Virtual goods allocation dynamic mapping and elastic storage space management method and system

Abstract

The invention relates to the technical field of warehouse management, and discloses a virtual goods allocation dynamic mapping and elastic warehouse space management method and system. The method comprises the following steps: acquiring a storage space multi-source data set including goods allocation physical attributes, goods flow records and storage environment parameters; generating a topological relation graph based on the physical attributes of the goods allocation, generating a virtual goods allocation allocation strategy according to the goods flow record, and processing the storage environment parameters to generate capacity demand fluctuation characteristics; and inputting the result into a multi-dimensional decision model to generate an optimization vector, constructing an elastic adjustment strategy graph through a constraint satisfaction algorithm, and outputting a virtual goods allocation dynamic mapping scheme. The system comprises a multi-source data acquisition module, a topology modeling module, a strategy generation module, a capacity prediction module, a multi-dimensional decision module and an elastic adjustment module. The storage space utilization rate and the operation efficiency can be improved, environmental changes can be effectively coped with, and the enterprise operation cost is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of warehouse management, and in particular to a method and system for dynamic mapping of virtual cargo spaces and flexible warehouse space management. Background Art

[0002] In the context of the rapid development of the modern logistics industry, warehouse management, as a key link in the supply chain, has a profound impact on the operating costs and competitiveness of enterprises through its efficiency and space utilization. The traditional warehouse space management model has many drawbacks and is difficult to meet the current complex and changing warehouse needs.

[0003] Traditional warehouse management usually adopts a fixed cargo space allocation method, that is, the goods are fixedly placed in specific physical cargo spaces. This method can maintain a certain management order when the types and flow of goods are relatively stable. However, with the diversification of market demand and the booming development of the e-commerce industry, the frequency, types and flow of goods entering and leaving the warehouse have shown great volatility. For example, during e-commerce promotion activities, the volume of goods entering and leaving the warehouse may surge several times or even dozens of times in a short period of time, and fixed cargo space allocation cannot flexibly respond to such traffic changes. The cargo spaces of some popular commodities may be severely congested, while other cargo spaces are idle, resulting in low overall utilization of storage space and a significant increase in the cost of storing and handling goods.

[0004] At the same time, traditional cargo space management does not make full use of the physical properties of cargo spaces. In the fixed cargo space mode, the physical properties of cargo spaces, such as spatial coordinates and load-bearing capacity, are simply considered as basic conditions for cargo storage. For example, some cargo spaces with stronger load-bearing capacity may be used to store lighter goods, while heavier goods need to be transported to a more distant area because they cannot find suitable cargo spaces. This not only wastes cargo space resources, but also increases the time and labor costs in the cargo handling process, reducing the efficiency of warehousing operations.

[0005] In addition, traditional warehouse management lacks effective analysis and utilization of storage environment parameters. Storage environment parameters such as temperature, humidity, ventilation conditions, etc., will have a significant impact on the storage quality and safety of goods. However, under the traditional management model, the allocation of cargo space is rarely adjusted dynamically according to changes in environmental parameters. For some temperature-sensitive medicines or foods, if the storage environment temperature is too high or too low, it may cause the goods to deteriorate and be damaged, causing huge economic losses to the company. Moreover, due to the lack of monitoring and analysis of environmental parameters, it is impossible to predict changes in storage capacity demand in advance, making it difficult for companies to quickly make reasonable storage space adjustment decisions when faced with emergencies.

[0006] In terms of the records of goods flow, the traditional management mode only simply records the inbound and outbound information of goods, without deeply mining and analyzing data such as the associated paths of goods and the frequency of inbound and outbound. This results in the inability to conduct scientific storage location allocation according to the flow laws of goods, and the inability to achieve rapid inbound and outbound of goods and efficient storage. For example, for goods combinations that often enter and leave the warehouse simultaneously, if they can be placed in adjacent storage locations, the handling distance and time can be greatly reduced, but the traditional management mode is difficult to achieve this.

[0007] With the development of advanced technologies such as big data and artificial intelligence, the limitations of the traditional warehousing management mode have become increasingly prominent. Enterprises urgently need a method and system that can comprehensively consider multi-source data, realize dynamic mapping of virtual storage locations and elastic management of warehousing space, so as to improve warehousing management efficiency, reduce costs, enhance the competitiveness of enterprises, and meet the rapid development needs of the modern logistics industry. Summary of the Invention

[0008] The purpose of the present invention is to provide a method and system for dynamic mapping of virtual storage locations and elastic management of warehousing space to solve the problems raised in the above background technology.

[0009] To achieve the above purpose, the present invention provides the following technical solution: A method for dynamic mapping of virtual storage locations and elastic management of warehousing space, the method includes:

[0010] Obtain a multi-source data set of the warehousing space; the multi-source data includes the physical attributes of storage locations, the records of goods flow, and the warehousing environment parameters; the physical attributes of storage locations include spatial coordinates, load-bearing capacity, and storage location status tags, and the records of goods flow include the frequency of inbound and outbound, the associated paths of goods, and the operation timestamps;

[0011] Based on the physical attributes of the storage locations, generate a topological relationship graph of the warehousing space through a graph structure modeling algorithm, and the topological relationship graph includes the connection strength and hierarchical dependence weights between storage location nodes;

[0012] According to the records of goods flow, generate a virtual storage location allocation strategy through a dynamic decision optimization algorithm, and the strategy includes virtual storage location creation rules, allocation priorities, and mapping update conditions;

[0013] Perform time series segmentation processing on the warehousing environment parameters to generate capacity demand fluctuation characteristics;

[0014] Input the topological relationship graph, the virtual storage location allocation strategy, and the capacity demand fluctuation characteristics into a multi-dimensional decision model to generate an optimization vector of the warehousing space;

[0015] Based on the optimized vector, construct an elastic adjustment strategy graph through a constraint satisfaction algorithm, and output a dynamic mapping scheme for virtual storage locations; the nodes of the elastic adjustment strategy graph represent warehousing operation units, and the edges represent the timing constraints and resource consumption weights between operations.

[0016] Preferably, the topological relationship graph of the warehousing space is generated through a graph structure modeling algorithm, including:

[0017] Perform spatial dimension normalization processing on the physical attributes of the storage locations to generate standardized storage location nodes;

[0018] Based on the adjacency relationship between the storage location nodes, divide the space area through a hierarchical clustering algorithm, and calculate the topological compactness of each area;

[0019] Generate a hierarchical connection strength according to the topological compactness and a preset dependence threshold;

[0020] Combine the hierarchical connection strength with the storage location status label to form a multi-dimensional topological relationship graph.

[0021] Preferably, the virtual storage location allocation strategy is generated through a dynamic decision optimization algorithm, including:

[0022] Perform path correlation analysis on the cargo flow records, and extract high-frequency operation paths and cargo combination patterns;

[0023] Initialize a candidate strategy pool based on the virtual storage location creation rule, and calculate the allocation priority score of each strategy;

[0024] Screen the optimal allocation strategy that meets the mapping update conditions through a strategy iteration algorithm, and update the virtual storage location mapping table.

[0025] Preferably, the time series segmentation processing of the warehousing environment parameters includes:

[0026] Divide dynamic time windows according to the operation timestamps, and extract the mean and variance of the environment parameters within each window;

[0027] Identify the abnormal change interval of the capacity demand through a fluctuation detection algorithm, and mark it as a key fluctuation segment;

[0028] Compare the key fluctuation segment with the historical capacity data to generate the capacity demand fluctuation characteristics.

[0029] Preferably, the multi-dimensional decision model includes a feature fusion module and a constraint mapping module, and the feature fusion module includes:

[0030] Perform normalization processing on the hierarchical dependence weights in the topological relationship graph to obtain a first fusion vector;

[0031] Discretely encode the allocation priority scores in the virtual storage location allocation strategy to generate a second fusion vector;

[0032] Perform trend fitting calculation on the capacity demand fluctuation characteristics, extract the long-term evolution pattern, and obtain a third fusion vector;

[0033] Combine the first fusion vector, the second fusion vector, and the third fusion vector into a unified decision sequence through a weighted superposition layer.

[0034] Preferably, the constraint mapping module includes:

[0035] Align the unified decision sequence in the spatial dimension to generate a constraint correlation matrix;

[0036] Extract the dependency features between operation units through a hierarchical attention mechanism to generate a dependency mapping matrix;

[0037] Perform matrix convolution operation on the constraint correlation matrix and the dependency mapping matrix to generate constraint fusion features;

[0038] Superimpose the constraint fusion features and the original unified decision sequence through residual connection, and output the optimized vector of the storage space.

[0039] Preferably, constructing an elastic adjustment strategy graph through a constraint satisfaction algorithm includes:

[0040] Initialize the node attributes according to the storage operation units, and generate an edge weight matrix based on the resource consumption weights;

[0041] Use the optimized vector as the node state, and the edge weight matrix consists of the priority of the time sequence constraint and the resource consumption weight;

[0042] Iteratively update the constraint satisfaction degree of each node through a cost function, and adjust the edge weight matrix;

[0043] Generate an optimal elastic adjustment sequence covering all operation units according to the adjusted edge weight matrix.

[0044] Preferably, the parameter optimization method of the hierarchical clustering algorithm includes:

[0045] Calculate the initial clustering radius and the minimum regional density according to the historical storage location distribution data, traverse the parameter combinations through a grid search algorithm, and select the parameters with the highest matching degree between the clustering result and the manually divided area;

[0046] Dynamically adjust the clustering radius and the minimum regional density according to the matching degree, and optimize the division accuracy of the hierarchical connection strength.

[0047] Preferably, the construction method of the cost function includes:

[0048] Define the cost value between nodes as the balance coefficient of the timing constraint and the resource consumption weight;

[0049] Initialize the constraint satisfaction degree of each node to zero, and the cost value of the starting point to the preset reference value;

[0050] Calculate the minimum cumulative cost value of each node based on the previous nodes through the dynamic programming algorithm, and record the optimal adjustment path;

[0051] Derive the complete elastic adjustment sequence reversely according to the optimal adjustment path.

[0052] Preferably, the present invention further includes a virtual storage location dynamic mapping and elastic storage space management system, and the system includes:

[0053] Multi-source data acquisition module: used to acquire the multi-source data set of the storage space, and the multi-source data includes the physical attributes of the storage location, the goods flow record and the storage environment parameters; wherein, the physical attributes of the storage location include the space coordinates, the load-bearing capacity and the storage location status label, and the goods flow record includes the inbound and outbound frequency, the goods association path and the operation timestamp;

[0054] Topological modeling module: configured to generate a topological relationship graph of the storage space based on the physical attributes of the storage location through a graph structure modeling algorithm, and the topological relationship graph includes the connection strength and the hierarchical dependence weight between the storage location nodes;

[0055] Policy generation module: used to generate a virtual storage location allocation policy according to the goods flow record through a dynamic decision optimization algorithm, and the policy includes the virtual storage location creation rule, the allocation priority and the mapping update condition;

[0056] Capacity prediction module: perform time series segmentation processing on the storage environment parameters to generate the capacity demand fluctuation characteristics;

[0057] Multi-dimensional decision module: input the topological relationship graph, the virtual storage location allocation policy and the capacity demand fluctuation characteristics into a multi-dimensional decision model to generate an optimized vector of the storage space; the multi-dimensional decision model includes a feature fusion module and a constraint mapping module;

[0058] Elastic adjustment module: used to construct an elastic adjustment policy graph based on the optimized vector through a constraint satisfaction algorithm, and output a virtual storage location dynamic mapping scheme; the nodes of the elastic adjustment policy graph represent the storage operation units, and the edges represent the timing constraints and the resource consumption weights between the operations.

[0059] Compared with the prior art, the beneficial effects of the present invention are:

[0060] In terms of improving the utilization rate of warehouse space, this method comprehensively analyzes multi-source data such as the physical attributes of storage locations, the records of goods flow, and the parameters of the warehouse environment. Based on the physical attributes of storage locations, a graph structure modeling algorithm is used to generate a topological relationship graph of the warehouse space, which can clearly present the connection strength and hierarchical dependence weights between storage location nodes, thus more reasonably planning the layout of storage locations. For storage locations with different load-bearing capacities, goods can be accurately allocated according to their weights to avoid waste of resources. According to the records of goods flow, the generated virtual storage location allocation strategy can dynamically adjust the storage location allocation according to the frequency of goods inbound and outbound and the associated paths. Goods with high inbound and outbound frequencies are allocated to more easily operable virtual storage locations, reducing the handling distance and improving the space utilization rate. During major e-commerce promotions, the system can quickly adjust the storage locations, allowing popular goods to be concentrated in convenient areas and enhancing the actual utilization efficiency of the warehouse space.

[0061] From the perspective of improving the efficiency of warehouse operations, through path correlation analysis of the records of goods flow, high-frequency operation paths and goods combination patterns are extracted, providing a strong basis for virtual storage location allocation. This optimizes the goods handling paths and reduces unnecessary handling operations. In actual warehouse operations, handling equipment such as forklifts do not need to frequently travel between different areas to handle goods, saving a large amount of time. At the same time, the construction of the elastic adjustment strategy graph, based on the optimization vector and constraint satisfaction algorithm, clarifies the timing constraints and resource consumption weights between warehouse operation units, ensuring the orderly progress of various operations and further enhancing the operation efficiency.

[0062] In response to changes in the warehouse environment, the time series of warehouse environment parameters is segmented to generate the characteristics of capacity demand fluctuations. In this way, the system can predict in advance the changes in warehouse capacity demand and timely adjust the dynamic mapping scheme of virtual storage locations. During the high-temperature period in summer, for warehouses storing temperature-sensitive goods, the system can adjust the storage locations in advance according to the temperature change trend, transfer these goods to areas with appropriate temperatures, ensure the storage quality of goods, and avoid losses of goods caused by environmental changes.

[0063] In terms of the adaptability and scalability of the system, the algorithms and models in this invention have good adaptability. The parameters of the hierarchical clustering algorithm can be optimized according to the historical storage location distribution data to make the spatial region division more accurate. The construction method of the cost function can calculate the optimal adjustment path through the dynamic programming algorithm to ensure the efficient operation of the system in different warehouse scenarios. As the business of the enterprise develops and the scale and complexity of the warehouse increase, the system can easily adapt to new warehouse requirements by adjusting the algorithm parameters and optimizing the model structure, and has strong scalability.

[0064] The invention can also reduce the operating costs of enterprises. The efficient utilization of warehousing space and the optimization of operation processes reduce the costs of goods storage and handling. The accurate allocation of storage locations and environmental management reduce the risks of goods damage and deterioration, avoiding economic losses. In terms of labor costs, due to the improvement of operation efficiency, the required manpower is reduced, further reducing the operating costs of enterprises, enhancing the competitiveness of enterprises in the market, and providing strong support for the sustainable development of enterprises. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] Figure 1 is the working principle diagram of the virtual storage location dynamic mapping and flexible warehousing space management method of the present invention;

[0066] Figure 2 is the working flow chart for generating the virtual storage location allocation strategy;

[0067] Figure 3 is the working flow chart for processing the warehousing environment parameters to generate fluctuation characteristics;

[0068] Figure 4 is the flow chart of the constraint mapping module of the multi-dimensional decision-making model. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0069] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0070] Please refer to Figures 1 - 4 , the present invention relates to a virtual storage location dynamic mapping and flexible warehousing space management method and system, and the following will elaborate on its specific implementation manners in detail.

[0071] The multi-source data obtained covers the physical attributes of storage locations, the records of goods flow, and the warehousing environment parameters. In terms of the physical attributes of storage locations, the spatial coordinates are used to accurately determine the position of the storage location in the warehousing space, the load-bearing capacity determines the upper limit of the weight of goods that the storage location can carry, and the storage location status label indicates whether the storage location is currently idle, occupied, or under maintenance, etc. In the records of goods flow, the frequency of inbound and outbound reflects the frequency of goods entering and leaving the warehouse, the goods association path reflects the trajectory of goods moving in the warehouse, and the operation timestamp records the specific time when each inbound and outbound operation occurs. The warehousing environment parameters include the temperature, humidity, ventilation conditions, etc. in the warehouse, and these data will affect the storage conditions of goods and the operation efficiency of the warehouse.

[0072] Using the graph structure modeling algorithm, a topological relationship graph is generated based on the physical attributes of the storage locations. This graph includes the connection strength and hierarchical dependence weights between the storage location nodes. The connection strength reflects the degree of closeness of the association between the storage locations, and the hierarchical dependence weight reflects the importance relationship between storage locations at different levels.

[0073] With the help of the dynamic decision optimization algorithm, a virtual storage location allocation strategy is generated based on the goods flow records. This strategy includes virtual storage location creation rules, allocation priorities, and mapping update conditions to achieve reasonable allocation and effective management of virtual storage locations.

[0074] Perform time series segmentation processing on the warehousing environment parameters to generate the capacity demand fluctuation characteristics. By analyzing the changes in the warehousing environment parameters in different time periods, the fluctuation rules of the warehouse capacity demand are mined, providing a basis for subsequent warehousing space optimization.

[0075] Input the topological relationship graph, virtual storage location allocation strategy, and capacity demand fluctuation characteristics into the multi-dimensional decision model. This model comprehensively considers various factors and generates a warehousing space optimization vector to provide a direction for the optimization and adjustment of the warehousing space.

[0076] Based on the optimization vector, use the constraint satisfaction algorithm to construct an elastic adjustment strategy graph. The nodes in the graph represent warehousing operation units, and the edges represent the timing constraints and resource consumption weights between operations. Finally, a virtual storage location dynamic mapping scheme is output to achieve elastic management of the warehousing space and dynamic mapping of virtual storage locations.

[0077] The present invention will be further described below in conjunction with Embodiments 1 to 5:

[0078] Embodiment 1:

[0079] In this embodiment, the process of generating the warehousing space topological relationship graph through the graph structure modeling algorithm is elaborated in detail. In an actual warehousing scenario, assume there is a large automated stereoscopic warehouse with a large number of storage locations. After obtaining the physical attribute data of the storage locations, perform spatial dimension normalization processing on it. Spatial dimension normalization processing is to convert data such as spatial coordinates of different scales to a unified standard range. For example, normalize the length, width, and height coordinate values of the storage locations to the interval [0, 1]. This can eliminate the influence of the data dimensions of different dimensions and facilitate subsequent calculations and analyses. After processing, standardized storage location nodes are generated, and each standardized storage location node contains information such as normalized spatial coordinates, load-bearing capacity, and storage location status labels.

[0080] Based on the adjacency relationship between storage location nodes, a hierarchical clustering algorithm is used to divide the space area. The adjacency relationship refers to whether two storage locations are adjacent in space. If they are adjacent, it is considered that they have an adjacency relationship. The hierarchical clustering algorithm is a bottom-up clustering method. First, each storage location node is regarded as a separate class, and then the classes are continuously merged according to the similarity between the nodes until a certain termination condition is met. In this embodiment, the Euclidean distance between the storage location nodes is calculated to measure the similarity, and the closer the distance, the higher the similarity. During the merging process, the topological compactness of each region is calculated. The topological compactness is calculated by the formula:

[0081]

[0082] Where is the number of storage location nodes in the region, is the Euclidean distance between the -th and the -th storage location nodes in the region. The topological compactness reflects the degree of aggregation of the storage location nodes in the region, and the smaller the value, the closer the nodes are aggregated.

[0083] The hierarchical connection strength is generated based on the topological compactness and a preset dependency threshold. The preset dependency threshold is a value set according to the actual layout and operation experience of the warehouse. For example, it is set to . When the topological compactness of the region is less than , it indicates that the storage location nodes in this region are closely aggregated and the connection strength between them is large; otherwise, the connection strength is small. Combining the hierarchical connection strength with the storage location status labels forms a multi-dimensional topological relationship graph. Storage location status labels such as the idle label is set to 0, the occupied label is set to 1, etc. In the multi-dimensional topological relationship graph, the storage location nodes with different status labels represent their relationships through connection edges with different strengths, thus constructing a topological relationship graph that can comprehensively reflect the storage space structure, providing an intuitive and effective data basis for subsequent warehousing management decisions.

[0084] Example 2:

[0085] This embodiment focuses on the process of generating a virtual storage location allocation strategy through a dynamic decision optimization algorithm. Taking an e-commerce warehouse as an example, this warehouse has a wide variety of goods and frequent inbound and outbound operations. First, a path correlation analysis is performed on the goods flow records. The goods flow records contain a large amount of inbound and outbound data. By analyzing the goods correlation paths and operation timestamps in these data, high-frequency operation paths and goods combination patterns can be extracted. For example, through data statistics over a period of time, it is found that several certain commodities are often shipped out together and their outbound paths are similar, thus forming a high-frequency operation path and goods combination pattern.

[0086] Initialize the candidate policy pool based on the virtual location creation rules. The virtual location creation rules can be formulated according to the actual needs and operational characteristics of the warehouse. For example, virtual locations can be created based on factors such as the category, size, and frequency of inbound and outbound of goods. Suppose virtual locations are created according to the category of goods, and goods of the same category are assigned to the same virtual location. For each virtual location allocation policy, calculate its allocation priority score , and the calculation formula is:

[0087]

[0088] where, represents the frequency of inbound and outbound of goods, represents the matching degree between the virtual location and the high-frequency operation path, represents the compliance degree between the virtual location and the storage requirements of goods (such as temperature, humidity, etc.). , , are weight coefficients, which are set according to the focus of warehouse operation. For example, if more emphasis is placed on the inbound and outbound efficiency, then can have a relatively larger value.

[0089] Screen the optimal allocation policy that meets the mapping update conditions through the policy iteration algorithm, and update the virtual location mapping table. The policy iteration algorithm is a method of continuously optimizing the policy. Each iteration will adjust the policy according to the current allocation situation and feedback information. The mapping update conditions can be set such that when there are significant changes in the frequency of inbound and outbound of goods, or when the storage requirements of the warehouse change, the virtual location mapping table needs to be updated. In each iteration process, compare the allocation priority scores of different policies, select the policy with the highest score as the current optimal allocation policy, and update the virtual location mapping table to ensure that the allocation of virtual locations can always adapt to the actual operation situation of the warehouse.

[0090] Example 3:

[0091] This example details the specific process of performing time series segmentation processing on warehouse environment parameters. Taking a food warehouse as an example, warehouse environment parameters are crucial for the storage quality of goods. First, divide dynamic time windows according to the operation timestamps. The operation timestamps record the time of each inbound and outbound operation. Based on this time information, the time can be divided into dynamic time windows of different lengths. For example, one day can be used as a time window. Within each time window, extract the mean and variance of the environment parameters. Suppose the environment parameter is the temperature in the warehouse , in the th time window, the calculation formula for the temperature mean is:

[0092]

[0093] Among them, is the number of times the temperature is recorded within the -th time window, is the -th temperature value recorded within the -th time window. The variance is calculated by the formula:

[0094] The variance reflects the fluctuation of the temperature within this time window.

[0095] Identify the abnormal change interval of the capacity demand through the fluctuation detection algorithm and mark it as the key fluctuation segment. The fluctuation detection algorithm can adopt statistical methods. For example, set a threshold. When the temperature mean or variance exceeds this threshold, it is considered that the capacity demand may have an abnormal change. Assume that the threshold of the temperature mean is . When or , mark the -th time window as the key fluctuation segment. Compare the key fluctuation segment with the historical capacity data to generate the capacity demand fluctuation characteristics. The historical capacity data records the actual capacity usage of the warehouse in different past time periods. By comparing the environmental parameters of the key fluctuation segment and the historical capacity data, the fluctuation law of the warehouse capacity demand under different environmental parameter changes can be analyzed. For example, when the temperature rises, the storage period of some foods shortens, and more temporary storage spaces may be required, resulting in an increase in the capacity demand. These fluctuation laws constitute the capacity demand fluctuation characteristics and provide an important basis for the optimization of the storage space.

[0096] Example 4:

[0097] This example details the working principle and process of the feature fusion module in the multi-dimensional decision-making model. Still taking the above e-commerce warehouse as an example, in the feature fusion module, first normalize the hierarchical dependence weights in the topological relationship graph. The hierarchical dependence weights reflect the importance relationship between different hierarchical storage locations. Since the value ranges of these weights may be different, in order to facilitate subsequent fusion calculations, normalization processing is required. Assume that the hierarchical dependence weight is , and the normalized weight is calculated by the formula:

[0098]

[0099] Among them, and are the minimum and maximum values among all the hierarchical dependence weights respectively. After normalization processing, the first fusion vector is obtained.

[0100] Discretely encode the allocation priority scores in the virtual storage location allocation strategy to generate a second fusion vector. The allocation priority score is a numerical value calculated according to the virtual storage location allocation strategy and is used to measure the advantages and disadvantages of different allocation strategies. Discrete encoding is to convert the continuous allocation priority scores into discrete encoded values. For example, the scores can be divided into several intervals, and each interval corresponds to an encoded value. Suppose the range of the allocation priority score is [0, 100], which is divided into three intervals: [0, 30) corresponds to the encoded value 0, [30, 70) corresponds to the encoded value 1, and [70, 100] corresponds to the encoded value 2. In this way, the allocation priority scores are converted into discrete encodings to form the second fusion vector.

[0101] Perform trend fitting calculation on the capacity demand fluctuation characteristics, extract the long-term evolution pattern, and obtain the third fusion vector. The capacity demand fluctuation characteristics contain the information of the change of the warehouse capacity demand over time. Through trend fitting calculation, its long-term evolution trend can be found. Using the polynomial fitting method, assume that the change of the capacity demand over time can be fitted by a polynomial function and determine the coefficients of the polynomial by the least squares method , so as to extract the long-term evolution pattern of the capacity demand and obtain the third fusion vector.

[0102] Finally, merge the first fusion vector, the second fusion vector, and the third fusion vector into a unified decision sequence through a weighted superposition layer. The weighted superposition layer assigns a weight to each vector according to the importance of different vectors, and then adds them up to obtain the unified decision sequence. Suppose the first fusion vector is , the second fusion vector is , the third fusion vector is , and the weights are , , , and the unified decision sequence is calculated by the formula:

[0103]

[0104] Through such a feature fusion process, various different types of feature information are fused together to provide a more comprehensive and accurate basis for subsequent warehousing space optimization decisions.

[0105] Example 5:

[0106] This embodiment focuses on elaborating on constructing an elastic adjustment strategy graph and related content through a constraint satisfaction algorithm. Taking a logistics transfer warehouse as an example, when constructing the elastic adjustment strategy graph, first initialize the node attributes according to the warehousing operation units, and generate an edge weight matrix based on the resource consumption weights. The warehousing operation units include operations such as inbound, outbound, sorting, and storage of goods. For each operation unit, initialize its node attributes, such as the operation type, operation time, required resources, etc. The resource consumption weight reflects the relative magnitude of resource consumption between different operations. Assume that the operation to operation has a resource consumption weight of , and generate an edge weight matrix , and the matrix element .

[0107] Use the optimization vector as the node state, and the edge weight matrix consists of the priority of the timing constraint and the resource consumption weight. The optimization vector is generated by a multi-dimensional decision model, which contains information such as the direction and degree of warehousing space optimization. Map the elements in the optimization vector to the state of each node. For example, if a certain element in the optimization vector represents the adjustment direction of a certain storage location, then use it as the state of the corresponding operation unit node. The priority of the timing constraint in the edge weight matrix is determined according to the sequence relationship between operations. For example, the inbound operation must be carried out before the storage operation, so the priority of the timing constraint of the edge from the inbound operation to the storage operation is relatively high.

[0108] Iteratively update the constraint satisfaction degree of each node through a cost function and adjust the edge weight matrix. The cost function is used to measure the cost of each node under various constraint conditions. When constructing the cost function, define the cost value between nodes as the balance coefficient of the timing constraint and the resource consumption weight. Assume that the cost value from node to node is , and the balance coefficient is , then , where is the priority of the timing constraint from node to node . Initialize the constraint satisfaction degree of each node to zero, and the starting point cost value is a preset reference value, for example, set to . Calculate the minimum cumulative cost value of each node based on the previous nodes through the dynamic programming algorithm and record the optimal adjustment path. The dynamic programming algorithm is a method that decomposes the original problem into relatively simple sub-problems and saves the solutions of the sub-problems to avoid repeated calculations. In this embodiment, starting from the starting node, calculate the minimum cumulative cost value of each node in turn. Assume that the minimum cumulative cost value of node is , then , where is the predecessor node of . Record the optimal adjustment path for each node, that is, the path corresponding to the minimum cumulative cost value from the starting point to this node.

[0109] Derive the complete elastic adjustment sequence reversely according to the optimal adjustment path. Starting from the end node, trace back reversely along the recorded optimal adjustment path, and determine the execution order of each operation unit in turn, so as to generate the optimal elastic adjustment sequence covering all operation units. Finally, generate an elastic adjustment strategy diagram according to the adjusted edge weight matrix, output the virtual storage location dynamic mapping scheme, and realize the efficient management of the storage space and the reasonable dynamic mapping of the virtual storage location.

[0110] It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements not only includes those elements, but also includes other elements not expressly listed, or also includes elements inherent to such process, method, article or device.

[0111] Although the embodiments of the present invention have been shown and described, those of ordinary skill in the art can understand that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for dynamic mapping of virtual cargo space and flexible storage space management, characterized in that: include: Acquire a multi-source data set of the storage space; the multi-source data includes physical attributes of the cargo space, cargo flow records and storage environment parameters; the physical attributes of the cargo space include space coordinates, load-bearing capacity and cargo space status labels, and the cargo flow records include in-and-out frequency, cargo associated paths and operation timestamps; Based on the physical properties of the cargo space, a topological relationship diagram of the storage space is generated through a graph structure modeling algorithm, wherein the topological relationship diagram includes the connection strength and hierarchical dependency weights between cargo space nodes; According to the cargo flow records, a virtual cargo space allocation strategy is generated through a dynamic decision optimization algorithm, wherein the strategy includes virtual cargo space creation rules, allocation priorities, and mapping update conditions; Performing time series segmentation processing on the storage environment parameters to generate capacity demand fluctuation characteristics; Input the topological relationship diagram, virtual cargo space allocation strategy and capacity demand fluctuation characteristics into a multidimensional decision model to generate a storage space optimization vector; Based on the optimization vector, a flexible adjustment strategy graph is constructed through a constraint satisfaction algorithm, and a virtual cargo space dynamic mapping solution is output; the nodes of the flexible adjustment strategy graph represent storage operation units, and the edges represent timing constraints and resource consumption weights between operations.

2. A method for dynamic mapping of virtual cargo space and flexible storage space management according to claim 1, characterized in that: The topological relationship diagram of the storage space is generated by the graph structure modeling algorithm, including: Normalizing the physical properties of the cargo location in spatial dimensions to generate standardized cargo location nodes; Based on the adjacency relationship between cargo location nodes, the spatial area is divided through a hierarchical clustering algorithm, and the topological density of each area is calculated; Generate hierarchical connection strength based on topological tightness and preset dependency threshold; Combine the hierarchical connection strength with the cargo location status label to form a multi-dimensional topological relationship diagram.

3. A method for dynamic mapping of virtual cargo space and flexible storage space management according to claim 1, characterized in that: The generation of a virtual cargo space allocation strategy through a dynamic decision optimization algorithm includes: Perform path correlation analysis on the cargo flow records to extract high-frequency operation paths and cargo combination patterns; Initialize the candidate strategy pool based on the virtual cargo space creation rules and calculate the allocation priority score of each strategy; The optimal allocation strategy that meets the mapping update conditions is screened through the policy iteration algorithm, and the virtual cargo location mapping table is updated.

4. A method for dynamic mapping of virtual cargo space and flexible storage space management according to claim 1, characterized in that: The time series segmentation processing of the storage environment parameters includes: Divide the dynamic time window according to the operation timestamp and extract the mean and variance of the environmental parameters in each window; The fluctuation detection algorithm is used to identify abnormal changes in capacity demand and mark them as key fluctuation segments; Compare key fluctuation segments with historical capacity data to generate capacity demand fluctuation characteristics.

5. A method for dynamic mapping of virtual cargo space and flexible storage space management according to claim 1, characterized in that: The multidimensional decision model includes a feature fusion module and a constraint mapping module, and the feature fusion module includes: Normalizing the hierarchical dependency weights in the topological relationship graph to obtain a first fusion vector; Discretely encode the allocation priority score in the virtual cargo space allocation strategy to generate a second fusion vector; Performing trend fitting calculation on the capacity demand fluctuation characteristics, extracting the long-term evolution pattern, and obtaining a third fusion vector; The first fusion vector, the second fusion vector and the third fusion vector are combined into a unified decision sequence through a weighted superposition layer.

6. A method for dynamic mapping of virtual cargo space and flexible storage space management according to claim 5, characterized in that: The constraint mapping module comprises: Align the spatial dimensions of the unified decision sequence to generate a constraint association matrix; The dependency features between operation units are extracted through the hierarchical attention mechanism to generate a dependency mapping matrix; Perform matrix convolution operation on the constraint association matrix and the dependency mapping matrix to generate constraint fusion features; The constraint fusion features are superimposed on the original unified decision sequence through residual connection, and the storage space optimization vector is output.

7. A method for dynamic mapping of virtual cargo space and flexible storage space management according to claim 1, characterized in that: The elastic adjustment strategy diagram is constructed by using a constraint satisfaction algorithm, including: Initialize node attributes according to the storage operation unit and generate edge weight matrix based on resource consumption weight; The optimization vector is used as the node state, and the edge weight matrix is ​​composed of the priority of the timing constraint and the resource consumption weight; Update the constraint satisfaction of each node through the cost function iteration and adjust the edge weight matrix; The optimal elastic adjustment sequence covering all operation units is generated according to the adjusted edge weight matrix.

8. A method for dynamic mapping of virtual cargo space and flexible storage space management according to claim 2, characterized in that: The parameter optimization method of the hierarchical clustering algorithm includes: The initial clustering radius and minimum area density are calculated based on the historical cargo location distribution data. The parameter combinations are traversed through the grid search algorithm to select the parameters with the highest matching degree between the clustering results and the manually divided areas. The clustering radius and minimum area density are dynamically adjusted according to the matching degree to optimize the accuracy of hierarchical connection strength division.

9. A method for dynamic mapping of virtual cargo space and flexible storage space management according to claim 7, characterized in that: The method for constructing the cost function includes: Define the cost value between nodes as the balance coefficient between timing constraints and resource consumption weights; Initialize the constraint satisfaction of each node to zero and the starting cost value to the preset benchmark value; Calculate the minimum cumulative cost of each node based on the previous node through the dynamic programming algorithm, and record the optimal adjustment path; A complete elastic adjustment sequence is generated by reverse deduction based on the optimal adjustment path.

10. A virtual cargo space dynamic mapping and flexible storage space management system, characterized in that: include: Multi-source data acquisition module: used to acquire a multi-source data set of the storage space, the multi-source data includes physical attributes of the cargo space, cargo flow records and storage environment parameters; wherein the physical attributes of the cargo space include space coordinates, load-bearing capacity and cargo space status label, and the cargo flow records include in-and-out frequency, cargo associated path and operation timestamp; Topological modeling module: configured to generate a topological relationship diagram of the storage space based on the physical properties of the cargo space through a graph structure modeling algorithm, wherein the topological relationship diagram includes the connection strength and hierarchical dependency weight between cargo space nodes; Strategy generation module: used to generate a virtual cargo space allocation strategy based on the cargo flow record through a dynamic decision optimization algorithm, wherein the strategy includes virtual cargo space creation rules, allocation priority and mapping update conditions; Capacity prediction module: performs time series segmentation processing on the storage environment parameters to generate capacity demand fluctuation characteristics; Multidimensional decision module: input the topological relationship diagram, virtual cargo space allocation strategy and capacity demand fluctuation characteristics into a multidimensional decision model to generate a storage space optimization vector; the multidimensional decision model includes a feature fusion module and a constraint mapping module; Elastic adjustment module: used to construct an elastic adjustment strategy graph based on the optimization vector through a constraint satisfaction algorithm, and output a dynamic mapping solution for virtual cargo space; the nodes of the elastic adjustment strategy graph represent storage operation units, and the edges represent timing constraints and resource consumption weights between operations.

Images