression. It takes the storage volume of goods in the third warehouse knowledge network data as the independent variable, and takes a certain indicator in the handling path related to the storage volume in the fourth warehouse knowledge network data (such as the load rate of the handling equipment, because the storage volume will affect the load of the handling equipment) as the dependent variable. By performing linear regression analysis on a large amount of sample data (data of multiple cargo locations), the coefficient of the regression equation is obtained. This coefficient is part of the conversion coefficient between the two data sets regarding the feature of the storage volume of goods.

[0086] The server also performs similar analysis on other common features (such as the relationship between the frequency of goods entering and leaving the warehouse and the time of goods handling, etc.), and combines these analysis results to obtain the conversion coefficient shared between the third warehouse knowledge network data and the fourth warehouse knowledge network data. This conversion coefficient may be a vector or matrix containing multiple elements, each element corresponding to the conversion relationship of different common features.

[0087] Step A130: performing linear conversion on the standardized third warehouse knowledge network data and the standardized fourth warehouse knowledge network data respectively according to the conversion coefficient.

[0088] For the third warehousing knowledge network data after standardization conversion, the server performs linear conversion according to the conversion coefficient obtained. Taking cargo location 1 as an example, it is assumed that the cargo storage volume of cargo location 1 after standardization conversion is 1.5, and the standardized value of cargo in and out frequency is 0.8. The coefficient corresponding to the cargo storage volume in the conversion coefficient is 2, and the coefficient corresponding to the cargo in and out frequency is 3.

[0089] For the storage volume of goods, the result after linear transformation is: 1.5×2=3; for the frequency of goods entering and leaving the warehouse, the result after linear transformation is: 0.8×3=2.4. The server performs linear transformation on each feature of each cargo location in the third warehousing knowledge network data in this way.

[0090] For the fourth warehouse knowledge network data after standardization, linear conversion is also performed according to the conversion coefficient. For example, for the standardized value of the handling distance of cargo location 1 in the fourth warehouse knowledge network data is -0.5, the standardized value of the supplier supply stability score is 1.2, the coefficient corresponding to the handling distance in the conversion coefficient is 1.5, and the coefficient corresponding to the supplier supply stability score is 0.8.

[0091] For the transport distance, the result after linear transformation is: -0.5×1.5=-0.75; for the supplier supply stability score, the result after linear transformation is: 1.2×0.8=0.96. The server performs such a linear transformation operation on the relevant features of each cargo location in the fourth warehouse knowledge network data.

[0092] Step A140, after performing the linear conversion, jumps to the step of determining the knowledge vector data of multiple first associated storage locations of storage location a in the third storage knowledge network data and each first associated storage location, and obtaining the knowledge vector data of the storage location a from the fourth storage knowledge network data as the first knowledge vector data of the storage location a.

[0093] After the above linear conversion, the server returns to the task of determining the multiple first associated storage locations of storage location a (assuming that storage location a is location 5) in the third storage knowledge network data.

[0094] The server re-analyzes factors such as the goods flow relationship and feature similarity of location 5. From the perspective of the goods flow relationship, since the characteristics such as the frequency of goods entering and leaving the warehouse in the third warehousing knowledge network data after linear transformation have changed, the server re-evaluates the flow direction of the goods based on the new frequency data. For example, the original flow of goods from location 3 to location 5 may be based on the previous frequency of entering and leaving the warehouse, but after linear transformation, the new frequency of entering and leaving the warehouse may change this flow relationship. If the new frequency shows that the flow of goods from location 4 to location 5 is more frequent, then location 4 is more likely to be determined as the first associated storage location of location 5.

[0095] From the perspective of feature similarity, for example, the cargo storage volume of cargo location 5 after linear transformation is a new value, and the server determines feature similarity based on the comparison of this new value with the cargo storage volume of other cargo locations. If the cargo storage volume of cargo location 6 after linear transformation is similar to that of cargo location 5, and other features such as cargo type are also similar, then cargo location 6 may also be determined as the first associated storage location of cargo location 5.

[0096] For each determined first associated storage location, the server obtains its knowledge vector data. Taking location 4 as an example, its knowledge vector data now contains information such as the cargo storage volume after linear transformation, the cargo in and out frequency, etc. For example, the cargo storage volume after linear transformation is 4 (obtained through the previous calculation), the cargo in and out frequency is 3 (also the result after linear transformation), etc.

[0097] At the same time, the server obtains the knowledge vector data of cargo location 5 (storage cargo location a) from the fourth warehouse knowledge network data after linear transformation as its first knowledge vector data. For example, the knowledge vector data of cargo location 5 in the fourth warehouse knowledge network data now contains information such as the transportation distance after linear transformation and the supplier supply stability score. If the transportation distance after linear transformation is 2 meters, the supplier supply stability score is 0.9, etc., this knowledge vector data is determined as the first knowledge vector data of cargo location 5.

[0098] In a possible implementation, the matrix of the storage optimization strategy is X×X, the plurality of first associated storage locations are M first associated storage locations, X is a positive integer, and M is equal to the square of X. Step A140 includes: Step A141, obtaining the allocation migration information a corresponding to the storage location a, wherein the allocation migration information a includes M allocation migration parameters.

[0099] In the e-commerce warehousing scenario, assume that the storage location a is location 10. The server obtains the allocation migration information a corresponding to location 10 from the database module that specifically stores the allocation migration information. Since M is equal to the square of X (assuming X=3, then M=9), the allocation migration information a contains 9 allocation migration parameters.

[0100] These transfer parameters are pre-calculated and set according to the overall planning and needs of the warehouse. For example, the first transfer parameter may represent the amount of goods to be transferred between shelf 10 and the first storage shelf in the target knowledge sub-network. If the warehouse receives a batch of new goods recently, and according to the sales forecast and inventory layout strategy, it plans to transfer a certain amount of goods from shelf 10 to the first storage shelf in the target knowledge sub-network, this amount is recorded in the first transfer parameter, assuming it is 50 pieces.

[0101] The second allocation migration parameter may be related to the allocation direction. For example, 0 means no allocation is required, 1 means allocation from storage location 10 to the target storage location, and -1 means allocation from the target storage location to storage location 10. If, according to the inventory flow requirements, goods need to be transferred from storage location 10 to the second storage location in the target knowledge subnetwork, then this parameter may be 1.

[0102] Other allocation migration parameters may also involve information such as the time schedule of allocation and the priority of goods. For example, the third allocation migration parameter indicates the time priority of allocation, and the higher the value, the higher the priority of allocation. If a certain type of goods needs to be allocated to the third storage location in the target knowledge subnetwork as soon as possible according to the sales activity schedule, this parameter may be set to a higher value, such as 8 (assuming the priority value range is 1-10).

[0103] Step A142, in the third storage knowledge network data, a target knowledge sub-network is determined with the storage location that is compatible with the storage location a as the center, and the target knowledge sub-network includes M storage locations.

[0104] The server operates in the third warehouse knowledge network data. Since the knowledge vector adaptation and other operations have been performed on the storage locations before, the adaptation relationship between the locations has been determined. For location 10 (storage location a), the server first finds a storage location that is compatible with location 10.

[0105] Assume that the storage location that matches the location 10 is the location 11. The target knowledge sub-network is determined with the location 11 as the center. The determination of this target knowledge sub-network is based on factors such as the layout relationship of the storage locations, the flow relationship of goods, and the similarity of the types of goods.

[0106] Based on the warehouse layout information, the server selects storage locations around location 11 to construct the target knowledge sub-network. For example, from the layout point of view, the eight locations around location 11 (assuming that the warehouse layout is in the form of a nine-square grid, with location 11 in the center) are given priority to join the target knowledge sub-network. At the same time, the server also considers the flow of goods. If there is frequent flow of goods between a location and location 11, it may be included in the target knowledge sub-network even if it is far away in layout. For example, although location 15 is not around the nine-square grid of location 11, since goods often flow between location 11 and location 15, location 15 is also included in the target knowledge sub-network.

[0107] According to the setting, the target knowledge subnetwork must contain M (9) storage locations. The server finally determines that these 9 storage locations constitute the target knowledge subnetwork by comprehensively considering factors such as layout, cargo flow, and cargo type similarity. For example, these 9 storage locations may include location 9, location 10, location 11, location 12, location 15, location 16, location 19, location 20, and location 21.

[0108] Step A143, based on the i-th allocation and migration parameter of the allocation and migration information a, the layout relationship of the storage node where the i-th storage location in the target knowledge subnetwork is located is allocated and migrated to generate a target storage node, i∈[1,M].

[0109] For i=1, the server obtains the first allocation migration parameter of the allocation migration information a. Assume that the first allocation migration parameter indicates that 50 pieces of goods are allocated from storage location 10 to the first storage location in the target knowledge subnetwork (assuming it is storage location 9), and the allocation direction is from storage location 10 to storage location 9.

[0110] The server performs a deployment migration process on the layout relationship of the storage node where the cargo location 9 is located according to the deployment migration parameter. In the third storage knowledge network data, the layout relationship of the storage node includes the location of the cargo location, the relationship between the cargo location and the surrounding equipment (such as the handling channel, shelf layout, etc.), etc. Since 50 pieces of goods need to be deployed to cargo location 9, the server needs to consider whether the remaining storage space of cargo location 9 is sufficient. If the current remaining space of cargo location 9 is 80 pieces, then this batch of goods can be accommodated.

[0111] The server also adjusts the relationship between cargo position 9 and its surrounding equipment. For example, if the goods are transported by forklift, the server needs to ensure that the forklift channel can meet the transportation needs after the goods are allocated. If the forklift channel was originally on the left side of cargo position 9, but the frequency of goods entering and exiting may increase after the goods are allocated, the server may adjust the layout of the forklift channel, widen it or change its direction to improve the transportation efficiency. In this way, the target storage node is generated. This target storage node is a storage layout structure composed of the relationship between cargo position 9 and its surrounding equipment after adjustment according to the allocation migration parameters.

[0112] For i=2, assume that the second allocation migration parameter indicates that a certain type of goods will be allocated from the second storage location (assuming it is location 12) in the target knowledge subnetwork to location 10, and there is a time priority requirement. The server performs allocation migration processing on the storage node where location 12 is located based on this parameter. The server first checks the storage situation of the goods in location 12 to determine the location and quantity of the specific type of goods to be allocated. Then, considering the time priority of the allocation, adjust the scheduling relationship between location 12 and the surrounding handling equipment (such as conveyor belts, forklifts, etc.) to ensure that the goods can be allocated to location 10 within the specified time. For example, adjust the running time of the conveyor belt or the scheduling order of the forklift to meet the allocation requirements, thereby generating the target storage node.

[0113] According to this method, the server performs allocation and migration processing on the storage nodes where the corresponding storage locations in the target knowledge subnetwork are located for each allocation and migration parameter i from 1 to M (9), and generates the corresponding target storage nodes.

[0114] Step A144, taking the storage location located at the target storage node in the third warehousing knowledge network data as the i-th first associated storage location of storage location a in the third warehousing knowledge network data, and obtaining the knowledge vector data of the i-th first associated storage location from the third warehousing knowledge network data.

[0115] In this embodiment, for i=1, after the previous allocation and migration process, the target storage node is a structure composed of the adjusted cargo location 9 and its peripheral equipment relationship, etc. The server uses cargo location 9 located at this target storage node as the first first associated storage location of storage location a (cargo location 10) in the third storage knowledge network data.

[0116] Then, the server obtains the knowledge vector data of the cargo location 9 from the third warehouse knowledge network data. This knowledge vector data contains various characteristic information of the cargo location 9, such as the cargo storage volume, cargo type, cargo in and out frequency, etc. Assume that the cargo storage volume of cargo location 9 is 30 pieces, the cargo type is electronic products, and the cargo in and out frequency is 2 times a week.

[0117] For i=2, the target storage node is a structure composed of the adjusted cargo location 12 and its peripheral equipment relationships. The server uses cargo location 12 as the second first associated storage cargo location of cargo location 10. The knowledge vector data of cargo location 12 is obtained from the third storage knowledge network data, for example, the cargo storage capacity of cargo location 12 is 60 pieces, the cargo type is daily necessities, and the cargo in and out frequency is 3 times a week.

[0118] According to this method, for i from 1 to M (9), the server takes the storage location located at the target storage node as the corresponding first associated storage location of location 10, and obtains their knowledge vector data.

[0119] In a possible implementation, the allocation migration information a corresponding to the storage location a is the ath allocation migration information in the target allocation migration information network. Before acquiring the allocation migration information a corresponding to the storage location a, the method further includes: Step B110, polling each storage location in the first storage knowledge network data, and determining the association of the currently polled storage location b in the first storage knowledge network data. The association of the storage location b includes M storage locations, and b is a positive integer and is not greater than the size of the storage locations in the first storage knowledge network data.

[0120] In this embodiment, in the e-commerce warehousing scenario, the first warehousing knowledge network data contains a large number of warehousing storage locations. The server starts to poll these warehousing storage locations. Assuming that the scale of the warehousing storage locations in the first warehousing knowledge network data is 100, the server starts to poll from the first warehousing storage location (ie, b=1).

[0121] For storage location 1 (storage location b), the server determines its associated M storage locations. The determination of associated storage locations is based on multiple factors, such as the relationship between the flow of goods, the physical layout relationship of the locations, and the similarity of the types of goods.

[0122] From the perspective of the goods flow relationship, the server analyzes the historical goods in and out records. If it is found that goods often flow from storage location 3, storage location 5 and storage location 8 to storage location 1, and also often flow from storage location 1 to storage location 12, storage location 15 and storage location 18, then these locations may be preliminarily determined as locations associated with storage location 1.

[0123] From the perspective of the physical layout of storage locations, the storage locations surrounding storage location 1, such as storage location 2 and storage location 4 (if they are adjacent in the warehouse layout) will also be considered as associated locations.

[0124] Considering the similarity of the types of goods, if storage location 1 stores electronic products and storage location 7 also stores accessories related to electronic products, storage location 7 may also be included in the associated locations. By combining these factors, the server determines the association of storage location 1 (storage location b), which includes M storage locations (assuming M=9, determined according to the previously set relationship).

[0125] Then the server continues to poll the next storage location, such as storage location 2 (at this time b=2), and determines its associated M storage locations in the same way until all 100 storage locations in the first storage knowledge network data are polled.

[0126] Step B120, based on the knowledge vector data of each storage location in the association of the storage location b, perform M times of difference tradeoff calculation and analysis on the knowledge vector data of the storage location b to generate the allocation migration information of the storage location b. The allocation migration information of the storage location b includes M allocation migration parameters, and the cth allocation migration parameter among the M allocation migration parameters is generated by the cth difference tradeoff calculation and analysis, c∈[1,M].

[0127] After determining the M storage locations associated with storage location b (for example, b=1), the server obtains the knowledge vector data of these storage locations. Assume that the associated storage locations of storage location 1 are storage location 3, storage location 5, storage location 8, storage location 12, storage location 15, storage location 18, storage location 2, storage location 4, and storage location 7 (a total of M=9).

[0128] For storage location 3, its knowledge vector data may include that the storage volume of goods is 50 pieces, the frequency of goods entering and leaving the warehouse is 3 times a week, the average residence time of goods is 2 days, etc.; for storage location 5, its knowledge vector data may include that the storage volume of goods is 40 pieces, the frequency of goods entering and leaving the warehouse is 2 times a week, the average residence time of goods is 3 days, etc., and so on.

[0129] The server performs 9 (M) differential trade-off calculations and analyses on the knowledge vector data of storage location 1. For the first differential trade-off calculation and analysis (c=1, corresponding to storage location 3), the server compares the knowledge vector data of storage location 1 and storage location 3. For example, comparing the storage volume of goods, if the storage volume of goods in storage location 1 is 60 pieces, the server calculates the difference between the two storage volumes (60-50=10 pieces), and then converts this difference into a transfer migration parameter according to the pre-set trade-off rules. If the difference is positive and large, it may mean that a certain amount of goods needs to be transferred from storage location 1 to storage location 3, and this transfer volume (assuming 5 pieces) becomes the first transfer migration parameter.

[0130] For the second difference trade-off calculation analysis (c=2, corresponding to storage location 5), other knowledge vector data elements such as the frequency of goods entering and leaving the warehouse are compared. If the frequency of goods entering and leaving the warehouse at storage location 1 is 4 times per week, and that at storage location 5 is 2 times per week, according to the trade-off rule, a transfer migration parameter may be generated to indicate the direction and quantity of goods transfer between storage location 1 and storage location 5. For example, it may indicate that part of the goods are transferred from storage location 1 to storage location 5, and the transfer quantity is 10 pieces (derived according to the specific trade-off calculation rule), and this 10 pieces is the second transfer migration parameter.

[0131] According to this method, the server performs differential trade-off calculation and analysis on the 9 storage locations associated with storage location 1 in turn, and generates allocation and migration information of storage location 1 including 9 allocation and migration parameters.

[0132] The server then repeats this process for the next storage location (such as b=2) until the allocation and migration information is generated for each storage location in the first storage knowledge network data.

[0133] Step B130: after all storage locations in the first storage knowledge network data are polled, a basic allocation and migration information network is generated, wherein the basic allocation and migration information network includes allocation and migration information of each storage location in the first storage knowledge network data.

[0134] When the server completes polling of all 100 storage locations (assuming the scale is 100) in the first storage knowledge network data and generates allocation migration information for each storage location, it starts to generate a basic allocation migration information network.

[0135] This basic allocation migration information network can be viewed as a data structure, where each element corresponds to the allocation migration information of a storage location. For example, the first element is the allocation migration information of storage location 1 (containing 9 allocation migration parameters, calculated according to the previous calculation), the second element is the allocation migration information of storage location 2, and so on, until the 100th element is the allocation migration information of storage location 100.

[0136] This basic allocation and migration information network as a whole reflects the allocation and migration demand relationship between each storage location in the first warehouse knowledge network data based on the current knowledge vector data, and is the basis for subsequent warehouse optimization processing.

[0137] Step B140, performing storage optimization processing on the basic deployment migration information network to generate a target deployment migration information network, wherein the target deployment migration information network includes A deployment migration information.

[0138] The server performs warehouse optimization processing on the basic deployment migration information network. First, the server considers the overall warehouse layout and operation strategy. For example, if the warehouse has different areas dedicated to storing different types of goods (such as area A for electronic products and area B for daily necessities), this area division principle should be followed during the optimization process.

[0139] The server checks whether the allocation migration information in the basic allocation migration information network complies with the regional division rules. If it is found that the allocation migration information of storage location 1 (located in area A) has goods allocated to storage location 50 (located in area B, but the goods type does not match) and does not comply with the special cross-area allocation strategy, the server will adjust the allocation migration parameters.

[0140] At the same time, the server also considers the equipment resources of the warehouse. If a certain dispatch migration information indicates that a large amount of goods needs to be moved in a short period of time, but the transportation capacity of the handling equipment cannot meet the requirements, the server will readjust the dispatch migration parameters. For example, the dispatch quantity will be reduced or the dispatch time will be extended to adapt to the limitations of equipment resources.

[0141] In addition, the server will also consider factors such as the timeliness of the goods. If some goods have a high timeliness requirement, when optimizing the allocation and migration information, priority will be given to ensuring the rapid allocation of these goods. For example, for the time-sensitive goods stored in storage location 20, its allocation and migration information will be adjusted so that it can be allocated to the target storage location more quickly.

[0142] Through these warehouse optimization processes, the server generates a target allocation migration information network. This target allocation migration information network still contains A allocation migration information (assuming that A is the same as the storage location scale in the first warehouse knowledge network data, that is, 100), but these allocation migration information have been optimized to better meet the requirements of the overall layout of the warehouse, equipment resources, and cargo characteristics.

[0143] In a possible implementation, step S130 includes: Step S131, based on the knowledge vector data of the plurality of first associated storage locations, perform Y times of difference tradeoff calculation analysis on the first knowledge vector data of the storage location a to generate Y difference tradeoff calculation parameters. One round of difference tradeoff calculation analysis is used to obtain one difference tradeoff calculation parameter, and Y is an integer greater than 1.

[0144] Step S132, constructing a target difference tradeoff calculation analysis result based on the Y difference tradeoff calculation parameters.

[0145] In a possible implementation, the knowledge vector data of each first associated storage location and the first knowledge vector data of the storage location a both include vector values ​​under Z knowledge dimensions, where Z is a positive integer.

[0146] Step S131 includes: Step S1311: cluster the Z knowledge dimensions to generate Q knowledge dimension clusters, wherein each knowledge dimension cluster includes Z / Q knowledge dimensions, and Q is a positive integer.

[0147] Step S1312: Determine the target knowledge dimension cluster for performing the d-th difference tradeoff calculation analysis from the Q knowledge dimension clusters, where d∈[1,Y].

[0148] Step S1313: poll each knowledge dimension in the target knowledge dimension cluster, and use the currently polled knowledge dimension as the current knowledge dimension.

[0149] Step S1314, based on the vector value of each first associated storage location under the current knowledge dimension, perform difference weight calculation and analysis on the vector value of the storage location a under the current knowledge dimension to generate a difference weight vector value under the current knowledge dimension.

[0150] Step S1315, after all knowledge dimensions in the target knowledge dimension cluster are polled, the difference tradeoff vector values ​​under each knowledge dimension in the target knowledge dimension cluster are accumulated to generate the dth difference tradeoff calculation parameter.

[0151] In this embodiment, in the e-commerce warehousing scenario, it is assumed that storage location a is location 10, and multiple first associated storage locations have been determined, such as location 11, location 12, location 13, etc. The knowledge vector data of each first associated storage location and the first knowledge vector data of storage location a both include vector values ​​under Z knowledge dimensions (assuming Z=6), and these knowledge dimensions may include the storage volume of goods, the frequency of goods entering and leaving the warehouse, the average residence time of goods, the type of goods, the distance between the location and the entrance and exit, and the utilization rate of equipment around the location.

[0152] The server needs to perform Y times (assuming Y=3) difference trade-off calculation analysis.

[0153] First, cluster the Z=6 knowledge dimensions to generate Q knowledge dimension clusters (assuming Q=3). Each knowledge dimension cluster includes Z / Q=2 knowledge dimensions. The first knowledge dimension cluster includes the storage volume and type of goods, the second knowledge dimension cluster includes the frequency of goods entering and leaving the warehouse and the average residence time of goods, and the third knowledge dimension cluster includes the distance between the cargo location and the entrance and exit and the utilization rate of the equipment around the cargo location.

[0154] For the first difference trade-off calculation analysis (d=1), the server determines the first knowledge dimension cluster (including cargo storage volume and cargo type) from the three knowledge dimension clusters as the target knowledge dimension cluster.

[0155] For the second difference trade-off calculation analysis (d=2), the server determines the second knowledge dimension cluster (including the frequency of goods entering and leaving the warehouse and the average residence time of goods) as the target knowledge dimension cluster.

[0156] For the third difference trade-off calculation analysis (d=3), the server determines the third knowledge dimension cluster (including the distance from the cargo location to the entrance and exit and the utilization rate of the equipment around the cargo location) as the target knowledge dimension cluster.

[0157] Taking the first difference trade-off calculation analysis (d=1) as an example, the target knowledge dimension cluster includes cargo storage quantity and cargo type.

[0158] The server first polls the knowledge dimension of cargo storage capacity and uses it as the current knowledge dimension.

[0159] Then, for the second polling, the server polls the knowledge dimension of the cargo type, and the cargo type becomes the current knowledge dimension.

[0160] When the current knowledge dimension is the cargo storage capacity, assume that the cargo storage capacity of storage location a (location 10) is 50 pieces. The cargo storage capacity of location 11 is 40 pieces, the cargo storage capacity of location 12 is 60 pieces, the cargo storage capacity of location 13 is 30 pieces, and so on.

[0161] The server calculates the difference weights according to a preset difference weighting method, such as calculating the absolute value of the difference. For cargo location 11, the difference weighting vector value is |50-40|=10 pieces; for cargo location 12, the difference weighting vector value is |50-60|=10 pieces; for cargo location 13, the difference weighting vector value is |50-30|=20 pieces, and so on.

[0162] When the current knowledge dimension is the type of goods, assuming that the type of goods is coded (such as 1 for electronic products, 2 for daily necessities, etc.), the type of goods in shelf 10 is coded as 1, the type of goods in shelf 11 is coded as 1, the type of goods in shelf 12 is coded as 2, and the type of goods in shelf 13 is coded as 1. The server sets a difference value according to the difference in the type of goods (for example, 0 for the same type and 1 for different types). For shelf 11, the difference trade-off vector value is 0; for shelf 12, the difference trade-off vector value is 1; for shelf 13, the difference trade-off vector value is 0.

[0163] Taking the first difference trade-off calculation analysis (d=1) as an example, the target knowledge dimension cluster includes cargo storage quantity and cargo type.

[0164] After polling the two knowledge dimensions of cargo storage quantity and cargo type and obtaining the difference weight vector values ​​under each knowledge dimension, the server accumulates these difference weight vector values.

[0165] Assume that the difference weight vector values ​​of the cargo storage volume are 10 pieces, 10 pieces, and 20 pieces (corresponding to cargo location 11, cargo location 12, and cargo location 13), and the difference weight vector values ​​of the cargo type are 0, 1, and 0 (corresponding to cargo location 11, cargo location 12, and cargo location 13). If the difference weight vector values ​​of the cargo storage volume are directly added to 10+10+20=40 pieces, the difference weight vector values ​​of the cargo type are added to 0+1+0=1. According to the set accumulation rule (for example, multiply the difference value of the cargo type by a coefficient, assuming it is 10), convert the difference value of the cargo type to the same unit of measurement as the cargo storage volume, and get 1×10=10. Then add the two together, 40+10=50, and this 50 is the first difference weight calculation parameter.

[0166] In the same way, the server operates the second difference tradeoff calculation analysis (d=2) and the third difference tradeoff calculation analysis (d=3) to generate the second and third difference tradeoff calculation parameters respectively.

[0167] The server has generated Y (assuming Y=3) difference tradeoff calculation parameters, which are 50 (the first difference tradeoff calculation parameter), 30 (the second difference tradeoff calculation parameter, calculated by similar steps to the above), and 40 (the third difference tradeoff calculation parameter).

[0168] The server constructs the target difference trade-off calculation analysis result. This result can be an array or data structure containing the three difference trade-off calculation parameters. For example, the target difference trade-off calculation analysis result = [50, 30, 40], which reflects the comprehensive difference trade-off between storage location a and the first associated storage location in different knowledge dimension clusters.

[0169] In a possible implementation, the matrix of the storage optimization strategy is X×X, the multiple first associated storage locations are M first associated storage locations, and X is a positive integer.

[0170] Step S1314 includes: Step S1314-1, obtaining a difference trade-off component d for performing the d-th difference trade-off calculation and analysis, wherein the difference trade-off component d includes Z / Q difference trade-off dimensions, and each difference trade-off dimension includes M influencing factors.

[0171] Step S1314-2, respectively calculate the difference between the vector value of each first associated storage location under the current knowledge dimension and the vector value of the storage location a under the current knowledge dimension, and generate M vector difference values.

[0172] Step S1314-3, based on the M influencing factors in the e-th difference weight dimension included in the difference weight component d, the M vector difference values ​​are fused and calculated to generate the difference weight vector value under the current knowledge dimension, where e is equal to the order code of the current knowledge dimension in the target knowledge dimension cluster.

[0173] In this embodiment, in the e-commerce warehousing scenario, it is assumed that storage location a is location 10, and the matrix of the warehousing optimization strategy is X×X (assuming X=3, then M=X²=9, that is, there are 9 first-associated storage locations, such as location 11, location 12, ..., location 19), and the knowledge vector data of each first-associated storage location and storage location a contains Z knowledge dimensions (assuming Z=6), and the Z knowledge dimensions are clustered into Q knowledge dimension clusters (assuming Q=3), and each knowledge dimension cluster contains Z / Q=2 knowledge dimensions.

[0174] When performing the dth (assuming d=1) difference tradeoff calculation analysis, the server obtains difference tradeoff component 1 from the database module storing the difference tradeoff components. This difference tradeoff component 1 includes Z / Q=2 difference tradeoff dimensions.

[0175] For the first difference trade-off dimension, it contains M=9 influencing factors. These influencing factors are pre-set according to various characteristics and business rules of warehousing. For example, the influencing factors may be related to the importance of goods, the priority of cargo locations, the adaptability of handling equipment, etc. Assume that the first influencing factor is related to the importance of goods, and the higher the value, the more important the goods; the second influencing factor is related to the priority of cargo locations, and the higher the value, the higher the priority of cargo locations in the warehouse layout, etc.

[0176] Similarly, the second difference trade-off dimension also contains 9 influencing factors, which may be related to other warehousing-related factors, such as the timeliness of goods and the storage efficiency of cargo space.

[0177] Let's continue to assume that the current knowledge dimension is the cargo storage capacity (in a certain knowledge dimension cluster), and the cargo storage capacity of storage location a (location 10) is 50 pieces.

[0178] For the first associated storage location 11, the storage volume of goods is 40 pieces, so the vector difference value between the two is |50-40|=10 pieces.

[0179] For cargo location 12, the cargo storage capacity is 60 pieces, and the vector difference value is |50-60|=10 pieces.

[0180] Using the same method, the vector difference values ​​of cargo location 13, cargo location 14, ..., cargo location 19 and cargo location 10 under the knowledge dimension of cargo storage volume are calculated, generating a total of M=9 vector difference values, such as [10,10,20,5,15,0,3,7,12] (these values ​​are only examples).

[0181] We continue to assume that the order encoding of the current knowledge dimension in the target knowledge dimension cluster is e=1 (i.e., the first difference trade-off dimension in the corresponding difference trade-off component d).

[0182] The 9 influencing factors in the first difference trade-off dimension of the difference trade-off component d (related to the importance of goods, priority of cargo locations, etc.) are [0.1, 0.2, 0.3, 0.15, 0.05, 0.1, 0.05, 0.03, 0.02] (these values ​​are only examples and are set according to actual warehousing business rules).

[0183] The server integrates the previously generated M=9 vector difference values ​​[10,10,20,5,15,0,3,7,12] with the nine influencing factors. For example, using the weighted summation method, the difference weight vector value is calculated as follows: 10×0.1+10×0.2+20×0.3+5×0.15+15×0.05+0×0.1+3×0.05+7×0.03+12×0.02=1+2+6+0.75+0.75+0+0.15+0.21+0.24=11.05. This 11.05 is the difference weight vector value under the current knowledge dimension of cargo storage volume.

[0184] If the order code of the current knowledge dimension in the target knowledge dimension cluster is e=2 (corresponding to the second difference trade-off dimension in the difference trade-off component d), the 9 influencing factors in the second difference trade-off dimension are used to fuse the M vector difference values ​​according to the same weighted summation method to obtain another difference trade-off vector value, which reflects the comprehensive difference situation after considering multiple warehousing-related factors under the current knowledge dimension.

[0185] In a possible implementation, the matrix of the storage optimization strategy is X×X, the plurality of first associated storage locations are M first associated storage locations, and X is a positive integer. The target difference tradeoff calculation analysis result includes difference tradeoff calculation parameters of Y knowledge dimensions, where Y>M.

[0186] Step S140 includes: Step S141, calling the knowledge dimension optimizer to perform knowledge dimension optimization on the target difference trade-off calculation and analysis result to generate M-dimensional data.

[0187] Step S142: Use the i-th dimension data in the M-dimensional data as the first storage scheduling association degree between the storage location a and the i-th first associated storage location, where i∈[1,M].

[0188] In this embodiment, in the e-commerce warehousing scenario, it is assumed that storage location a is location 10, and the matrix of the warehousing optimization strategy is X×X (assuming X=4, then M=X²=16, that is, there are 16 first-associated storage locations), and the target difference trade-off calculation analysis results include the difference trade-off calculation parameters of Y knowledge dimensions (assuming Y=20).

[0189] The server calls the knowledge dimension optimizer, which is a specially designed algorithm module that can integrate and optimize multiple knowledge dimensions in the target difference trade-off calculation analysis results.

[0190] The knowledge dimension optimizer first analyzes the 20 difference tradeoff calculation parameters in the target difference tradeoff calculation analysis results. These difference tradeoff calculation parameters reflect the comprehensive difference tradeoffs between storage location a and the first associated storage location in different knowledge dimension clusters, such as the comprehensive differences in multiple knowledge dimensions such as the storage volume of goods, the frequency of goods entering and leaving the warehouse, and the type of goods.

[0191] The optimizer uses the principal component analysis (PCA) algorithm to optimize the knowledge dimension. It calculates the covariance matrix between the parameters of the difference trade-off calculation, and then finds the eigenvalues ​​and eigenvectors of the covariance matrix. According to the size of the eigenvalue, the most important M=16 eigenvectors are selected (because there are 16 first-association storage locations), and the target difference trade-off calculation analysis results are projected onto the subspace formed by these 16 eigenvectors.

[0192] After this projection operation, M-dimensional data is generated. This M-dimensional data is a data representation in a new and optimized knowledge dimension space, which can more effectively reflect the relationship between storage location a and the first associated storage location, remove some redundant information and highlight the key dimensional relationship.

[0193] Next, the server obtains the M-dimensional data generated by the knowledge dimension optimizer. Assume that the M-dimensional data is an array containing 16 elements, such as [0.1, 0.2, 0.3, 0.15, 0.05, 0.1, 0.05, 0.03, 0.02, 0.12, 0.18, 0.22, 0.08, 0.09, 0.11, 0.07].

[0194] For i=1, the server uses the first dimension data 0.1 as the first storage scheduling correlation between storage location a (location 10) and the first first associated storage location (assuming it is location 11). This correlation indicates the closeness of storage locations 10 and 11 in storage scheduling. The higher the value, the stronger the correlation between the two. For example, in storage scheduling operations such as cargo allocation and handling sequence, a high correlation means that they are more likely to affect each other or have a closer operational connection.

[0195] For i=2, 0.2 is used as the first storage scheduling association between storage location 10 and the second first associated storage location (assuming it is storage location 12). In the same way, the server maps each dimension data in the M-dimensional data to the first storage scheduling association between storage location a and the corresponding first associated storage location, until i=16, completing the association setting for all 16 first associated storage locations. These associations will provide an important basis for the subsequent generation of storage optimization strategies corresponding to storage location a.

[0196] In a possible implementation, the matrix of the storage optimization strategy is X×X, the multiple first associated storage locations are M first associated storage locations, and X is a positive integer.

[0197] Step S140 further includes: Step S143, performing smoothing filtering processing on the second warehousing knowledge network data to generate candidate warehousing knowledge network data, and determining M second associated storage locations of the storage location a in the candidate warehousing knowledge network data.

[0198] Step S144, obtaining the second storage scheduling association degree between the storage location a and each second associated storage location.

[0199] Step S145, integrating the i-th second storage scheduling correlation degree with the estimated i-th first storage scheduling correlation degree to generate the i-th global storage scheduling correlation degree corresponding to the storage location a, where i∈[1,M].

[0200] Step S146: generating a storage optimization strategy corresponding to the storage location a according to the M global storage scheduling associations corresponding to the storage location a.

[0201] In this embodiment, in the e-commerce warehousing scenario, the second warehousing knowledge network data contains rich warehousing scheduling behavior data and other information. The server performs smoothing and filtering on the second warehousing knowledge network data. For example, for the goods in and out time data, if the original data has a lot of fluctuations (possibly due to data collection errors or temporary operation fluctuations), the server uses the moving average method for smoothing and filtering.

[0202] Assume that the time of goods entering and leaving the warehouse is recorded hourly in the original data, and fluctuates greatly within a certain day. The server calculates the average time of goods entering and leaving the warehouse in the past 3 hours to replace the original data of each hour, so as to obtain smoother goods entering and leaving the warehouse time data. For other data, such as the operating speed of handling equipment, similar methods are also used for smoothing and filtering to generate candidate warehouse knowledge network data.

[0203] Then, the server determines M second associated storage locations of storage location a (eg, storage location 10) in the candidate storage knowledge network data. The basis for determining the associated storage locations includes the goods flow relationship, the physical location relationship of the storage locations, and the similarity of the goods types.

[0204] From the perspective of cargo flow relationships, the server analyzes cargo handling records in the candidate warehouse knowledge network data. If it is found that cargo often flows from cargo location 13, cargo location 15, and cargo location 18 to cargo location 10, and also often flows from cargo location 10 to cargo location 20, cargo location 22, and cargo location 25, etc., these cargo locations may be preliminarily determined as cargo locations associated with cargo location 10.

[0205] From the perspective of the physical location of the storage locations, the storage locations surrounding storage location 10, such as storage location 9 and storage location 11 (if they are adjacent in the warehouse layout) will also be considered as associated locations.

[0206] Considering the similarity of the types of goods, if the shelf 10 stores electronic products and the shelf 17 also stores accessories related to electronic products, the shelf 17 may also be included in the associated shelf. By combining these factors, the server determines M (assuming M=9) second associated storage locations of the shelf 10 (storage location a) in the candidate storage knowledge network data, such as shelf 9, shelf 11, shelf 13, shelf 15, shelf 17, shelf 18, shelf 20, shelf 22 and shelf 25.

[0207] Next, for storage location a (location 10) and the first second associated storage location (location 9), the server determines the second knowledge vector data of location 10 (for example, including information such as cargo storage volume, cargo in / out frequency, and cargo average residence time) based on the candidate storage knowledge network data. Assume that the cargo storage volume of location 10 is 50 pieces, the cargo in / out frequency is 3 times per week, and the cargo average residence time is 2 days. At the same time, the knowledge vector data of location 9 is determined. Assume that the cargo storage volume of location 9 is 40 pieces, the cargo in / out frequency is 2 times per week, and the cargo average residence time is 3 days.

[0208] Then, based on the knowledge vector data of cargo location 9, the server performs a difference weighting calculation and analysis on the second knowledge vector data of cargo location 10. For example, the difference in cargo storage volume is calculated to be |50-40|=10 pieces, the difference in cargo in and out frequency is |3-2|=1 time / week, and the difference in average cargo residence time is |2-3|=1 day. According to the pre-set weighing rules (for example, these differences are weighted and summed, and the weights are set according to the importance of the business, assuming that the weight of cargo storage volume is 0.5, the weight of cargo in and out frequency is 0.3, and the weight of average cargo residence time is 0.2), a value is calculated. After this value is converted (for example, the calculation result is mapped to between 0 and 1 through a mapping function), the second storage scheduling correlation between cargo location 10 and cargo location 9 is obtained, which is assumed to be 0.6.

[0209] In the same way, the server calculates the second storage scheduling correlation between cargo location 10 and other second associated storage cargo locations (cargo location 11, cargo location 13, etc.). For example, the second storage scheduling correlation between cargo location 10 and cargo location 11 is calculated to be 0.4, and the second storage scheduling correlation between cargo location 10 and cargo location 13 is calculated to be 0.7, and so on.

[0210] Then, assume that for i=1, the first second storage scheduling correlation is the correlation between cargo location 10 and cargo location 9, which is 0.6, and the first first storage scheduling correlation (previously estimated) is 0.4. The server uses the weighted average method for integration (the weight can be set according to the actual situation, assuming that both weights are 0.5). The first global storage scheduling correlation is calculated to be (0.6×0.5+0.4×0.5)=0.5.

[0211] For i=2, assume that the second storage scheduling correlation between cargo location 10 and cargo location 11 is 0.4, and the second storage scheduling correlation is 0.3. Similarly, weighted average (weights are all 0.5) is used to obtain the second global storage scheduling correlation of (0.4×0.5+0.3×0.5)=0.35. According to this method, the server integrates each correlation of i from 1 to M (assuming M=9) to generate 9 global storage scheduling correlations, such as [0.5, 0.35, 0.65, 0.55, 0.45, 0.6, 0.4, 0.5, 0.48].

[0212] Next, the server generates a storage optimization strategy for storage location a (location 10) based on the M (9) global storage scheduling correlations. If the global storage scheduling correlation is high, for example, the global storage scheduling correlation between location 10 and location 13 is 0.65, this means that in the storage optimization strategy, operations such as cargo allocation and equipment scheduling between location 10 and location 13 should be closer.

[0213] The server can construct an X×X (assuming X=3, M=X²=9) matrix to represent the storage optimization strategy. The element (i,j) in the matrix represents the optimization relationship between storage location a (location 10) and the jth associated storage location (the order in the M associated storage locations). For example, for location 13 (assuming it is the third associated storage location) with a high degree of relevance to global storage scheduling, the element (1,3) in the matrix may represent a higher priority value (such as 3, indicating a higher priority in operations such as goods allocation).

[0214] For a storage location with a low correlation with global storage scheduling, such as storage location 11 (assuming the correlation is 0.35), the element (1,2) in the matrix may represent a lower priority value (such as 1, indicating a lower priority in operations such as cargo allocation). By constructing the matrix in this way, the server generates a storage optimization strategy corresponding to storage location a (storage location 10). This strategy will guide operations such as cargo allocation and handling equipment scheduling in the warehouse to improve the efficiency of storage operations.

[0215] In a possible implementation, step S144 includes: Step S1441, determining the second knowledge vector data of storage location a according to the second storage knowledge network data, and determining the knowledge vector data of each second associated storage location.

[0216] In this embodiment, in the e-commerce warehousing scenario, the second warehousing knowledge network data contains rich warehousing information. The server determines the second knowledge vector data of the storage location a (eg, location 10) from the second warehousing knowledge network data.

[0217] This second knowledge vector data contains information from multiple dimensions. For example, in terms of cargo storage capacity, the server finds that the current storage capacity of cargo location 10 is 50 pieces; in terms of cargo in and out frequency, by analyzing historical data, it is found that the cargo in and out frequency of cargo location 10 is 3 times a week; in terms of average cargo residence time, it is calculated that the average residence time of cargo in cargo location 10 is 2 days; in terms of cargo type, it is determined that electronic products are stored in cargo location 10; in terms of the distance from the cargo location to the entrance and exit, it is measured or calculated that the distance from the cargo location 10 to the entrance and exit is 20 meters; in terms of the utilization rate of equipment around the cargo location, according to the equipment usage records, it is known that the utilization rate of the handling equipment (such as forklifts) around cargo location 10 is 0.6 (indicating the proportion of the busyness of the equipment).

[0218] For each second-associated storage location, the server also determines its knowledge vector data in the same way. Taking location 9 (a second-associated storage location) as an example, the query shows that its cargo storage volume is 40 pieces, the frequency of cargo in and out of the warehouse is 2 times a week, the average cargo residence time is 3 days, the cargo type is electronic product accessories, the distance from the entrance and exit is 18 meters, and the peripheral equipment utilization rate is 0.5. Taking location 11 as an example, its cargo storage volume is 60 pieces, the frequency of cargo in and out of the warehouse is 4 times a week, the average cargo residence time is 1.5 days, the cargo type is daily necessities, the distance from the entrance and exit is 22 meters, and the peripheral equipment utilization rate is 0.7. The server performs such data query and knowledge vector data determination operations for all M second-associated storage locations (assuming M=9, including location 9, location 11, etc.).

[0219] Step S1442, performing a difference weight calculation analysis on the second knowledge vector data of the storage location a based on the knowledge vector data of the M second associated storage locations, and generating a candidate difference weight calculation analysis result.

[0220] The server begins to perform a difference trade-off calculation analysis on storage location a (location 10) and M (9) second associated storage locations.

[0221] First, compare the cargo storage volume dimension. For cargo location 9, its cargo storage volume is 40 pieces, and the cargo storage volume of cargo location 10 is 50 pieces. The difference between the two is calculated as |50-40|=10 pieces. For cargo location 11, its cargo storage volume is 60 pieces, and the difference with cargo location 10 is |50-60|=10 pieces. In this way, the difference values ​​of cargo storage volume between cargo location 10 and all 9 second-associated storage locations are calculated, such as [10,10,5,15,0,3,7,12,8] (these values ​​correspond to 9 second-associated storage locations).

[0222] In the dimension of goods in and out frequency, the goods in and out frequency of cargo location 9 is 2 times per week, and that of cargo location 10 is 3 times per week, with a difference of |3-2|=1 time / week. For cargo location 11 (4 times per week), the difference is |3-4|=1 time / week. Similarly, the difference values ​​of cargo location 10 and all second-associated storage locations in this dimension are calculated, such as [1,1,2,0,1,0.5,1.5,0.5,2].

[0223] For the dimension of average cargo residence time, the average cargo residence time of cargo in storage location 9 is 3 days, and that of cargo location 10 is 2 days, with a difference of |2-3|=1 day. The difference values ​​of cargo location 10 and all second-associated storage locations in this dimension are calculated, such as [1, 0.5, 0.5, 1.5, 2, 0, 1, 0, 1].

[0224] In the goods type dimension, if codes are used to represent the goods type (such as 1 for electronic products and 2 for daily necessities), location 10 is 1, location 9 is 1, and the difference is 0; location 11 is 2, and the difference is 1. Calculate the difference between location 10 and all second-associated storage locations in this dimension, such as [0,1,0,0,1,0,1,0,0].

[0225] For the distance dimension between the storage location and the entrance and exit, the distance between storage location 9 and the entrance and exit is 18 meters, and the distance between storage location 10 and the entrance and exit is 20 meters. The difference is |20-18|=2 meters. Calculate the difference between storage location 10 and all the second associated storage locations in this dimension, such as [2,2,5,3,1,4,0,1,3].

[0226] In the dimension of equipment utilization around the cargo location, the equipment utilization of cargo location 9 is 0.5, and that of cargo location 10 is 0.6, with a difference of |0.6-0.5|=0.1. Calculate the difference between cargo location 10 and all second-associated storage cargo locations in this dimension, such as [0.1, 0.1, 0.2, 0.05, 0.15, 0.05, 0.05, 0.1, 0.08].

[0227] The server calculates the difference values ​​of these dimensions according to the preset weighing rules. For example, the weighted summation method is adopted, assuming that the weight of cargo storage volume is 0.2, the weight of cargo in and out frequency is 0.2, the weight of cargo average residence time is 0.1, the weight of cargo type is 0.1, the weight of cargo location distance to the entrance and exit is 0.2, and the weight of the utilization rate of equipment around the cargo location is 0.2.

[0228] Taking cargo location 9 as an example, a comprehensive value is calculated: (10×0.2+1×0.2+1×0.1+0×0.1+2×0.2+0.1×0.2)=2+0.2+0.1+0+0.4+0.02=2.72. In the same way, the comprehensive values ​​of cargo location 10 and all second-associated storage locations are calculated. These comprehensive values ​​constitute the candidate difference trade-off calculation analysis results, such as [2.72, 2.85, 3.1, 2.45, 2.3, 1.85, 2.2, 2.1, 2.5].

[0229] Step S1443, based on the candidate difference trade-off calculation analysis result, estimate the second storage scheduling association degree between the storage location a and each second associated storage location.

[0230] In this embodiment, the server estimates the second storage scheduling association degree between storage location a (location 10) and each second associated storage location based on the candidate difference trade-off calculation analysis result generated previously.

[0231] The server uses a mapping function to convert the numerical value in the candidate difference trade-off calculation analysis result into a correlation value between 0 and 1. For example, for the candidate difference trade-off calculation analysis result value 2.72 corresponding to cargo location 9, the mapping function y=1 / (1+exp(-x)) (where x=2.72) is used to calculate y≈0.93. This 0.93 is the second storage scheduling correlation between cargo location 10 and cargo location 9, indicating the closeness between the two in terms of storage scheduling.

[0232] For the candidate difference trade-off calculation analysis result value 2.85 corresponding to the cargo location 11, the above mapping function is also used to calculate y≈0.94. According to this method, the server calculates the second storage scheduling correlation between the cargo location 10 and all 9 second associated storage cargo locations, such as [0.93, 0.94, 0.95, 0.91, 0.90, 0.88, 0.92, 0.91, 0.93]. These correlations will play an important role in the subsequent storage optimization strategy generation and other operations.

[0233] In a possible implementation, step S150 includes: Step S151, obtaining the allocation migration information a corresponding to the storage location a, wherein the allocation migration information a includes M allocation migration parameters.

[0234] In this embodiment, in the e-commerce warehousing scenario, it is assumed that the storage location a is location 10. The server obtains the allocation migration information a corresponding to location 10 from the previously generated and stored allocation migration information library. This allocation migration information library is constructed in the previous warehousing optimization process, which contains the allocation migration information of each storage location.

[0235] The deployment migration information a contains M deployment migration parameters. For example, M=9 (assuming that the storage optimization strategy matrix set previously is X×X, X=3, so M=X²=9). These 9 deployment migration parameters each represent different deployment information.

[0236] The first allocation migration parameter may indicate the amount of goods to be allocated between location 10 and the first associated storage location. Assuming this value is 50 pieces, it means that according to the warehouse optimization strategy, 50 pieces of goods need to be allocated from location 10 to the first associated storage location or in the opposite direction (the allocation direction is determined by the positive or negative value of the parameter or other indicators).

[0237] The second deployment migration parameter may be related to the deployment priority. For example, a value of 3 (assuming the priority range is 1-5) means that this deployment operation has a medium to high priority in the overall deployment plan.

[0238] The third allocation migration parameter may involve the time schedule of the allocation. For example, it may indicate that the allocation operation needs to be completed within the next 24 hours, which is determined based on factors such as the timeliness of the goods and the overall operation plan of the warehouse.

[0239] Other allocation migration parameters may also include the adaptability of the cargo type (for example, ensuring that the type of cargo to be allocated matches the storage requirements of the target storage location), the selection of handling equipment (such as specifying whether to use a forklift or a conveyor belt for cargo handling), etc. The server fully obtains these 9 allocation migration parameters to provide a basis for subsequent warehouse optimization operations.

[0240] Step S152, mapping the storage location a to the first warehousing knowledge network data, and generating a corresponding position corresponding to the storage location a.

[0241] In this embodiment, the first warehousing knowledge network data includes basic information of each storage location in the warehouse, such as the location number, location coordinates, location size, etc. The server maps storage location a (location 10) to the first warehousing knowledge network data.

[0242] In the first warehouse knowledge network data, each storage location has a unique identifier and corresponding location information. The server searches for the corresponding record in the data structure by the location number (location 10). Assume that the first warehouse knowledge network data is stored in a table format, with one column being the location number and the other columns being the related location information. The server determines the corresponding position of location 10 in the first warehouse knowledge network data by searching for the row with location number 10.

[0243] This corresponding position not only contains the basic information of the cargo location 10 itself, but also is related to the relationship between other storage cargo locations in the data structure. For example, its relative position relationship with surrounding cargo locations (adjacent relationship in warehouse layout, etc.) and its logical position relationship in the entire storage knowledge network (such as upstream and downstream relationship in the goods flow path, etc.). By determining this corresponding position, the server can accurately analyze and optimize the cargo location 10 and its related storage operations within the framework of the first storage knowledge network data.

[0244] Step S153, determining the M third associated storage locations of the storage location a in the first storage knowledge network data and the knowledge vector data of each third associated storage location based on the corresponding position and the M allocation migration parameters in the allocation migration information a.

[0245] In this embodiment, the server determines M third associated storage locations according to the corresponding position of the previously determined storage location a (location 10) in the first storage knowledge network data and the M (9) allocation and migration parameters in the allocation and migration information a.

[0246] Assume that the first allocation migration parameter indicates that 50 pieces of goods are allocated from storage location 10 to another storage location, and according to the corresponding location information, among the storage locations around storage location 10, storage location 11 is directly associated with storage location 10 in the goods flow path and complies with the allocation direction (assuming that the allocation direction is from storage location 10 to storage location 11), then storage location 11 may be determined as the first third associated storage location.

[0247] For cargo location 11 (the first third associated storage cargo location), the server obtains its knowledge vector data. This knowledge vector data contains various information about cargo location 11 in the first storage knowledge network data. For example, in terms of cargo storage capacity, the query shows that the current storage capacity is 40 pieces; the frequency of cargo in and out of the warehouse is 2 times a week; the average residence time of cargo is 3 days; the cargo type is electronic product accessories; the distance between the cargo location and the entrance and exit is 18 meters; the utilization rate of the equipment surrounding the cargo location is 0.5, etc.

[0248] According to other allocation migration parameters in the allocation migration information a, other third-associated storage locations are further determined. For example, the second allocation migration parameter (related to the allocation priority) and the third allocation migration parameter (related to the allocation schedule) may further screen and determine other qualified locations. Assume that according to these parameters, location 13 also meets the requirements and is determined as the second third-associated storage location. The server also obtains the knowledge vector data of location 13, such as the storage volume of goods is 60 pieces, the frequency of goods in and out of the warehouse is 4 times a week, the average residence time of goods is 1.5 days, the type of goods is daily necessities, the distance from the entrance and exit is 22 meters, the utilization rate of peripheral equipment is 0.7, etc.

[0249] According to this method, the server determines 9 third associated storage locations in the first storage knowledge network data based on the 9 allocation migration parameters in the allocation migration information a, and obtains the knowledge vector data of each third associated storage location.

[0250] Step S154: performing storage optimization on the knowledge vector data of the M third associated storage locations according to the storage optimization strategy corresponding to the storage location a.

[0251] In this embodiment, it is assumed that the storage optimization strategy corresponding to storage location a (location 10) is represented in the form of a 3×3 (X=3) matrix, in which the elements reflect the optimization relationship between location 10 and the associated storage locations.

[0252] For the first third associated storage location (storage location 11), according to the corresponding element in the storage optimization strategy matrix, it is assumed that the element indicates that the priority of the goods allocation is high (for example, the value in the matrix is ​​4, indicating a high priority, and the value range is 1-5). If the first allocation migration parameter in the allocation migration information a indicates that 50 pieces of goods are allocated from storage location 10 to storage location 11, the server adjusts the goods storage capacity information in the knowledge vector data of storage location 11 according to this priority. The original storage capacity of goods in storage location 11 is 40 pieces. After adding the 50 pieces of goods allocated, the updated storage capacity of goods is 90 pieces.

[0253] At the same time, due to the increase in the amount of goods allocated, the frequency of goods entering and leaving the warehouse at the cargo location 11 may be affected. The server recalculates the frequency of goods entering and leaving the warehouse at the cargo location 11 based on historical data and preset rules. Assuming that the frequency of goods entering and leaving the warehouse is adjusted from 2 times a week to 3 times a week based on factors such as the type of goods and the operation mode of the warehouse, this dimension in the knowledge vector data of the cargo location 11 is updated.

[0254] For the dimension of the distance between the cargo location and the entrance and exit, if a large amount of goods is allocated to cargo location 11, it may affect the handling efficiency. The server will evaluate whether it is necessary to adjust the distance relationship between cargo location 11 and the entrance and exit (represented in the knowledge vector data) based on the warehouse layout and equipment operation. For example, if it is found that the channel around cargo location 11 will be congested after the goods are allocated, the equivalent distance between cargo location 11 and the entrance and exit may be marked in the knowledge vector data as increased (for example, from 18 meters to 20 meters, where the equivalent distance is a comprehensive evaluation after considering factors such as congestion).

[0255] In the same way, the server optimizes the knowledge vector data of the other eight third-related storage locations based on the storage optimization strategy corresponding to storage location a (storage location 10). For example, for storage location 13 (the second third-related storage location), according to the priority, allocation quantity and other information in the storage optimization strategy, adjust the relevant dimensions in the knowledge vector data such as the storage volume of goods, the frequency of goods entering and leaving the warehouse, the average residence time of goods, the distance between the storage location and the entrance and exit, and the utilization rate of the equipment around the storage location, so as to achieve the optimization of the entire storage system under the first storage knowledge network data framework.

[0256] Figure 2 An embodiment of the present application provides an intelligent warehousing optimization scheduling system 100 based on the Internet of Things and AI, including a processor 1001 and a memory 1003 and an application code stored in the memory 1003. The processor 1001 executes the above application code to implement the steps of the sensor data processing method applied to smart photovoltaics.

[0257] Figure 2 The shown intelligent warehousing optimization and scheduling system 100 based on the Internet of Things and AI includes: a processor 1001 and a memory 1003. Among them, the processor 1001 and the memory 1003 are connected, such as through a bus 1002. Optionally, the intelligent warehousing optimization and scheduling system 100 based on the Internet of Things and AI may also include a transceiver 1004, which can be used for data interaction between the server and other servers, such as data transmission and / or data reception. It should be noted that the transceiver 1004 is not limited to one in actual scheduling, and the structure of the intelligent warehousing optimization and scheduling system 100 based on the Internet of Things and AI does not constitute a limitation on the embodiments of the present application.

[0258] Processor 1001 may be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array) or other programmable logic devices, transistor logic devices, hardware components or any combination thereof. It may implement or execute various exemplary logic blocks, modules and circuits described in conjunction with the disclosure of this application. Processor 1001 may also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, etc.

[0259] The bus 1002 may include a path to transmit information between the above components. The bus 1002 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus. The bus 1002 may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 2 Only one thick line is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0260] The memory 1003 may be a ROM (Read Only Memory) or other types of static storage devices that can store static information and instructions, a RAM (Random Access Memory) or other types of dynamic storage devices that can store information and instructions, or an EEPROM (Electrically Erasable Programmable Read Only Memory), a CD-ROM (Compact Disc Read Only Memory) or other optical disk storage, optical disk storage (including compressed optical disk, laser disk, optical disk, digital versatile disk, Blu-ray disk, etc.), magnetic disk storage medium, other magnetic storage devices, or any other medium that can be used to carry or store application code and can be read by a computer, without limitation herein.

[0261] The memory 1003 is used to store application code for executing the embodiment of the present application, and the execution is controlled by the processor 1001. The processor 1001 is used to execute the application code stored in the memory 1003 to implement the steps shown in the above method embodiment.

[0262] The server includes but is not limited to: mobile terminals such as mobile phones, notebook computers, PADs, etc. and fixed terminals such as digital TVs, desktop computers, etc.

[0263] An embodiment of the present application provides a computer-readable storage medium on which application code is stored. When the application code is executed by a processor, the steps and corresponding contents of the aforementioned method embodiment can be implemented.

[0264] It should be understood that, although each operation step is indicated by arrows in the flowchart of the embodiment of the present application, the implementation order of these steps is not limited to the order indicated by the arrows. Unless there is clear explanation in this article, in some implementation scenarios of the embodiment of the present application, the implementation steps in each flowchart can be performed in other orders according to demand. In addition, some or all of the steps in each flowchart may include multiple sub-steps or multiple stages according to actual implementation scenarios, and some or all of these sub-steps or stages may be executed at the same time, and each sub-step or stage in these sub-steps or stages may also be executed at different times respectively. In different scenarios at the execution time, the execution order of these sub-steps or stages can be flexibly configured based on demand, and the embodiment of the present application does not limit this.

[0265] The above is only an optional implementation method for some implementation scenarios of the present application. It should be pointed out that for ordinary technicians in this technical field, without departing from the technical concept of the scheme of the present application, other similar implementation methods based on the technical ideas of the present application are also within the protection scope of the embodiments of the present application.

Claims

1. An intelligent warehousing optimization scheduling method based on the Internet of Things and AI, characterized in that: The method includes: Acquire first warehouse knowledge network data to be optimized, and acquire second warehouse knowledge network data; the first warehouse knowledge network data and the second warehouse knowledge network data are generated by constructing knowledge networks of different IoT warehouse management depths for the same warehouse area, and the knowledge vector data in the second warehouse knowledge network data includes warehouse scheduling behavior data; Determine multiple first associated storage locations of storage location a and knowledge vector data of each first associated storage location based on the first storage knowledge network data, and determine the first knowledge vector data of the storage location a based on the second storage knowledge network data; a∈[1, A], A is the scale of the storage location in the second storage knowledge network data; Based on the knowledge vector data of the plurality of first associated storage locations, performing a difference tradeoff calculation and analysis on the first knowledge vector data of the storage location a to generate a target difference tradeoff calculation and analysis result; Based on the target difference trade-off calculation and analysis results, the first storage scheduling correlation between the storage location a and each first associated storage location is estimated, and the storage optimization strategy corresponding to the storage location a is generated by using the estimated first storage scheduling correlations; According to the storage optimization strategy corresponding to the storage location a, the first storage knowledge network data is optimized.

2. The intelligent warehousing optimization scheduling method based on the Internet of Things and AI according to claim 1 is characterized in that: The feature detail granularity of the first warehouse knowledge network data is smaller than the feature detail granularity of the second warehouse knowledge network data, and the first warehouse knowledge network data is in the knowledge representation domain; The step of determining a plurality of first associated storage locations of storage location a and knowledge vector data of each first associated storage location based on the first storage knowledge network data, and determining the first knowledge vector data of storage location a based on the second storage knowledge network data, includes: Performing feature enhancement prediction processing on the first warehouse knowledge network data to generate third warehouse knowledge network data, wherein the feature detail granularity of the third warehouse knowledge network data is equal to the feature detail granularity of the second warehouse knowledge network data; Acquire a knowledge mapping toolset, and acquire third warehouse knowledge network data generated by performing feature enhancement prediction processing on the first warehouse knowledge network data; Generate a source knowledge graph in the knowledge mapping tool set based on the second warehouse knowledge network data, and generate a target knowledge framework in the knowledge mapping tool set based on the third warehouse knowledge network data; Using the knowledge mapping toolset, the source knowledge graph is mapped to the knowledge representation domain where the target knowledge framework is located, and fourth warehouse knowledge network data is generated; In the knowledge representation domain, knowledge vector adaptation is performed on the fourth warehouse knowledge network data and the third warehouse knowledge network data, so that two warehouse cargo locations in the same warehouse node in the fourth warehouse knowledge network data and the third warehouse knowledge network data are adapted to each other; After knowledge vector adaptation, the knowledge vector data of multiple first associated storage locations and each first associated storage location of storage location a in the third storage knowledge network data are determined, and the knowledge vector data of storage location a is obtained from the fourth storage knowledge network data as the first knowledge vector data of storage location a.

3. The intelligent warehousing optimization scheduling method based on the Internet of Things and AI according to claim 2 is characterized in that: After performing knowledge vector adaptation, the method further includes: Performing standardized conversion on the third warehouse knowledge network data and the fourth warehouse knowledge network data respectively to generate the standardized converted third warehouse knowledge network data and the standardized converted fourth warehouse knowledge network data; Acquire a conversion coefficient shared between the third warehouse knowledge network data and the fourth warehouse knowledge network data; According to the conversion coefficient, linear conversion is performed on the third warehouse knowledge network data after the standardization conversion and the fourth warehouse knowledge network data after the standardization conversion respectively; After performing the linear conversion, jump to the step of determining multiple first associated storage locations of storage location a in the third storage knowledge network data and the knowledge vector data of each first associated storage location, and obtaining the knowledge vector data of the storage location a from the fourth storage knowledge network data as the first knowledge vector data of the storage location a.

4. The intelligent warehousing optimization scheduling method based on the Internet of Things and AI according to claim 2 or 3 is characterized in that: The matrix of the storage optimization strategy is X×X, the plurality of first associated storage locations are M first associated storage locations, X is a positive integer, M is equal to the square of X, and the knowledge vector data of the plurality of first associated storage locations and each first associated storage location in the third storage knowledge network data for determining the storage location a includes: Acquire the allocation migration information a corresponding to the storage location a, wherein the allocation migration information a includes M allocation migration parameters; In the third storage knowledge network data, a target knowledge sub-network is determined with the storage location that is compatible with the storage location a as the center, and the target knowledge sub-network includes M storage locations; Based on the i-th allocation migration parameter of the allocation migration information a, the layout relationship of the storage node where the i-th storage location in the target knowledge subnetwork is located is allocated and migrated to generate a target storage node, i∈[1,M]; The storage location located at the target storage node in the third storage knowledge network data is used as the i-th first associated storage location of storage location a in the third storage knowledge network data, and the knowledge vector data of the i-th first associated storage location is obtained from the third storage knowledge network data; The allocation migration information a corresponding to the storage location a is the ath allocation migration information in the target allocation migration information network. Before acquiring the allocation migration information a corresponding to the storage location a, the method further includes: Polling each storage location in the first storage knowledge network data, and determining the association of the currently polled storage location b in the first storage knowledge network data; wherein the association of the storage location b includes M storage locations, and b is a positive integer and is not greater than the size of the storage locations in the first storage knowledge network data; Based on the knowledge vector data of each storage location in the association of the storage location b, M times of difference tradeoff calculation and analysis are performed on the knowledge vector data of the storage location b to generate the allocation and migration information of the storage location b; wherein the allocation and migration information of the storage location b includes M allocation and migration parameters, and the cth allocation and migration parameter among the M allocation and migration parameters is generated by the cth difference tradeoff calculation and analysis, c∈[1,M]; After all storage locations in the first storage knowledge network data are polled, a basic allocation migration information network is generated; wherein the basic allocation migration information network includes allocation migration information of each storage location in the first storage knowledge network data; The basic deployment migration information network is subjected to storage optimization processing to generate a target deployment migration information network, wherein the target deployment migration information network includes A deployment migration information.

5. The intelligent warehousing optimization scheduling method based on the Internet of Things and AI according to claim 1 is characterized in that: The step of performing difference weighted calculation and analysis on the first knowledge vector data of the storage location a based on the knowledge vector data of the plurality of first associated storage locations to generate a target difference weighted calculation and analysis result includes: According to the knowledge vector data of the plurality of first associated storage locations, performing Y difference tradeoff calculation and analysis on the first knowledge vector data of the storage location a, and generating Y difference tradeoff calculation parameters; wherein one round of difference tradeoff calculation and analysis is used to obtain one difference tradeoff calculation parameter, and Y is an integer greater than 1; Based on the Y difference tradeoff calculation parameters, a target difference tradeoff calculation analysis result is constructed.

6. The intelligent warehousing optimization scheduling method based on the Internet of Things and AI according to claim 5 is characterized in that: The knowledge vector data of each first associated storage location and the first knowledge vector data of the storage location a both include vector values ​​under Z knowledge dimensions, where Z is a positive integer; The step of performing Y difference weight calculation and analysis on the first knowledge vector data of the storage location a based on the knowledge vector data of the plurality of first associated storage locations to generate Y difference weight calculation parameters includes: Clustering the Z knowledge dimensions to generate Q knowledge dimension clusters; wherein each knowledge dimension cluster includes Z / Q knowledge dimensions, and Q is a positive integer; Determine, from the Q knowledge dimension clusters, a target knowledge dimension cluster for performing the d-th difference tradeoff calculation analysis; wherein d∈[1,Y]; Polling each knowledge dimension in the target knowledge dimension cluster, and taking the currently polled knowledge dimension as the current knowledge dimension; According to the vector value of each first associated storage location under the current knowledge dimension, a difference weighted calculation and analysis is performed on the vector value of the storage location a under the current knowledge dimension to generate a difference weighted vector value under the current knowledge dimension; After all knowledge dimensions in the target knowledge dimension cluster are polled, the difference weight vector values ​​under each knowledge dimension in the target knowledge dimension cluster are accumulated to generate the dth difference weight calculation parameter.

7. The intelligent warehousing optimization scheduling method based on the Internet of Things and AI according to claim 6 is characterized in that: The matrix of the storage optimization strategy is X×X, the plurality of first associated storage locations are M first associated storage locations, and X is a positive integer; The step of performing difference weight calculation and analysis on the vector value of the storage location a under the current knowledge dimension according to the vector value of each first associated storage location under the current knowledge dimension to generate the difference weight vector value under the current knowledge dimension includes: Obtaining a difference tradeoff component d for performing the d-th difference tradeoff calculation and analysis, wherein the difference tradeoff component d includes Z / Q difference tradeoff dimensions, and each difference tradeoff dimension includes M influencing factors; Calculate the difference between the vector value of each first associated storage location under the current knowledge dimension and the vector value of the storage location a under the current knowledge dimension, and generate M vector difference values; According to the M influencing factors in the e-th difference trade-off dimension included in the difference trade-off component d, the M vector difference values ​​are fused and calculated to generate the difference trade-off vector value under the current knowledge dimension; wherein e is equal to the order code of the current knowledge dimension in the target knowledge dimension cluster.

8. The intelligent warehousing optimization scheduling method based on the Internet of Things and AI according to claim 1 is characterized in that: The matrix of the storage optimization strategy is X×X, the plurality of first associated storage locations are M first associated storage locations, and X is a positive integer; the target difference trade-off calculation analysis result includes difference trade-off calculation parameters of Y knowledge dimensions, and Y>M; The estimating the first storage scheduling association degree between the storage location a and each first associated storage location based on the target difference trade-off calculation analysis result includes: Calling a knowledge dimension optimizer to perform knowledge dimension optimization on the target difference trade-off calculation and analysis results to generate M-dimensional data; The i-th dimension data in the M-dimensional data is used as the first storage scheduling association degree between the storage location a and the i-th first associated storage location; wherein, i∈[1,M].

9. The intelligent warehousing optimization scheduling method based on the Internet of Things and AI according to claim 1 is characterized in that: The matrix of the storage optimization strategy is X×X, the plurality of first associated storage locations are M first associated storage locations, and X is a positive integer; The method of using the estimated first storage scheduling associations to generate a storage optimization strategy corresponding to the storage location a includes: Performing smoothing filtering on the second warehousing knowledge network data to generate candidate warehousing knowledge network data, and determining M second associated warehousing locations of the warehousing location a in the candidate warehousing knowledge network data; Obtaining a second storage scheduling correlation degree between the storage location a and each second associated storage location; Integrate the i-th second storage scheduling correlation with the estimated i-th first storage scheduling correlation to generate the i-th global storage scheduling correlation corresponding to the storage location a; where i∈[1,M]; Generate a storage optimization strategy corresponding to the storage location a according to the M global storage scheduling associations corresponding to the storage location a; Wherein, the obtaining of the second storage scheduling correlation between the storage location a and each second associated storage location includes: Determine the second knowledge vector data of storage location a according to the second storage knowledge network data, and determine the knowledge vector data of each second associated storage location; Based on the knowledge vector data of the M second associated storage locations, performing a difference tradeoff calculation analysis on the second knowledge vector data of the storage location a to generate a candidate difference tradeoff calculation analysis result; Based on the candidate difference trade-off calculation and analysis results, estimating the second storage scheduling association degree between the storage location a and each second associated storage location; The step of performing storage optimization on the first storage knowledge network data according to the storage optimization strategy corresponding to the storage location a includes: Acquire the allocation migration information a corresponding to the storage location a, wherein the allocation migration information a includes M allocation migration parameters; Corresponding the storage location a to the first storage knowledge network data to generate a corresponding position corresponding to the storage location a; Determine, according to the corresponding position and the M allocation migration parameters in the allocation migration information a, the M third associated storage locations of the storage location a in the first storage knowledge network data and the knowledge vector data of each third associated storage location; According to the storage optimization strategy corresponding to the storage location a, storage optimization is performed on the knowledge vector data of the M third associated storage locations.

10. An intelligent warehousing optimization and scheduling system based on the Internet of Things and AI, characterized in that: The intelligent warehousing optimization and scheduling system based on the Internet of Things and AI includes a processor and a machine-readable storage medium, in which machine-executable instructions are stored. The machine-executable instructions are loaded and executed by the processor to implement the intelligent warehousing optimization and scheduling method based on the Internet of Things and AI according to any one of claims 1 to 9.

Images