Warehouse distribution intelligent management visualization method and system based on artificial intelligence

Abstract

The invention discloses a warehouse distribution intelligent management visualization method and system based on artificial intelligence. The method comprises the following steps: S1, collecting and preprocessing warehouse distribution data; s2, constructing a state scheduling matrix, optimizing a distribution path by using an ant colony resonance network, and strengthening a high-frequency path; s3, generating a task allocation table, clustering delivery paths, and constructing a conflict matrix based on path crossing, order overlapping and resource occupation; s4, in combination with the conflict matrix and the scheduling matrix, calculating a structure entropy and a distribution entropy, and generating a task entropy graph; s5, generating a tree information entropy graph, mapping task nodes, adjusting graph layout, and reflecting scheduling priority and resource load; s6, the path optimization result, the task table and the entropy graph are synchronously rendered, and an interactive visual scheduling interface is generated; according to the method, path optimization and entropy graph visualization methods are fused, intelligent scheduling and graph presentation of warehouse allocation tasks are achieved, efficiency is improved, conflicts are reduced, and decision intuition is enhanced.

Description

Technical Field

[0001] The present invention relates to the technical field of warehousing and distribution management, and particularly to an intelligent management visualization method and system for warehousing and distribution based on artificial intelligence. Background Art

[0002] In the process of modern logistics and supply chain management, warehousing and distribution, as an important link connecting production and consumption, directly determine the agility and cost structure of enterprise operations in terms of efficiency and coordination. Especially in the context of the rapid development of the e-commerce, manufacturing, and retail industries, the coordinated management of warehousing and distribution has become a key part of the intelligent logistics system. At present, a large number of warehousing and distribution scheduling strategies based on rule engines and manual experience are widely used in the industry. Although they have a certain adaptability in static scenarios, when facing the logistics scheduling requirements of dynamic, highly concurrent, and multi-dimensional coordination, there are often problems such as lagging response, resource waste, and suboptimal paths, and they cannot meet the refined management goals of intelligence, real-time, and high efficiency.

[0003] In the field of warehousing and distribution management, traditional systems mostly use fixed processes and manual configuration for task assignment and path scheduling. Such methods often rely on manually setting priority rules or historical experience to build models, lacking the ability of adaptive learning and real-time optimization, and it is difficult to cope with the fluctuations brought by different times, different regions, and diverse order structures. For example, when the order volume surges, vehicle resources are tense, or warehouse inventory is abnormal, traditional scheduling strategies usually cannot adjust the path or assign tasks according to the real-time status, resulting in delivery delays, resource conflicts, and even the phenomenon of "dead zone warehouses". At the same time, due to the mutual coupling of multi-dimensional elements such as warehouse resources, vehicle resources, and personnel resources, it is difficult for the scheduling mechanism formulated based on experience to coordinate the operating relationships between various nodes, and the overall efficiency of the system is limited.

[0004] On the other hand, although some enterprises have tried to introduce artificial intelligence algorithms for path optimization and task prediction in recent years, most of them still stay at the level of isolated algorithm operation and fail to closely integrate the AI analysis results with the warehousing and distribution business processes. For example, the path optimization algorithm is only used for single-task assignment and fails to continuously update and feedback learn; or the prediction model cannot be linked with the actual task assignment mechanism and only exists as an auxiliary decision-making tool without truly driving the evolution of the scheduling strategy. In addition, many current intelligent scheduling platforms lack visualization expression capabilities, making it difficult for managers to understand task distribution, resource occupancy, and conflict bottlenecks through an intuitive interface, further restricting the implementation and optimization of intelligent scheduling strategies.

[0005] In terms of visualization, current mainstream platforms typically use tables, line charts, or simple heat maps to display task status and path information. However, these platforms are unable to reflect the dynamic evolution of resources and the parallelization of multiple tasks in complex warehousing and distribution scenarios. This is especially true in scenarios with frequent resource conflicts or dense scheduling nodes. Traditional visualization methods fail to provide a clear logical structure and operational guidance, often leaving users stranded with information redundancy and difficult judgment. Furthermore, existing visualization methods are mostly static graphics, lacking a mechanism to connect with real-time backend data, and are unable to reflect the impact of path adjustments on the global resource layout, resulting in a lack of verifiability and traceability of scheduling results.

[0006] From the perspective of underlying data organization, warehouse and distribution data is typically characterized by high dimensionality, heterogeneity, and strong temporal correlations, involving multiple dimensions such as inventory data, order information, vehicle status, and task execution records. Existing platforms fail to construct a unified scheduling matrix during data modeling, resulting in fragmented data processing flows. This leads to inconsistent algorithm input dimensions and confusing data semantics, hindering the construction of a complete scheduling closed loop. Furthermore, key variables in resource scheduling exhibit implicit resonance and coupling relationships, such as the degree of intersection between delivery routes, the feedback effect between vehicle-task overlap, and inventory fluctuations. Current technologies lack intelligent mechanisms capable of characterizing and utilizing these coupling characteristics.

[0007] Furthermore, while information entropy theory has been extensively studied in information processing and feature extraction, research on constructing task entropy graphs based on resource conflicts and scheduling complexity is relatively new in the field of warehousing and distribution task scheduling. Existing methods are unable to effectively characterize the load complexity and path coupling strength of scheduling nodes using information entropy metrics. This makes it difficult to highlight key bottlenecks when visualizing task layouts and to form a priority hierarchy between tasks, hindering refined management and policy optimization.

[0008] Therefore, how to provide an artificial intelligence-based visualization method and system for intelligent warehouse and distribution management is an urgent problem that needs to be solved by those skilled in the art. Summary of the Invention

[0009] One purpose of the present invention is to propose a visualization method and system for intelligent warehouse and distribution management based on artificial intelligence. The present invention makes full use of state scheduling matrix modeling, ant colony resonance network path optimization, resource conflict factor evaluation and information entropy function construction technology, and describes in detail the whole process of realizing path enhancement, task allocation, conflict modeling and entropy graph visualization based on multi-source warehouse and distribution data. It has the advantages of high scheduling accuracy, high resource utilization, strong conflict prediction ability and strong visual interactivity.

[0010] According to an embodiment of the present invention, a method for visualizing intelligent warehouse and distribution management based on artificial intelligence includes the following steps:

[0011] S1. Collect warehouse and distribution data and perform preprocessing;

[0012] S2. Based on the preprocessed warehouse and distribution data, construct a state scheduling matrix, optimize the distribution path in the state scheduling matrix using an ant colony resonance network, and strengthen the selection of high-frequency paths through a node resonance mechanism;

[0013] S3. Generate a task assignment table according to the path optimization result and the state scheduling matrix, and perform clustering calculation on the distribution paths in the task assignment table. Use path intersection degree, order overlap degree, and resource occupancy ratio as evaluation indicators to construct a resource conflict matrix;

[0014] S4. Based on the resource conflict matrix and the state scheduling matrix, construct an information entropy function, calculate the structural entropy and distribution entropy of each scheduling node, and generate a task entropy graph structure with weight identifiers;

[0015] S5. Generate a two-dimensional tree-shaped information entropy graph according to the task entropy graph structure, perform tree mapping on each task node in the task assignment table, and adjust the tree graph layout to reflect task priority, resource load, and path coupling strength;

[0016] S6. Synchronously render the path optimization result, the task assignment table, and the information entropy graph to generate a nested visual scheduling interface, and implement real-time refresh and interactive query operations.

[0017] Optionally, the warehouse and distribution data includes inventory quantity, order information, inbound and outbound time, distribution path, vehicle location, and cargo status. The preprocessing includes missing value filling, format standardization, timestamp alignment, and data cleaning operations;

[0018] Optionally, the scheduling state matrix uses order information as the index field, and inventory quantity, vehicle location, and cargo status as the feature fields, and generates a scheduling time series according to the inbound and outbound time and the distribution path;

[0019] Optionally, the task assignment table includes order number, starting warehouse node, target distribution node, optimal distribution path, and corresponding time period;

[0020] Optionally, the specific content of S2 includes:

[0021] S21. Construct a state scheduling matrix M based on the preprocessed warehouse and distribution data. The state scheduling matrix where n represents the total number of orders, m represents the number of dimensions of the scheduling features, f1 represents the inventory quantity field, f2 represents the current vehicle location field, f3 represents the current cargo status field, and m ij represents the value in the i-th row and j-th column of the state scheduling matrix, that is, the value of the i-th order on the j-th scheduling feature;

[0022] S22. Optimize the distribution path field in the state scheduling matrix M using the ant colony resonance network. Initialize the path graph G = (V, E), where V represents the set of scheduling nodes and E represents the set of path edges. Define the path pheromone matrix Path resonance function R ij (t) represents the resonance intensity between node v i and node v j at time t. The update formula for the path pheromone weight matrix is as follows:

[0023]

[0024] Among them, Φ ij (t + 1) represents the path pheromone matrix between node v i and node v j at time t + 1. ρ represents the pheromone evaporation coefficient, α represents the path enhancement factor, d ij represents the path distance from node v i to node v j . K represents the number of ant individuals, ω k represents the path preference weight of the k-th ant individual, represents the resonance response of the k-th ant to the path (v i , v j ) at time t;

[0025] S23. Calculate the path reinforcement probability matrix based on the path pheromone weight matrix The path selection probability calculation formula is:

[0026]

[0027] Among them, P ij (t) represents the probability of being selected from node v i to node v j at time t. Φ ij (t) represents the path pheromone matrix between node v i and node v j at time t. Φ il (t) represents the path pheromone matrix between node v i and node v l at time t. γ represents the weight exponent of the pheromone factor, δ represents the weight exponent of the heuristic function factor, η ij (t) represents the expected heuristic function value of the path (v i , v j ) at time t. σ ij (t) represents the path (v i , v j)The normalized historical usage frequency at time t, σ il (t) represents the normalized historical usage frequency of the path (v i , v l ) at time t, N i represents the set of nodes adjacent to node v i ;

[0028] S24. Extract the path with the maximum probability value between each pair of nodes in the path enhancement probability matrix as the path optimization result for task assignment calculation.

[0029] Optionally, the S3 specifically includes:

[0030] S31. Based on the path enhancement probability matrix P and the state scheduling matrix M, construct a task assignment table T, where the task assignment table T = {τ1, τ2, …, τ n},where τ i = <o i , s i , d i , r i , t i >, o i represents the i-th order number, s i represents the starting warehouse node, d i represents the target delivery node, r i represents the optimal path index, which comes from the path with the maximum probability value in P, and t i represents the estimated delivery time period;

[0031] S32. Perform clustering analysis on all path fields r i in the task assignment table to generate path clustering clusters C = {C1, C2, …, C k}, and construct a path cross-degree matrix

[0032]

[0033] where k is the set number of clusters, X ab represents the path cross-degree between path clustering clusters C a and C b , |r i ∩r j | represents the number of overlapping nodes between path r i and path r j , |r i | represents the total number of nodes in path r i , min(|r i |, |r j |) is the smaller value in the path lengths, and |C a | represents cluster Ca The number of tasks in the medium

[0034] S33. Calculate the order overlap matrix based on the task assignment table Among them, the order overlap O ij Indicates order o i And order o j The occupancy overlap degree in the same resource interval, and n represents the total number of orders;

[0035] Combined with the path intersection matrix X, the order overlap matrix O and the vehicle resource scheduling matrix Construct a resource conflict matrix

[0036]

[0037] Among them, R ij Indicates task τ i And task τ j The resource conflict value between them, λ1, λ2, λ3 are conflict weighting coefficients, satisfying λ1 + λ2 + λ3 = 1, δ ia Indicates task τ i Whether it belongs to cluster C a δ jb Indicates task τ j Whether it belongs to cluster C b When belonging, it is 1, otherwise it is 0, v iq Indicates task τ i The usage value on the qth vehicle resource, v jq Indicates task τ j The usage value on the qth vehicle resource, p represents the number of dimensions of the vehicle resource, k is the set number of clusters, and max(·) represents the maximum value function;

[0038] S34. Store the resource conflict matrix as the basis for task conflict evaluation, and use it to construct the information entropy function and generate visualization graphics.

[0039] ]>Optionally, the construction method of the order overlap matrix O is: according to the time period and resource node corresponding to each order in the task assignment table, count the number of times that any two orders occupy the same resource node in the same time period, and calculate the overlap degree between orders based on the time overlap interval and resource coincidence degree, and generate a symmetric matrix O with dimensions of n×n. Among them, the order overlap O ij Indicates order o i And order o j The occupancy overlap degree in the same resource interval, and n represents the total number of orders. [[ID=[]64]]

[0040] Optionally, the specific content of S4 includes:

[0041] S41. Based on the resource conflict matrix R and the status scheduling matrix M, obtain the task set Θ = {θ1, θ2, …, θ q} corresponding to each scheduling node, where represents the subset of tasks assigned to the j-th scheduling node, q represents the total number of scheduling nodes, and T represents the task allocation table;

[0042] S42. Construct the structural entropy H s (j) and the distribution entropy H d (j) of the scheduling node. The calculation expressions of the information entropy function are as follows:

[0043]

[0044]

[0045] where H j represents the task entropy value of the j-th scheduling node, β1 and β2 are the weighting coefficients of the structural entropy and the distribution entropy, satisfying β1 + β2 = 1, m ik represents the value of the i-th task in the k-th feature dimension of the status scheduling matrix, m i′k represents the value of the i'-th task in the k-th feature dimension of the status scheduling matrix, R il represents the conflict value between tasks τ i and task τ l in the resource conflict matrix, |θ j | represents the number of tasks on scheduling node j, log2(·) represents the logarithmic function, n represents the total number of orders, and m represents the number of dimensions of the scheduling features;

[0046] S43. According to the task entropy values H j of all scheduling nodes, construct the task entropy graph structure G H = (V H , E H , W H ), where V H represents the set of scheduling nodes, E H represents the set of resource flow edges between scheduling nodes, W H represents the weight identifier corresponding to the task entropy value H j of each scheduling node. Each node v j ∈ V H is assigned the weight H j .

[0047] Optionally, the S5 specifically includes:

[0048] S51. Based on the task entropy graph structure G H = (V H , E H , WH ) With the task assignment table T = {τ1, τ2, …, τ n}, construct the task node mapping relationship μ: T → V H , where τ i represents the i-th task in the task assignment table, and μ(τ i ) = v j ∈ V H means mapping the task τ i to the scheduling node v j . If the node identifiers of s i and v j are the same, the mapping is established;

[0049] S52. Construct a two-dimensional tree structure T H = (N, L) based on the task node mapping relationship, where N represents the set of task nodes, represents the directed parent-child relationship. If there is a delivery path transfer relationship between the task τ i and the task τ j , then let τ i → τ j be a directed edge and add it to L;

[0050] S53. Calculate the graphic layout weight Ψ i of each task node in the two-dimensional tree information entropy graph, and construct the weight function Ψ:

[0051]

[0052] where Ψ i represents the node layout weight of the task τ i in the tree graph, H j represents the task entropy value of the j-th scheduling node, R il represents the conflict value between the task τ i and the task τ l in the resource conflict matrix, r i represents the optimal path of the task τ i , κ1, κ2, κ3 are weight factors, satisfying κ1 + κ2 + κ3 = 1, n represents the total number of orders, max(·) represents the maximum value function, and q represents the total number of scheduling nodes;

[0053] S54. Dynamically adjust the size, color, and branch width of the task nodes in the tree information entropy graph according to the graphic layout weight to form a visual hierarchical structure, so that the nodes with high task priority, heavy resource load, and high path coupling intensity occupy prominent positions in the graph.

[0054] A warehouse distribution intelligent management visualization system based on artificial intelligence according to an embodiment of the present invention includes:

[0055] A data processing module for collecting warehouse distribution data and performing preprocessing;

[0056] A path optimization module for constructing a status scheduling matrix based on the preprocessed warehouse distribution data, optimizing the distribution path in the status scheduling matrix using an ant colony resonance network, and strengthening the selection of high-frequency paths through a node resonance mechanism;

[0057] A task allocation module for generating a task allocation table according to the path optimization result and the status scheduling matrix, and performing clustering calculation on the distribution paths in the task allocation table, using path intersection degree, order overlap degree, and resource occupancy ratio as evaluation indicators to construct a resource conflict matrix;

[0058] An information entropy construction module for constructing an information entropy function based on the resource conflict matrix and the status scheduling matrix, calculating the structural entropy and distribution entropy of each scheduling node, and generating a task entropy graph structure with weight identifiers;

[0059] An information entropy graph visualization module for generating a two-dimensional tree-shaped information entropy graph according to the task entropy graph structure, performing tree mapping on each task node in the task allocation table, and adjusting the tree graph layout to reflect task priority, resource load, and path coupling strength;

[0060] A scheduling rendering module for synchronously rendering the path optimization result, the task allocation table, and the information entropy graph, generating a nested visual scheduling interface, and implementing real-time refresh and interactive query operations.

[0061] The beneficial effects of the present invention are:

[0062] First of all, by constructing a unified status scheduling matrix, the present invention effectively integrates inventory information, order information, inbound and outbound time, distribution path, vehicle location, and cargo status in the warehouse distribution data, solves the problems of heterogeneous data sources, inconsistent formats, and fragmented processing processes in the prior art, and provides a high-quality data basis for subsequent intelligent scheduling and path optimization. By introducing an ant colony resonance network and constructing a path pheromone weight matrix, the present invention realizes the reinforcement learning and adaptive optimization of the distribution path, overcomes the drawbacks of traditional scheduling methods relying on static rules and untimely path updates, and significantly improves the global optimality and dynamic adaptability of path decision-making.

[0063] Secondly, in the task allocation process of the present invention, a joint evaluation mechanism of path intersection degree, order overlap degree, and resource occupancy ratio is introduced, and a resource conflict matrix is constructed, thereby realizing the quantitative modeling and risk identification of resource conflict situations in the scheduling process, and improving the scientificity and robustness of task allocation. On this basis, the constructed task entropy graph structure can effectively characterize the structural entropy and distribution entropy of each scheduling node, and realize the measurement and classification of task complexity. This design breaks through the limitation of the existing technology in lacking quantitative descriptions of task bottlenecks and resource hotspots, making the scheduling strategy have stronger perception and response capabilities.

[0064] Finally, the present invention generates a two-dimensional tree-shaped information entropy graph based on the information entropy function, and combines the path optimization result with the task allocation table to construct a nested visual scheduling interface that can be refreshed in real time and interactively operated, solving the problems of weak expression ability of the visual interface and poor linkage with backend data in the existing technology. This visual interface can not only clearly present task priorities, resource loads, and path coupling states, but also support managers to make decisions and adjustments quickly based on the graphical interface, thereby realizing intelligent control, efficient execution, and precise optimization of warehouse distribution scheduling. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] The drawings are used to provide a further understanding of the present invention, and constitute a part of the specification. They are used to explain the present invention together with the embodiments of the present invention, and do not constitute a limitation to the present invention. In the drawings:

[0066] Figure 1 is a flowchart of a visual method for intelligent warehouse distribution management based on artificial intelligence proposed by the present invention;

[0067] Figure 2 is a flowchart for constructing a path pheromone and probability matrix of a visual method for intelligent warehouse distribution management based on artificial intelligence proposed by the present invention;

[0068] Figure 3 is a module structure diagram of a visual system for intelligent warehouse distribution management based on artificial intelligence proposed by the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0069] Now, the present invention will be further described in detail with reference to the accompanying drawings. These drawings are all simplified schematic diagrams, only illustrating the basic structure of the present invention in a schematic manner, so they only show the components related to the present invention.

[0070] Refer to Figure 1-2 , a visual method for intelligent warehouse distribution management based on artificial intelligence, includes the following steps:

[0071] S1. Collect warehouse distribution data and perform preprocessing;

[0072] S2. Based on the preprocessed warehouse and distribution data, construct a status scheduling matrix, optimize the distribution paths in the status scheduling matrix using an ant colony resonance network, and strengthen the selection of high-frequency paths through a node resonance mechanism;

[0073] S3. Generate a task assignment table according to the path optimization results and the status scheduling matrix, and perform clustering calculations on the distribution paths in the task assignment table. Use the path intersection degree, order overlap degree, and resource occupancy ratio as evaluation indicators to construct a resource conflict matrix;

[0074] S4. Based on the resource conflict matrix and the status scheduling matrix, construct an information entropy function, calculate the structural entropy and distribution entropy of each scheduling node, and generate a task entropy graph structure with weight identifiers;

[0075] S5. Generate a two-dimensional tree-shaped information entropy graph according to the task entropy graph structure, perform tree mapping on each task node in the task assignment table, and adjust the tree graph layout to reflect task priorities, resource loads, and path coupling strengths;

[0076] S6. Synchronously render the path optimization results, the task assignment table, and the information entropy graph to generate a nested visual scheduling interface, and implement real-time refresh and interactive query operations.

[0077] Through the full-process design from data collection, scheduling modeling, path optimization, task assignment, information entropy analysis to visual rendering, the present invention realizes the intelligent modeling and dynamic optimization control of the entire process tasks of warehouse and distribution, improves the scheduling efficiency, response speed, and overall resource utilization rate, and meets the intelligent scheduling requirements of warehouse and distribution under variable scenarios.

[0078] In this embodiment, the warehouse and distribution data includes inventory quantity, order information, inbound and outbound time, distribution path, vehicle location, and cargo status. The preprocessing includes missing value filling, format standardization, timestamp alignment, and data cleaning operations;

[0079] The present invention refines the definition of the content and preprocessing method of warehouse and distribution data, which helps to improve the quality and structural consistency of the original data, provides a solid data foundation for the construction of the subsequent status scheduling matrix and the input of AI algorithms, and solves the problems of data missing, structural chaos, and processing delay in traditional systems.

[0080] In this embodiment, the scheduling status matrix uses order information as the index field, inventory quantity, vehicle location, and cargo status as the feature fields, and generates a scheduling time sequence according to the inbound and outbound time and the distribution path;

[0081] The present invention clarifies the field design and timing construction rules of the state scheduling matrix, so that the scheduling data has a clear structured representation, improves the input interpretability and model convergence efficiency of the path optimization algorithm, and supports unified modeling of complex multi-dimensional warehouse and distribution states, enhancing the system's adaptability to scheduling scenarios.

[0082] In this embodiment, the task allocation table includes the order number, the starting storage node, the target delivery node, the optimal delivery path and the corresponding time period;

[0083] The present invention clarifies the mapping relationship between the core elements of the task in the task allocation table, which helps to improve the practical implementation of the path optimization results, builds a bridge between scheduling instructions and order execution, and enables the system to accurately connect warehousing and distribution behaviors during the execution process, reducing execution deviation and scheduling delay.

[0084] In this embodiment, S2 specifically includes:

[0085] S21, constructing a state scheduling matrix M based on the pre-processed warehouse and distribution data, the state scheduling matrix Where n represents the total number of orders, m represents the number of dimensions of the scheduling feature, f1 represents the inventory quantity field, f2 represents the vehicle's current location field, f3 represents the cargo's current status field, and m represents the number of dimensions of the scheduling feature. ij Represents the value of the i-th row and j-th column in the state scheduling matrix, that is, the value of the i-th order on the j-th scheduling feature;

[0086] S22. Use the ant colony resonance network to optimize the delivery path field in the state scheduling matrix M, initialize the path graph G = (V, E), where V represents the scheduling node set and E represents the path edge set, and define the path pheromone matrix Path resonance function R ij (t) represents the node v at time t i With node v j The resonance strength between them, the path pheromone weight matrix update formula is:

[0087]

[0088] Among them, Φ ij (t+1) indicates that at time t+1, node v i With node v j The path pheromone matrix between them, ρ represents the pheromone volatility coefficient, α represents the path enhancement factor, d ij Represents node v i To node v j The path distance, K represents the number of ant individuals, ω k represents the path preference weight of the kth ant individual, Denote the resonance response of the k-th ant to the path (v i , v j ) at time t;

[0089] S23. Calculate the path reinforcement probability matrix based on the path pheromone weight matrix The path selection probability calculation formula is:

[0090]

[0091] where P ij (t) represents the probability of being selected from node v i to node v j at time t, Φ ij (t) represents the path pheromone matrix between node v i and node v j at time t, Φ il (t) represents the path pheromone matrix between node v i and node v l at time t, γ represents the weight exponent of the pheromone factor, δ represents the weight exponent of the heuristic function factor, η ij (t) represents the expected heuristic function value of the path (v i , v j ) at time t, σ ij (t) represents the normalized historical usage frequency of the path (v i , v j ) at time t, σ il (t) represents the normalized historical usage frequency of the path (v i , v l ) at time t, N i represents the set of nodes adjacent to node v i ;

[0092] S24. Extract the path with the maximum probability value between each pair of nodes in the path reinforcement probability matrix as the path optimization result for task assignment calculation.

[0093] Through introducing the ant colony resonance network in path optimization and constructing the path pheromone weight matrix and the path reinforcement probability matrix, the present invention realizes the global learning and dynamic update of path selection, has strong adaptive ability and reinforcement memory characteristics, and significantly improves the intelligence and robustness of path planning compared with the traditional static path scheduling method.

[0094] In this embodiment, the S3 specifically includes:

[0095] S31. Based on the path reinforcement probability matrix P and the state scheduling matrix M, construct a task assignment table T, where the task assignment table T = {τ1, τ2, …, τ n}, where τ i => o i , s i , d i , r i , t i , o i represents the i-th order number, s i represents the starting warehouse node, d i represents the target delivery node, r i represents the optimal path index, which comes from the path with the largest probability value in P, and t i represents the estimated delivery time period;

[0096] S32. Conduct clustering analysis on all path fields r i in the task assignment table to generate path clustering clusters C = {C1, C2, …, C k}, and construct a path cross-degree matrix

[0097]

[0098] where k is the set number of clusters, X ab represents the path cross-degree between path clustering clusters C a and C b , |r i ∩ r j | represents the number of node coincidences between path r i and path r j , |r i | represents the total number of nodes in path r i , min(|r i |, |r j |) is the smaller value in the path lengths, and |C a | represents the number of tasks in cluster C a ;

[0099] S33. Based on the task assignment table, calculate the order overlap degree matrix where the order overlap degree O ij represents the occupancy overlap degree of order o i and order o j in the same resource interval, and n represents the total number of orders;

[0100] Combine the path cross-degree matrix X, the order overlap degree matrix O, and the vehicle resource scheduling matrix to construct a resource conflict matrix

[0101]

[0102] Among them, R ij represents the resource conflict value between task τ i and task τ j , λ1, λ2, λ3 are conflict weighting coefficients, satisfying λ1 + λ2 + λ3 = 1, and δ ia represents whether task τ i belongs to cluster C a , δ jb represents whether task τ j belongs to cluster C b , when it belongs, it is 1, otherwise it is 0, and v iq represents the usage value of task τ i on the qth vehicle resource, and v jq represents the usage value of task τ j on the qth vehicle resource. p represents the number of dimensions of the vehicle resource, k is the set number of clusters, and max(·) represents the maximum value function;

[0103] S34. Store the resource conflict matrix as the basis for task conflict evaluation, which is used to construct the information entropy function and generate the visualization graph.

[0104] Construct the resource conflict matrix based on path clustering and task attribute analysis, so that the present invention can perceive potential conflicts and bottleneck positions in advance during the task allocation stage, effectively avoid problems such as resource overload, path conflict, and delivery overlap, and improve the rationality of overall task allocation and the stability of system operation.

[0105] In this embodiment, the construction method of the order overlap degree matrix O is as follows: according to the time period and resource node corresponding to each order in the task allocation table, count the number of times any two orders occupy the same resource node within the same time period, calculate the overlap degree between orders based on the time overlap interval and resource coincidence degree, and generate a symmetric matrix O with dimensions of n×n, where the order overlap degree O ij represents the occupancy overlap degree of order o i and order o j in the same resource interval, and n represents the total number of orders.

[0106] The present invention maps resource conflict factors and task complexity to entropy value indicators by defining the calculation formulas of structural entropy and distribution entropy, forming a structured task entropy graph structure, which not only improves the system's ability to express complex task scheduling states, but also provides a quantitative basis for task priority sorting and visualization layout.

[0107] In this embodiment, the specific content of S4 includes:

[0108] S41. Based on the resource conflict matrix R and the state scheduling matrix M, obtain the task set Θ = {θ1, θ2, …, θ q} corresponding to each scheduling node, where represents the task subset assigned to the j-th scheduling node, q represents the total number of scheduling nodes, and T represents the task allocation table;

[0109] S42. Construct the structural entropy H s (j) and the distribution entropy H d (j) of the scheduling node. The calculation expressions of the information entropy function are as follows:

[0110]

[0111] Among them, H j represents the task entropy value of the j-th scheduling node, β1 and β2 are the weighting coefficients of the structural entropy and the distribution entropy, satisfying β1 + β2 = 1, m ik represents the value of the i-th task in the k-th feature dimension in the state scheduling matrix, m i′k represents the value of the i'-th task in the k-th feature dimension in the state scheduling matrix, R il represents the conflict value between tasks τ i and task τ l in the resource conflict matrix, |θ j | represents the number of tasks on scheduling node j, log2(·) represents the logarithmic function, n represents the total number of orders, and m represents the number of dimensions of the scheduling features;

[0112] S43. According to the task entropy values H j of all scheduling nodes, construct the task entropy graph structure G H = (V H , E H , W H ), where V H represents the set of scheduling nodes, E H represents the set of resource flow edges between scheduling nodes, W H represents the weight identifier corresponding to the task entropy value H j of each scheduling node. Each node v j ∈ V H is assigned the weight H j .

[0113] As a graphical expression form of the scheduling state, the task entropy graph structure not only provides a comprehensive description of the task load and the intensity of resource conflicts, but also constructs the flow relationship between tasks, enhances the controllability, logical hierarchy and graphical expression between tasks, and provides auxiliary decision-making support for task distribution and node scheduling.

[0114] In this embodiment, the specific content of S5 includes:

[0115] S51, based on the task entropy graph structure G H =(V H ,E H ,W H ) and the task allocation table T = {τ1,τ2,…,τ n}, construct the task node mapping relationship μ:T→V H , where τ i represents the i-th task in the task assignment table, μ(τ i )=v j ∈V H Indicates that the task τ i Mapped to scheduling node v j , if s i With v j If the node identifiers of are consistent, the mapping is established;

[0116] S52, construct a two-dimensional tree structure T based on the task node mapping relationship H =(N,L), where N represents the set of task nodes, Represents a directed parent-child relationship. If the task τ i With the task τ j If there is a delivery path transfer relationship, let τ i →τ j Add L to a directed edge;

[0117] S53, calculate the graphic layout weight Ψ of each task node in the two-dimensional tree information entropy graph i , construct the weight function Ψ:

[0118]

[0119] Among them, i Represents the task τ i The node layout weight in the tree graph, H j represents the task entropy value of the jth scheduling node, R il Represents the task τ in the resource conflict matrix i With the task τ l The conflict value between i Represents the task τ i The optimal path of , κ1, κ2, κ3 are weight factors, satisfying κ1+κ2+κ3=1, n represents the total number of orders, max(·) represents the maximum function, and q represents the total number of scheduling nodes;

[0120] S54. Dynamically adjust the sizes, colors, and branch widths of task nodes in the tree-shaped information entropy graph according to the graphic layout weights to form a visual hierarchy, so that nodes with high task priorities, heavy resource loads, and high path coupling intensities occupy prominent positions in the graph.

[0121] The present invention realizes the graphical expression of the complex task distribution state on a two-dimensional tree-shaped information entropy graph. By dynamically adjusting the representation methods of task nodes in the graph according to the graphic layout weights, it highlights key tasks and resource hotspots, improves the visual perception ability of managers for the global scheduling state, and enhances the interactivity and decision-making efficiency of the system.

[0122] Reference Figure 3 , a warehouse and distribution intelligent management visualization system based on artificial intelligence, including:

[0123] A data processing module for collecting warehouse and distribution data and performing preprocessing;

[0124] A path optimization module for constructing a state scheduling matrix based on the preprocessed warehouse and distribution data, optimizing the distribution paths in the state scheduling matrix using an ant colony resonance network, and strengthening the selection of high-frequency paths through a node resonance mechanism;

[0125] A task allocation module for generating a task allocation table according to the path optimization result and the state scheduling matrix, and performing clustering calculations on the distribution paths in the task allocation table. Using the path intersection degree, order overlap degree, and resource occupancy ratio as evaluation indicators, a resource conflict matrix is constructed;

[0126] An information entropy construction module for constructing an information entropy function based on the resource conflict matrix and the state scheduling matrix, calculating the structural entropy and distribution entropy of each scheduling node, and generating a task entropy graph structure with weight identifiers;

[0127] An information entropy graph visualization module for generating a two-dimensional tree-shaped information entropy graph according to the task entropy graph structure, performing a tree mapping on each task node in the task allocation table, and adjusting the tree graph layout to reflect task priorities, resource loads, and path coupling intensities;

[0128] A scheduling rendering module for synchronously rendering the path optimization result, the task allocation table, and the information entropy graph to generate a nested visual scheduling interface, and implementing real-time refresh and interactive query operations.

[0129] The warehouse and distribution intelligent management visualization system proposed by the present invention integrates multiple functional modules such as data processing, path optimization, task allocation, information entropy modeling, and graphic rendering, has high integration and intelligent response capabilities, and can be widely applied to intelligent logistics, supply chain scheduling, and large-scale warehouse and distribution scenarios to realize intelligent scheduling, visual display, and efficient execution of warehouse and distribution resources.

[0130] Example 1:

[0131] To verify the feasibility of the present invention in implementation, the present invention is applied to the intelligent warehousing and distribution center of a large e-commerce enterprise located in the Industrial Park of Suzhou. The daily order volume processed by this distribution center is about 50,000 to 70,000 orders, the supporting warehousing space exceeds 30,000 square meters, and the service covers six major cities in the East China region. Previously, the enterprise used a task scheduling method driven by a traditional rule engine, which had problems such as repeated distribution paths, frequent vehicle conflicts, slow inventory turnover, and low resource utilization rate. Especially during the peak promotion period, situations of warehouse congestion and delayed delivery were prone to occur, seriously affecting customer satisfaction and logistics cost control.

[0132] In this embodiment, first, by deploying the artificial intelligence warehousing and distribution intelligent management visualization method provided by the present invention, data such as inbound and outbound records, real-time inventory status, order structure, distribution route information, vehicle location, and task status at the warehousing end are collected, and missing value filling, field standardization, and timestamp alignment are performed through a unified data processing module. On this basis, the system constructs a status scheduling matrix, integrating the order dimension, inventory status, and distribution resource status for subsequent AI model calls.

[0133] Subsequently, the ant colony resonance network proposed by the present invention is used to optimize the distribution path. The system extracts path nodes from the scheduling matrix and constructs a path pheromone weight matrix based on the resonance mechanism. During the continuous five-day operation, the system strengthens the high-frequency paths and gradually replaces the inefficient paths by dynamically adjusting the path selection probability. In the optimized scheduling, compared with the original system, the average path coincidence rate decreased by 38%, and the vehicle cross-scheduling conflict decreased by 52%.

[0134] After completing the path optimization, the system automatically generates a task allocation table and calculates a resource conflict matrix based on the path intersection degree, order overlap degree, and resource occupancy ratio. For example, on March 28, 2025, a total of 43,279 distribution tasks were generated. Through cluster analysis, 18 types of path aggregation clusters were identified. The three regions with the highest path overlap density were marked as high-risk conflict regions by the system and were preferentially arranged for off-peak scheduling.

[0135] Based on the conflict matrix and the scheduling matrix, the system further calculates the structural entropy and distribution entropy of each scheduling node, forms a task entropy map, and maps it into a two-dimensional tree-shaped information entropy map. This graphical interface is deployed on the large screen of the scheduling center, and the system is refreshed in real time every 10 seconds, supporting schedulers to deeply view task details and predicted risks through click interaction.

[0136] The entire test period was from March 25th to March 31st, 2025. The system processed a total of 452,000 orders. The average length of the delivery route decreased from the original 17.6 kilometers to 13.9 kilometers. The average number of orders dispatched per bicycle per day increased by 23.4%. The resource utilization rate during peak hours increased from 65.7% to 84.2%. The average delay of the scheduling response feedback by the system was shortened to 1.7 seconds. The feedback information from the scheduling staff showed that through the visual task entropy map, high-load areas and resource bottlenecks could be quickly identified, reducing the number of manual scheduling interventions by more than 40%. The following is a comparison of some scheduling data in the embodiment:

[0137] Table 1 Comparison Table of Key Scheduling Indicators Before and After Application

[0138]

[0139]

[0140] As can be seen from the above table, after deploying the method of the present invention, the optimization of the warehouse distribution path and the effect of task scheduling have been significantly improved, various conflict indicators have decreased significantly, the system resource utilization efficiency and response speed have been significantly enhanced, bringing stable and efficient scheduling performance to the enterprise. At the same time, through the deployment of the visual information entropy map, the entire scheduling process has achieved a closed-loop from data-driven to graphic-assisted decision-making, verifying the feasibility and remarkable effect of the present invention in the actual logistics scenario.

[0141] The above is only a preferred specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention, according to the technical solution of the present invention and its inventive concept, makes equivalent substitutions or changes, and should be covered by the protection scope of the present invention.

Claims

1. An artificial intelligence-based visual method for intelligent management of warehousing and distribution, characterized in that, It includes the following steps: S1. Collect warehouse and distribution data and perform preprocessing; S2. Based on the preprocessed warehouse and distribution data, construct a state scheduling matrix, use the ant colony resonance network to optimize the distribution path in the state scheduling matrix, and strengthen the selection of high-frequency paths through the node resonance mechanism; S3. Generate a task assignment table according to the path optimization result and the state scheduling matrix, and perform clustering calculation on the distribution paths in the task assignment table. Use the path intersection degree, order overlap degree and resource occupancy ratio as evaluation indicators to construct a resource conflict matrix; S4. Based on the resource conflict matrix and the state scheduling matrix, construct an information entropy function, calculate the structural entropy and distribution entropy of each scheduling node, and generate a task entropy graph structure with weight identifiers; S5. Generate a two-dimensional tree-shaped information entropy graph according to the task entropy graph structure, perform tree mapping on each task node in the task assignment table, and adjust the tree graph layout to reflect the task priority, resource load and path coupling strength; S6. Synchronously render the path optimization result, task assignment table and information entropy graph to generate a nested visual scheduling interface, and implement real-time refresh and interactive query operations.

2. The visualized method for intelligent warehouse and distribution management based on artificial intelligence according to claim 1, characterized in that, The warehouse and distribution data includes inventory quantity, order information, inbound and outbound time, distribution path, vehicle location and cargo status. The preprocessing includes missing value filling, format standardization, timestamp alignment and data cleaning operations.

3. The visualized method for intelligent management of warehousing and distribution based on artificial intelligence according to claim 1, wherein The scheduling state matrix uses order information as the index field, inventory quantity, vehicle location and cargo status as the feature fields, and generates a scheduling time sequence according to the inbound and outbound time and distribution path.

4. An intelligent management visualization method for warehouse distribution based on artificial intelligence according to claim 1, characterized in that, The task assignment table includes order number, starting warehouse node, target distribution node, optimal distribution path and corresponding time period.

5. A visualization method for intelligent warehouse and distribution management based on artificial intelligence according to claim 1, characterized in that The specific content of S2 includes: S21. Construct a status scheduling matrix M based on the preprocessed warehouse distribution data. The status scheduling matrix where n represents the total number of orders, m represents the number of dimensions of scheduling features, f1 represents the inventory quantity field, f2 represents the current vehicle location field, f3 represents the current cargo status field, and m ij represents the value in the i-th row and j-th column of the status scheduling matrix, that is, the value of the i-th order on the j-th scheduling feature; S22. Use the ant colony resonance network to optimize the delivery path field in the state scheduling matrix M. Initialize the path graph G=(V, E), where V represents the set of scheduling nodes and E represents the set of path edges, and define the path pheromone matrix Path resonance function R ij (t) represents the resonance intensity between node v i and node v j at time t. The update formula for the path pheromone weight matrix is as follows: Among them, Φ ij (t + 1) represents the pheromone matrix of the path between node v i and node v j , ρ represents the pheromone evaporation coefficient, α represents the path enhancement factor, d ij represents the path distance from node v i to node v j , K represents the number of ant individuals, ω k represents the path preference weight of the k-th ant individual, represents the resonance response of the k-th ant to the path (v i , v j ) at time t; S23. Calculate the path reinforcement probability matrix based on the path pheromone weight matrix The path selection probability calculation formula is as follows: Among them, P ij (t) represents the probability of being selected from node v i to node v j at time t. Φ ij (t) represents the pheromone matrix of the path between node v i and node v j at time t. Φ il (t) represents the pheromone matrix of the path between node v i and node v l at time t. γ represents the weight exponent of the pheromone factor, δ represents the weight exponent of the heuristic function factor, and η ij (t) represents the expected heuristic function value of the path (v i , v j ) at time t. σ ij (t) represents the normalized historical usage frequency of the path (v i , v j ) at time t. σ il (t) represents the normalized historical usage frequency of the path (v i , v l ) at time t. N i represents the set of nodes adjacent to node v i ; S24. Extract the path with the maximum probability value between each pair of nodes in the path reinforcement probability matrix as the path optimization result for task assignment calculation.

6. The visualized method for intelligent management of warehouse distribution based on artificial intelligence according to claim 1, characterized in that, The specific content of S3 includes: S31. Based on the path reinforcement probability matrix P and the state scheduling matrix M, construct a task assignment table T, where the task assignment table T = {τ1, τ2, …, τ n}, where τ i = <o i , s i , d i , r i , t i >, o i represents the i-th order number, s i represents the starting warehouse node, d i represents the target delivery node, r i represents the optimal path index, which comes from the path with the largest probability value in P, and t i represents the estimated delivery time period; S32. For all path fields r in the task allocation table i perform clustering analysis to generate path clustering clusters C = {C1, C2, …, C k}, and construct a path cross-degree matrix where k is the set number of clusters, X ab represents the path intersection degree between path cluster C a and C b ; |r i ∩r j | represents the number of node coincidences between path r i and path r j ; |r i | represents the total number of nodes in path r i ; min(|r i |, |r j |) is the smaller value of the path lengths; |C a | represents the number of tasks in cluster C a ; S33. Calculate the order overlap degree matrix based on the task assignment table where the order overlap degree O ij represents the occupancy overlap degree of order o i and order o j in the same resource interval, and n represents the total number of orders; Combine the path intersection degree matrix X, the order overlap degree matrix O, and the vehicle resource scheduling matrix Construct a resource conflict matrix Among them, R ij represents the resource conflict value between task τ i and task τ j . λ1, λ2, λ3 are conflict weighting coefficients, satisfying λ1 + λ2 + λ3 = 1. δ ia represents whether task τ i belongs to cluster C a . δ jb represents whether task τ j belongs to cluster C b . When it belongs, it is 1, otherwise it is 0. v iq represents the usage value of task τ i on the q-th vehicle resource. v jq represents the usage value of task τ j on the q-th vehicle resource. p represents the number of dimensions of the vehicle resource, k is the set number of clusters, and max(·) represents the maximum value function; S34. Store the resource conflict matrix as the basis for task conflict evaluation, which is used to construct the information entropy function and generate visualization graphics.

7. A visualization method for intelligent management of warehouse and distribution based on artificial intelligence according to claim 6, characterized in that, The construction method of the order overlap degree matrix O is as follows: According to the time periods and resource nodes corresponding to each order in the task assignment table, count the number of times any two orders occupy the same resource node within the same time period, calculate the overlap degree between orders based on the time overlap interval and the resource coincidence degree, and generate a symmetric matrix O with a dimension of n×n, where the order overlap degree O ij represents order o i and order o j in the occupancy overlap degree within the same resource interval, and n represents the total number of orders.

8. A visualization method for intelligent warehouse distribution management based on artificial intelligence according to claim 1, characterized in that, The specific content of S4 includes: S41. Based on the resource conflict matrix R and the state scheduling matrix M, obtain the task set Θ = {θ1, θ2, …, θ q} corresponding to each scheduling node, where represents the task subset allocated to the j-th scheduling node, q represents the total number of scheduling nodes, and T represents the task allocation table; S42. Construct the structural entropy H of the scheduling node s (j) and the distribution entropy H d (j). The calculation expression of the information entropy function is as follows: Among them, H j represents the task entropy value of the j-th scheduling node, β1 and β2 are the weighted coefficients of the structural entropy and the distribution entropy, satisfying β1 + β2 = 1, m ik represents the value of the i-th task in the k-th feature dimension in the state scheduling matrix, m i′k represents the value of the i'-th task in the k-th feature dimension in the state scheduling matrix, R il represents the conflict value between task τ i and task τ l in the resource conflict matrix, |θ j | represents the number of tasks on the scheduling node j, log2(·) represents the logarithmic function, n represents the total number of orders, and m represents the number of dimensions of the scheduling features; S43. Based on the task entropy value H of all scheduling nodes j , construct a task entropy graph structure G H = (V H , E H , W H ), where V H represents the set of scheduling nodes, E H represents the set of resource flow edges between scheduling nodes, W H represents the weight identifier corresponding to the task entropy value H j of each scheduling node. Each node v j ∈ V H is assigned a weight H j .

9. A visualization method for intelligent management of warehouse and distribution based on artificial intelligence according to claim 1, characterized in that The specific content of S5 includes: S51. Based on the task entropy graph structure G H =(V H , E H , W H ) and the task assignment table T = {τ1, τ2, …, τ n}, construct the task node mapping relationship μ: T → V H , where τ i represents the i-th task in the task assignment table, and μ(τ i ) = v j ∈ V H means mapping the task τ i to the scheduling node v j . If the node identifiers of s i and v j are the same, the mapping is established; S52. Construct a two-dimensional tree structure T based on the task node mapping relationship H =(N, L), where N represents the set of task nodes, represents the directed parent-child relationship. If there is a delivery path transfer relationship between task τ i and task τ j , then let τ i →τ j be a directed edge and add it to L; S53. Calculate the graphic layout weight Ψ of each task node in the two-dimensional tree-shaped information entropy graph i , and construct a weight function Among them, Ψ i represents the node layout weight of task τ i in the tree diagram, H j represents the task entropy value of the j-th scheduling node, R il represents the conflict value between task τ i and task τ l in the resource conflict matrix, r i represents the optimal path of task τ i where κ1, κ2, κ3 are weight factors satisfying κ1 + κ2 + κ3 = 1, n represents the total number of orders, max(·) represents the maximum value function, and q represents the total number of scheduling nodes; S54. Dynamically adjust the size, color and branch width of the task nodes in the tree-shaped information entropy graph according to the graphic layout weight to form a visual hierarchical structure, so that the nodes with high task priority, heavy resource load and high path coupling strength occupy a prominent position in the graph.

10. An intelligent management visualization system for warehouse and distribution based on artificial intelligence, which executes an intelligent management visualization method for warehouse and distribution based on artificial intelligence according to any one of claims 1 to 7, characterized in that, It includes: A data processing module for collecting warehouse and distribution data and performing preprocessing; A path optimization module for constructing a state scheduling matrix based on the preprocessed warehouse and distribution data, using the ant colony resonance network to optimize the distribution path in the state scheduling matrix, and strengthening the selection of high-frequency paths through the node resonance mechanism; A task assignment module for generating a task assignment table according to the path optimization result and the state scheduling matrix, and performing clustering calculation on the distribution paths in the task assignment table. Use the path intersection degree, order overlap degree and resource occupancy ratio as evaluation indicators to construct a resource conflict matrix; An information entropy construction module for constructing an information entropy function based on the resource conflict matrix and the state scheduling matrix, calculating the structural entropy and distribution entropy of each scheduling node, and generating a task entropy graph structure with weight identifiers; The information entropy graph visualization module is used to generate a two-dimensional tree-shaped information entropy graph according to the task entropy graph structure, perform a tree-shaped mapping on each task node in the task assignment table, and adjust the tree graph layout to reflect task priorities, resource loads, and path coupling strengths; The scheduling rendering module is used to synchronously render the path optimization results, task assignment table, and information entropy graph, generate a nested visual scheduling interface, and implement real-time refreshing and interactive query operations.

Images