Commodity inventory management method and system based on cloud computing

Abstract

The invention relates to the field of inventory management and cloud computing application, in particular to a commodity inventory management method and system based on cloud computing, and the method comprises the steps: collecting commodity dynamic information, environment state data and commodity state data, and generating enhanced data; constructing a dynamic scene graph and simulating resource flow through a distributed simulation system; generating scene prediction data in combination with a quantum heuristic optimization algorithm; constructing a hierarchical inventory structure based on the prediction data, realizing resource dynamic allocation between inventory layers through a multi-agent collaborative optimization algorithm, and generating inventory optimization data; and recording a resource allocation task execution process, and generating transparent inventory state data by using a block chain technology. According to the invention, the inventory structure can be dynamically adjusted in real time, the inventory management precision is improved, the inventory overstock and shortage risks are effectively reduced, and accurate support is provided for supply chain management.

Description

Technical Field

[0001] The present invention relates to the fields of inventory management and cloud computing applications, and in particular to a commodity inventory management method and system based on cloud computing. Background Art

[0002] Through precise inventory control, enterprises can reduce the risk of overstocking while ensuring the smoothness and efficiency of the supply chain. In traditional methods, inventory management often relies on rules of thumb or a single classification method, such as ABC classification and XYZ classification. However, these methods fail to fully cope with complex market demands and changing commodity consumption patterns, resulting in insufficient flexibility and refinement in inventory management.

[0003] The existing technology mainly adopts the following methods: ABC classification method: Classify by commodity value and strengthen the management of high-value commodities. However, this method tends to ignore low-value commodities and affect the diversified market strategy.

[0004] XYZ classification method: classification based on demand fluctuation and predictability. Although it is suitable for commodities with obvious demand fluctuations, its classification relies on historical data and lacks dynamic adjustment capabilities. The main defects of these methods include: single classification standards, lack of dynamic adjustment capabilities, and insufficient adaptability to market changes, resulting in the risk of commodity backlog or out-of-stock. Summary of the invention

[0005] In view of the many problems existing in the above-mentioned prior art, the present invention provides a commodity inventory management method and system based on cloud computing. The present invention constructs a dynamic scene graph to represent the inventory status, uses a distributed simulation system to simulate resource flow, and combines a quantum heuristic optimization algorithm to adjust the scene weight and path cost to achieve dynamic allocation and optimization of inventory resources. Finally, transparent inventory status data is generated to provide real-time support for inventory decision-making and effectively reduce inventory risks.

[0006] A commodity inventory management method based on cloud computing includes the following steps: Collect product dynamic information, environmental status data and product status data, pre-process the collected data, and generate enhanced data; Based on the enhanced data, a dynamic scene graph is constructed, the scene nodes in the dynamic scene graph are deduced using a distributed simulation system, and the scene prediction data is generated in combination with a quantum heuristic optimization algorithm; Based on the scenario prediction data, a hierarchical inventory structure is constructed, which includes dynamic inventory, buffer inventory and strategic inventory. The dynamic allocation of resources among dynamic inventory, buffer inventory and strategic inventory is realized through a multi-agent collaborative optimization algorithm, and inventory optimization data is generated; Based on inventory optimization data, inventory resource allocation tasks are executed, and the execution process of inventory resource allocation tasks is recorded to generate transparent inventory status data.

[0007] Preferably, the dynamic information of the commodity is collected through an RFID tag, and the RFID tag records the unique identification, current location and inventory change information of the commodity; the environmental status data is collected through a temperature and humidity sensor, including the temperature and humidity of the storage environment; the commodity status data is collected through a pressure sensor and a high-definition camera, the pressure sensor is used to detect the pressure state of the commodity surface, and the high-definition camera is used to obtain the appearance image data of the commodity.

[0008] Preferably, the commodity dynamic information, environmental status data and commodity status data are cleaned, redundancy is removed and outlier screening is performed, and the outlier screening includes marking the temperature values ​​in the environmental status data that are lower than -10 degrees Celsius or higher than 50 degrees Celsius as invalid data; the screened data are formatted in a standardized manner, and the environmental status feature data and the commodity status feature data are extracted based on the biometric learning model, and the environmental status feature data and the commodity status feature data are prioritized in combination with a dynamic weight optimization algorithm to generate enhanced data.

[0009] Preferably, the dynamic scene graph is composed of scene nodes and scene edges, wherein the scene nodes represent standard scenes, extreme scenes and abnormal scenes, and the scene edges represent resource flow paths between scene nodes and their adjustment costs, and the adjustment costs are calculated according to the following formula: ,

[0010] in, Represents the node from the scene To scene node resource adjustment costs; Represents a scene node The weight of the scene node Priority; Indicates that the resource is from the scene node Adjust to scene node the time required; Indicates that the resource is from the scene node Adjust to scene node consumption.

[0011] Preferably, a distributed simulation system is used to simulate the resource flow of each scene node in the dynamic scene graph to generate initial resource allocation data for the scene node; a quantum heuristic optimization algorithm is used to dynamically adjust the scene weight distribution in the dynamic scene graph, and the scene edge weights of the dynamic scene graph are recalculated to generate scene prediction data, wherein the scene prediction data includes the optimization weight, resource demand and resource fluctuation range of each scene node.

[0012] Preferably, the hierarchical inventory structure includes dynamic inventory, buffer inventory and strategic inventory, wherein the dynamic inventory is used to meet short-term demand within 1 to 2 days, the buffer inventory is used to meet fluctuating demand within 3 to 7 days, and the strategic inventory is used to support long-term demand planning within 30 days; the dynamic inventory, buffer inventory and strategic inventory are dynamically connected through a resource allocation path to form a resource flow network.

[0013] Preferably, the dynamic allocation of resources among the dynamic inventory, the buffer inventory and the strategic inventory is achieved through a multi-agent collaborative optimization algorithm. Each inventory layer acts as an independent agent node, and when executing resource allocation, the allocation demand is calculated by sharing the current inventory status and resource demand information; the resource allocation path and resource allocation strategy are adjusted according to the feedback mechanism between agents, and inventory optimization data is generated according to scenario feedback.

[0014] Preferably, during the execution of the inventory resource allocation task, blockchain technology is used to record the execution process of the inventory resource allocation task, and the record content includes the source inventory layer, target inventory layer, allocation quantity, execution time and task completion status of the resource allocation; the blockchain record is stored in the form of transaction blocks, and each transaction block includes a unique identifier and specific data of task execution.

[0015] Preferably, the transparent inventory status data includes the current inventory status of the dynamic inventory, the buffer inventory and the strategic inventory, the execution result of the inventory resource allocation task and the real-time feedback information of the task completion rate.

[0016] A system for implementing the cloud computing-based commodity inventory management method described in the claims, comprising: Multimodal data collection module, used to collect commodity dynamic information, environmental status data and commodity status data, where commodity dynamic information is collected through RFID tags, environmental status data is collected through temperature and humidity sensors, and commodity status data is collected through pressure sensors and high-definition cameras; A data preprocessing module, which is in communication with the multimodal data acquisition module, is used to clean the collected data, remove redundancy and filter outliers, standardize the format of the filtered data, extract environmental state feature data and commodity state feature data using a biometric learning model, and generate enhanced data in combination with a dynamic weight optimization algorithm; A dynamic scene graph construction and deduction module is communicatively connected to the data preprocessing module, and is used to construct a dynamic scene graph based on the enhanced data, wherein the dynamic scene graph is composed of scene nodes and scene edges, the scene nodes represent standard scenes, extreme scenes and abnormal scenes, and the scene edges represent resource flow paths and adjustment costs between scene nodes; the dynamic scene graph construction and deduction module further uses a distributed simulation system to simulate resource flow of scene nodes in the dynamic scene graph, and adjusts the scene weight distribution and scene edge weight in the dynamic scene graph in combination with a quantum heuristic optimization algorithm to generate scene prediction data; A hierarchical inventory optimization module is communicatively connected to the dynamic scenario graph construction and deduction module, and is used to construct a hierarchical inventory structure based on the scenario prediction data, wherein the hierarchical inventory structure includes dynamic inventory, buffer inventory and strategic inventory; and realizes dynamic allocation of resources among dynamic inventory, buffer inventory and strategic inventory through a multi-agent collaborative optimization algorithm to generate inventory optimization data; The inventory resource execution module is communicatively connected with the hierarchical inventory optimization module, and is used to execute the inventory resource allocation task based on the inventory optimization data, and use blockchain technology to record the execution process of the inventory resource allocation task, and generate transparent inventory status data including the current inventory status and task execution results.

[0017] Compared with the prior art, the advantages and beneficial effects of the present invention are: The present invention uses dynamic scene graphs to fully reflect inventory changes, breaking through the singleness of traditional classification methods; The present invention combines quantum heuristic optimization algorithms to dynamically adjust resource allocation, improving prediction accuracy and resource utilization efficiency; The present invention uses a multi-agent collaborative optimization algorithm to achieve dynamic allocation among multiple layers of inventory, effectively solving the problems of inventory backlog and shortage. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is a schematic diagram of the process of the present invention; Figure 2 A schematic diagram of dynamic scene graph construction and resource flow in the present invention; Figure 3 It is a schematic diagram of the layered inventory structure in the present invention; Figure 4 This is a schematic diagram of multi-agent collaboration in the present invention; Figure 5 It is a structural block diagram of the system of the present invention. DETAILED DESCRIPTION

[0019] Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the present disclosure. In the following detailed description, for ease of explanation, many specific details are set forth to provide a comprehensive understanding of the embodiments of the present disclosure. However, it is obvious that one or more embodiments may also be implemented without these specific details. In addition, in the following description, descriptions of known structures and technologies are omitted to avoid unnecessary confusion of the concepts of the present disclosure.

[0020] The terms used herein are only for describing specific embodiments and are not intended to limit the present disclosure. The terms "comprise", "include", etc. used herein indicate the existence of the features, steps, operations and / or components, but do not exclude the existence or addition of one or more other features, steps, operations or components.

[0021] All terms (including technical and scientific terms) used herein have the meanings commonly understood by those skilled in the art unless otherwise defined. It should be noted that the terms used herein should be interpreted as having a meaning consistent with the context of this specification and should not be interpreted in an idealized or overly rigid manner.

[0022] like Figure 1 As shown, a commodity inventory management method based on cloud computing includes the following steps: Collect product dynamic information, environmental status data and product status data, pre-process the collected data, and generate enhanced data; Preferably, the dynamic information of the commodity is collected through an RFID tag, and the RFID tag records the unique identification, current location and inventory change information of the commodity; the environmental status data is collected through a temperature and humidity sensor, including the temperature and humidity of the storage environment; the commodity status data is collected through a pressure sensor and a high-definition camera, the pressure sensor is used to detect the pressure state of the commodity surface, and the high-definition camera is used to obtain the appearance image data of the commodity.

[0023] The dynamic information of the commodity is collected through an RFID tag, and the RFID tag communicates with the reader through a built-in radio frequency antenna, and uses radio frequency identification technology to obtain the unique identification, current location and inventory change information of the commodity in real time. In the present invention, the unique identification of the RFID tag is used to distinguish the identity information of each commodity. The identification is associated with the record of the commodity in the inventory system to ensure the uniqueness and accuracy of inventory management; the current location information is calculated by the reader in combination with the positioning algorithm to calculate the current storage or transportation location of the commodity, which helps to achieve visual management of dynamic inventory; the inventory change information records the warehousing, outbound and transit of the commodity, and updates the inventory status of the commodity in real time through the background data processing module. The non-contact information collection of RFID tags can significantly improve the efficiency and accuracy of data collection, avoiding errors and time waste in traditional manual inventory.

[0024] The environmental status data is collected through a temperature and humidity sensor, which monitors the temperature and humidity of the storage environment in real time. The core principle is based on the electrical signal response of temperature and humidity sensitive materials to environmental changes, which is converted into readable temperature and humidity data through a signal processing module. In the present invention, the collection of temperature and humidity data is used to evaluate the suitability of the storage environment, especially when storing perishable items or environmentally sensitive goods, to ensure that the goods are stored under optimal conditions. For example, the storage requirements for a batch of goods are a temperature range of 15°C to 25°C and a humidity range of 40% to 60%. The environmental data collected by the temperature and humidity sensor can be used for real-time comparison. When the environmental parameters exceed the set threshold, the system will trigger an early warning or automatic control equipment (such as air conditioning or dehumidifier) ​​to restore suitable conditions, thereby ensuring the quality of the goods.

[0025] The commodity status data is collected through a pressure sensor and a high-definition camera, wherein the pressure sensor is used to detect the pressure state of the commodity surface. The principle is based on the resistance change of the resistance strain gauge inside the sensor under pressure, which is converted into a pressure value through an amplification circuit. The present invention detects the stress conditions of commodities during storage or transportation through pressure data, for example, to determine whether a stack of goods is damaged due to excessive stacking. When the pressure sensor detects that the pressure of a commodity exceeds the set safety value, for example, when the pressure of a packaging box exceeds 200N, the system will mark the commodity as "abnormally stressed" and record its specific location information for timely adjustment.

[0026] High-definition cameras are used to obtain image data of the appearance of goods, and combined with computer vision technology, they can automatically detect and record the appearance of goods. Specifically, the image data collected by the camera is used to extract features through image processing algorithms, such as edge detection of damaged packaging areas or text recognition of product surface logos, to determine whether there are any abnormalities in the appearance of the goods or to verify the correctness of the goods. For example, in a certain logistics scenario, the image collected by the high-definition camera shows that there are tears on the packaging of a box of goods. The system can mark the product as "abnormal in appearance" through image comparison technology and notify warehouse management personnel for further inspection.

[0027] Preferably, the commodity dynamic information, environmental status data and commodity status data are cleaned, redundancy is removed and outlier screening is performed, and the outlier screening includes marking the temperature values ​​in the environmental status data that are lower than -10 degrees Celsius or higher than 50 degrees Celsius as invalid data; the screened data are formatted in a standardized manner, and the environmental status feature data and the commodity status feature data are extracted based on the biometric learning model, and the environmental status feature data and the commodity status feature data are prioritized in combination with a dynamic weight optimization algorithm to generate enhanced data.

[0028] In the present invention, data cleaning, redundancy removal and outlier screening of commodity dynamic information, environmental status data and commodity status data are key steps to ensure data reliability and decision accuracy. The principle of data cleaning is to identify and correct errors, omissions or inconsistencies in the data through algorithms. In the present invention, the collected data may generate noise or erroneous records due to sensor failure, transmission interruption or external interference, such as repeated commodity location data or outliers in environmental status data. Removing these invalid data through cleaning algorithms can significantly improve data quality.

[0029] The redundancy removal process removes duplicate records by comparing the timestamp and source of the data. For example, if the dynamic information of a product records the same location multiple times in a short period of time, the system will retain the latest record and delete other redundant data. This process reduces the waste of computing resources caused by data redundancy while ensuring the efficiency of updating the real-time status of the system.

[0030] The key to outlier screening is to set a reasonable threshold range. In the present invention, outlier screening of environmental status data is performed by setting the temperature threshold to -10 degrees Celsius to 50 degrees Celsius, and marking temperature data outside this range as invalid data. In specific implementation, the system performs screening through the following judgment logic: ,

[0031] in, Represents temperature data, To judge the result, 1 indicates valid data and 0 indicates invalid data. This screening method is suitable for commodities (such as food or medicine) that need to maintain a suitable temperature range in the storage environment. When the temperature value read by the sensor is -15 degrees Celsius, the system will mark the data as invalid and trigger the sensor calibration or equipment maintenance process to avoid affecting subsequent decisions.

[0032] Format standardization is to uniformly encode and format the cleaned data. For example, all timestamp formats are unified to the ISO 8601 standard (such as "2024-12-04T10:30:00Z") to ensure compatibility and consistency between different data sources. Through standardization, the system can store, retrieve and calculate data more efficiently.

[0033] The process of extracting environmental status feature data and product status feature data based on the biometric learning model involves the application of deep learning technology. The biometric learning model simulates the characteristics of biological neural networks and extracts important features related to product status and environmental status through training with a large amount of historical data. For example, the model may extract features such as temperature and humidity fluctuation rate, maximum value, and minimum value in environmental status data, and pressure mean value, pressure change rate, and surface deformation area in product status data. These features play a key role in determining whether the product is damaged or whether the environment is suitable. The feature data output by the model is expressed in vector form for subsequent analysis.

[0034] The dynamic weight optimization algorithm prioritizes different features after feature extraction. The core of weight optimization is to dynamically adjust the weight value according to the importance of the feature. For example, the system may calculate the weight through the following formula: ,

[0035] in, Representation characteristics The weight of Representation characteristics The importance rating of is the total number of features. Through the optimized weights, the system can focus on the feature data that is most critical to decision making. For example, for products that are sensitive to temperature and humidity, the weight of the temperature and humidity features in the environmental status data will be significantly higher than other features.

[0036] like Figure 2 As shown, based on the enhanced data, a dynamic scene graph is constructed, the scene nodes in the dynamic scene graph are deduced using a distributed simulation system, and the scene prediction data is generated in combination with a quantum heuristic optimization algorithm; Preferably, the dynamic scene graph is composed of scene nodes and scene edges, wherein the scene nodes represent standard scenes, extreme scenes and abnormal scenes, and the scene edges represent resource flow paths between scene nodes and their adjustment costs, and the adjustment costs are calculated according to the following formula: ,

[0037] in, Represents the node from the scene To scene node resource adjustment costs; Represents a scene node The weight of the scene node Priority; Indicates that the resource is from the scene node Adjust to scene node the time required; Indicates that the resource is from the scene node Adjust to scene node consumption.

[0038] In the present invention, the construction of a dynamic scene graph is a key link, and the definition of scene nodes and scene edges provides a basic framework for resource allocation and optimization. The scene nodes represent standard scenes, extreme scenes, and abnormal scenes. These scene nodes are used to characterize different business scenarios that may occur in inventory management. For example, standard scenes correspond to normal inventory demand and allocation, extreme scenes may involve temporary large-volume demand (such as promotional activities or sudden orders), and abnormal scenes are used to handle unpredictable inventory anomalies or supply chain disruptions (such as logistics delays or natural disasters).

[0039] Scene edges are used to describe the resource flow paths and their adjustment costs between scene nodes. These edges reflect the cost of resources flowing from one scene node to another through weights and cost values. Adjustment cost is the core parameter of scene optimization, and its calculation formula is described above.

[0040] The principle of this formula is to comprehensively consider the priority of scene nodes, the time required for resource adjustment, and resource consumption, and integrate the three into an adjustment cost value in a weighted manner. Specifically, the scene node weight Reflects the importance of the current node in the scene. For example, for a node with a sudden increase in demand, its weight will be dynamically increased to give priority to meeting its resource needs. Resource Adjustment Time Reflects the scene node To scene node The efficiency of the route, such as the shortest transportation time calculated by the traffic route optimization model. It includes actual expenses such as transportation energy consumption and labor costs.

[0041] In practical applications, dynamic scene graphs are used to simulate resource flows through distributed simulation systems. Assume that there are three scene nodes A, B, and C, where A represents a standard scene, B represents an extreme scene, and C represents an abnormal scene. At a certain moment, the weight of node A is , the weight of node B is , the weight of node C is The time required for resources to flow from A to B is hours, the consumption is Unit, calculate the adjustment cost: .

[0042] Similarly, the adjustment costs of other edges can also be calculated. In a dynamic scene graph, the system optimizes the resource flow path by calculating the adjustment costs of each edge in real time. , the system will give priority to adjusting resources from A to B to achieve the lowest cost resource allocation.

[0043] Through the real-time adjustment of the dynamic scene graph and its edge weights, the system can dynamically respond to changes in inventory demand. Especially in the case of peak promotions or supply chain disruptions, the dynamic scene graph can quickly reallocate resources to ensure business continuity. For example, during peak promotions, the weights of extreme scene nodes is dynamically promoted, thereby increasing the resource allocation priority from standard scenarios to extreme scenarios and reducing the possibility of out-of-stock situations.

[0044] Preferably, a distributed simulation system is used to simulate the resource flow of each scene node in the dynamic scene graph to generate initial resource allocation data for the scene node; a quantum heuristic optimization algorithm is used to dynamically adjust the scene weight distribution in the dynamic scene graph, and the scene edge weights of the dynamic scene graph are recalculated to generate scene prediction data, wherein the scene prediction data includes the optimization weight, resource demand and resource fluctuation range of each scene node.

[0045] In the present invention, using a distributed simulation system to simulate the resource flow of scene nodes in a dynamic scene graph is a key step in achieving accurate resource optimization. The core principle of the distributed simulation system is to divide the complex resource flow process into multiple independent scene nodes, and simulate the global resource flow through multi-node collaborative computing. Each scene node represents a specific business scenario in inventory management. For example, a standard scene node may represent conventional inventory demand, while an extreme scene node represents a temporary demand peak or an abnormal scenario (such as inventory shortage). Through a distributed architecture, the system can perform parallel computing on multiple scene nodes at the same time to improve simulation efficiency.

[0046] In the specific implementation, the distributed simulation system simulates the process of resources flowing from one node to another based on the scene nodes and scene edge information in the dynamic scene graph. The goal of resource flow simulation is to generate initial resource allocation data for each scene node, which includes the current resource stock, expected resource inflow and outflow of the node. For example, if the initial resource stock of a scene node is 100 units, the inflow is expected to be 50 units, and the outflow is expected to be 30 units, the system will generate the initial resource allocation data as follows: .

[0047] After generating the initial resource allocation data, the system further combines the quantum heuristic optimization algorithm to dynamically adjust the scene weight distribution. The quantum heuristic optimization algorithm is based on the principle of quantum computing, and realizes efficient search for the global optimal solution by constructing a probabilistic superposition state and a quantum interference mechanism. Specifically in the present invention, the quantum heuristic optimization algorithm takes the initial resource allocation data of the scene node as input, and dynamically adjusts the weight value of the node by evaluating the resource demand urgency, flow efficiency and resource fluctuation range of the node. For example, when the resource demand urgency of a scene node is high and the flow efficiency is low, the algorithm will increase the weight of the node to improve its resource allocation priority.

[0048] After adjusting the scene weight distribution, the system recalculates the weight of the scene edge in the dynamic scene graph. The scene edge weight represents the cost or efficiency of the resource flow path. The calculation formula of the scene edge weight is: By recalculating edge weights, the system is able to update the global structure of the scene graph to reflect the latest resource flow priorities and path selections.

[0049] Finally, the system generates scenario prediction data, which includes the optimization weight, resource demand, and resource fluctuation range of each scenario node. These data provide key inputs for the construction and optimization of the subsequent inventory hierarchy. For example, in a logistics warehousing network, scenario prediction data can help the system identify high-demand nodes and low-efficiency nodes, and adjust resource allocation strategies to meet the needs of key nodes. In the specific implementation, the optimization weight of a certain warehousing node A is 0.8, the resource demand is 200 units, and the resource fluctuation range is ±20 units. The scenario prediction data is as follows: .

[0050] Through the combination of distributed simulation system and quantum heuristic optimization algorithm, the present invention can realize efficient modeling and dynamic adjustment of dynamic scene graphs, and the generated scene prediction data provides accurate and real-time decision support for inventory optimization. At the same time, because the quantum heuristic optimization algorithm has the ability to search for the global optimal solution, the present invention is significantly superior to traditional algorithms in terms of resource allocation efficiency and cost control, can adapt to complex inventory management scenarios, and significantly improve the intelligent level of cloud computing commodity inventory management.

[0051] like Figure 3 As shown, based on the scenario prediction data, a hierarchical inventory structure is constructed, and the hierarchical inventory structure includes dynamic inventory, buffer inventory and strategic inventory. The dynamic allocation of resources among dynamic inventory, buffer inventory and strategic inventory is realized through a multi-agent collaborative optimization algorithm, and inventory optimization data is generated; In the present invention, the construction of a hierarchical inventory structure is a key link in optimizing resource allocation and improving response efficiency. The hierarchical inventory structure is divided into dynamic inventory, buffer inventory and strategic inventory according to the function and application scenario of the inventory. These three types of inventory meet the resource allocation needs of different time periods and demand types, thereby realizing hierarchical management and dynamic optimization of inventory resources.

[0052] Dynamic inventory is used to meet short-term demand within 1 to 2 days, mainly covering daily order fulfillment and quick response tasks, such as processing urgent orders from retailers. Dynamic inventory relies on the resource demand and resource fluctuation range in the scenario forecast data for rapid adjustment. Buffer inventory is used to cope with demand fluctuations within 3 to 7 days. Its main function is to provide flexibility in the supply chain and buffer the impact of short-term supply and demand mismatches, such as supporting fluctuating demand during promotional activities. Strategic inventory is used to support long-term demand planning of 30 days or more. Its main function is to ensure the long-term stability of the supply chain through early layout, such as preparing goods in advance before the seasonal demand peak.

[0053] The principle of building a hierarchical inventory structure is based on the node optimization weights and resource requirements in the scenario prediction data. The system divides inventory resources into layers through optimization algorithms to ensure that each layer of inventory can meet the corresponding demand type. For example, if the resource demand of a node A is 200 units and the optimization weight is 0.8, the system will prioritize the demand of high-weight nodes to the dynamic inventory layer, while the resource demand of low-weight nodes may be included in the buffer inventory or strategic inventory layer to balance the utilization efficiency and risk of resources.

[0054] In order to achieve efficient resource allocation between inventory levels, the solution adopts a multi-agent collaborative optimization algorithm. Figure 4As shown in the figure, each inventory layer is defined as an agent node, and these nodes perform collaborative optimization by sharing real-time inventory status and resource demand information. The core of the agent collaborative optimization algorithm is to dynamically adjust the resource allocation path and strategy between each layer of inventory through a feedback mechanism. For example, when the available inventory of the dynamic inventory layer is insufficient to meet the current demand, the agent node will request the buffer inventory layer to provide additional resources through a negotiation mechanism, while ensuring that the inventory of the strategic inventory layer is not over-consumed.

[0055] In practical applications, assume that the current available inventory in the dynamic inventory layer is 150 units, but the demand of node A is 200 units. The intelligent node will allocate 50 units of resources from the buffer inventory layer to the dynamic inventory layer according to the optimization weight of node A and the urgency of resource demand to meet the demand of node A. At the same time, the algorithm will dynamically update the resource status of each layer of inventory and generate inventory optimization data, which includes the available inventory, allocation path and execution time of the current inventory layer.

[0056] Through the hierarchical inventory structure and multi-agent collaborative optimization algorithm, the present invention can achieve refined management and dynamic allocation of resources, significantly improving the flexibility and response speed of the inventory management system. The design of the hierarchical inventory structure enables the system to reasonably balance short-term, medium-term and long-term demands, while the application of the intelligent collaborative optimization algorithm ensures the real-time and efficient allocation of inventory resources. Overall, while ensuring business continuity, the present invention optimizes inventory turnover efficiency and reduces inventory holding costs, providing strong technical support for commodity inventory management based on cloud computing.

[0057] Preferably, the hierarchical inventory structure includes dynamic inventory, buffer inventory and strategic inventory, wherein the dynamic inventory is used to meet short-term demand within 1 to 2 days, the buffer inventory is used to meet fluctuating demand within 3 to 7 days, and the strategic inventory is used to support long-term demand planning within 30 days; the dynamic inventory, buffer inventory and strategic inventory are dynamically connected through a resource allocation path to form a resource flow network.

[0058] In the present invention, the design of the hierarchical inventory structure aims to meet the inventory management needs of different time periods and demand scenarios. Through the division of dynamic inventory, buffer inventory and strategic inventory, refined management and efficient flow of resources are achieved. The dynamic inventory, buffer inventory and strategic inventory are dynamically connected through resource allocation paths, forming a resource flow network with flexible deployment capabilities, thereby ensuring that the system remains stable and efficient in the face of demand changes in different time dimensions.

[0059] Dynamic inventory is used to meet short-term demand within 1 to 2 days and is mainly suitable for daily inventory management scenarios, such as quickly processing retail orders or emergency needs. In the system, dynamic inventory dynamically adjusts inventory by analyzing the resource demand and fluctuation range in the scenario forecast data in real time. For example, during peak order periods, dynamic inventory can prioritize the allocation of resources from buffer inventory to quickly meet demand. Its core feature is fast response speed, but relatively small inventory, which is suitable for frequent small batch allocation.

[0060] Buffer stock is used to meet fluctuating demand within 3 to 7 days, mainly to solve the problem of short-term supply and demand imbalance. The design principle of buffer stock is to provide inventory flexibility. By maintaining an appropriate amount of safety inventory, it can effectively respond to sudden demand growth or supply chain disruptions. For example, during a promotion, the system will allocate some resources from strategic inventory to buffer inventory based on the demand fluctuation range in the scenario forecast data to alleviate the pressure of dynamic inventory. Buffer inventory ensures the continuity and stability of the system by balancing supply and demand fluctuations.

[0061] Strategic inventory is used to support long-term demand planning within 30 days, mainly for seasonal demand or long-term supply chain layout. The core function of strategic inventory is to ensure the stability of the supply chain in the long-term demand changes by reserving resources in advance. For example, before the holidays, the system will allocate resources to strategic inventory based on historical demand data forecasts to cope with possible demand peaks. Strategic inventory has a large capacity and a low allocation frequency, but it is of great significance to the long-term stability of the system.

[0062] The three layers of inventory are dynamically connected through resource allocation paths to form a resource flow network. The implementation of this network relies on the real-time input of scenario prediction data and the support of multi-agent collaborative optimization algorithms. The resource allocation path is optimized through the scenario nodes and scenario edges defined by the dynamic scenario graph. Specifically, when the resources of the dynamic inventory layer are insufficient to meet short-term demand, the system will automatically allocate resources from the buffer inventory layer through the resource allocation path and update the status of the resource flow path. For example, if the current inventory of a dynamic inventory node is 100 units and the demand is 150 units, the system will allocate 50 units of resources from the buffer inventory node through the resource allocation path to ensure that the demand of the dynamic inventory node is met.

[0063] The formation of a resource flow network enables flexible collaboration between inventory layers, and the system can dynamically adjust the resource volume of each inventory layer according to actual demand. For example, during a promotion, when demand fluctuates greatly, buffer inventory will take on the main allocation task, while strategic inventory will provide stable resource support for buffer inventory through the allocation path, ultimately ensuring the rapid response capability of dynamic inventory.

[0064] Through the design of a hierarchical inventory structure and its resource flow network, the present invention achieves refined management of short-term, medium-term and long-term inventory needs, effectively improving inventory turnover efficiency and supply chain stability. Dynamic inventory meets the need for rapid response, buffer inventory provides supply and demand balancing capabilities, and strategic inventory guarantees long-term stability, providing a highly flexible and intelligent solution for commodity inventory management. In addition, the dynamic adjustment of resource allocation paths ensures efficient collaboration between various layers of inventory, greatly reducing delays and resource waste in the inventory allocation process, and providing important technical support for commodity inventory management based on cloud computing.

[0065] Preferably, the dynamic allocation of resources among the dynamic inventory, the buffer inventory and the strategic inventory is achieved through a multi-agent collaborative optimization algorithm. Each inventory layer acts as an independent agent node, and when executing resource allocation, the allocation demand is calculated by sharing the current inventory status and resource demand information; the resource allocation path and resource allocation strategy are adjusted according to the feedback mechanism between agents, and inventory optimization data is generated according to scenario feedback.

[0066] In the present invention, the dynamic allocation of resources between dynamic inventory, buffer inventory and strategic inventory is realized through a multi-agent collaborative optimization algorithm, which is an important means of improving inventory management efficiency and responsiveness in the present invention. The core principle of the multi-agent collaborative optimization algorithm is to model each inventory layer as an independent agent node, each node makes autonomous decisions based on its current inventory status, resource requirements and the global goals of the system, and realizes overall optimization through interaction and feedback between agents.

[0067] In the system, dynamic inventory, buffer inventory and strategic inventory are respectively used as intelligent nodes, and each node has perception, decision-making and execution capabilities. The perception capability is reflected in the ability of the intelligent node to obtain its current inventory status and resource demand information in real time, such as inventory, demand and demand urgency; the decision-making capability is reflected in the ability of the intelligent node to calculate the amount of resources required to be allocated based on scenario prediction data and feedback mechanism; the execution capability refers to the ability of the intelligent node to interact with other nodes through resource allocation paths. Intelligent nodes exchange information through a shared communication network to ensure that each node works collaboratively on a global scale.

[0068] The process of dynamic resource allocation is based on the feedback mechanism between multiple agents. During the allocation process, the dynamic inventory node initiates a resource allocation request to the buffer inventory or strategic inventory node based on real-time demand. For example, when the dynamic inventory node detects that its current inventory is insufficient to meet short-term demand (for example, the inventory is 100 units and the demand is 150 units), the node will request an additional 50 units of resources from the buffer inventory node. The buffer inventory node decides whether to meet the request by calculating its own inventory status and demand pressure, and then feeds back the allocation result. If the buffer inventory cannot fully meet the request (such as only 30 units of resources can be provided), the system will further initiate a request to the strategic inventory node to meet the remaining 20 units of resource demand.

[0069] The agent collaborative optimization algorithm achieves resource allocation optimization through the following steps: First, each agent node calculates its resource demand priority, which is based on a comprehensive evaluation of inventory levels, demand urgency, and scenario prediction data. For example, the demand urgency of dynamic inventory nodes is usually higher than that of buffer inventory nodes, and the demand of buffer inventory nodes is higher than that of strategic inventory nodes. Secondly, the agent node dynamically adjusts the resource allocation path according to the priority. For example, if the cost of the current path is high (such as long transportation time or high energy consumption), the system will choose a better path for resource allocation. Finally, each node updates its inventory status according to the allocation results, and feeds the data back to the system center to generate inventory optimization data.

[0070] Inventory optimization data includes information such as current inventory status, resource allocation path, and allocation efficiency. For example, in one allocation, the initial inventory of the dynamic inventory node is 100 units, and the demand is 150 units. After the buffer inventory node allocates 30 units of resources, the strategic inventory node allocates the remaining 20 units of resources, and finally completes the resource allocation. The inventory optimization data generated by the system is recorded as follows: Inventory optimization data = {dynamic inventory status: 150, buffer inventory status: 470, strategic inventory status: 950, allocation path: buffer inventory → strategic inventory, allocation efficiency: 95%} Through this dynamic allocation mechanism, the multi-agent collaborative optimization algorithm can significantly improve the system's ability to respond to changes in inventory demand. For example, in promotional activities with large demand fluctuations, the dynamic inventory node can quickly obtain resource support from the buffer inventory node, and when the buffer inventory is insufficient, it can replenish resources from the strategic inventory in a timely manner, thus avoiding inventory shortages or over-reliance on a single inventory layer.

[0071] Based on inventory optimization data, inventory resource allocation tasks are executed, and the execution process of inventory resource allocation tasks is recorded to generate transparent inventory status data.

[0072] Preferably, during the execution of the inventory resource allocation task, blockchain technology is used to record the execution process of the inventory resource allocation task, and the record content includes the source inventory layer, target inventory layer, allocation quantity, execution time and task completion status of the resource allocation; the blockchain record is stored in the form of transaction blocks, and each transaction block includes a unique identifier and specific data of task execution.

[0073] In the present invention, the use of blockchain technology to record the execution process of inventory resource allocation tasks is an important means to ensure the transparency and immutability of allocation data and improve traceability efficiency. The core characteristics of blockchain technology include distributed storage, data immutability and transaction transparency. In the present invention, these characteristics are used to achieve comprehensive recording and auditing of the inventory resource allocation process.

[0074] During the execution of inventory resource allocation tasks, each resource allocation will generate a complete task record, which includes the source inventory layer, target inventory layer, allocation quantity, execution time, and task completion status of the resource allocation. These record data are packaged into transaction blocks and stored in the blockchain network. Each transaction block has a unique identifier, which is generated by a hash algorithm to ensure the uniqueness and integrity of the record. At the same time, each transaction block also includes specific data on task execution, such as the start time, end time, execution node, and operator information of the allocation task, to ensure the transparency and traceability of the allocation process.

[0075] The specific implementation process of blockchain records is as follows: When the resource allocation task is started, the system will automatically generate the basic information of the allocation task, such as the task ID, the identifiers of the source inventory layer and the target inventory layer, and the number of resources planned to be allocated. After the allocation task is completed, the system will integrate the execution result information (including the actual allocation quantity, allocation path, actual execution time and task completion status) with the basic information to form a complete task record. The record is packaged into a transaction block and linked to the previous block through a hash value to form a chain structure of the blockchain. For example, in an inventory allocation task, the dynamic inventory layer allocates 100 units of resources to the buffer inventory layer. The task ID is T001, the start time is "2024-12-05T10:00:00Z", the completion time is "2024-12-05T10:30:00Z", and the actual completion status is "successful". The corresponding transaction block content is as follows: { "Task ID": "T001", "Source Inventory Layer": "Dynamic Inventory Layer", "Target Stock Level": "Buffer Stock Level", "Allocation Quantity": 100, "Start Time": "2024-12-05T10:00:00Z", "Completion Time": "2024-12-05T10:30:00Z", "Task Status": "Success", "Previous block hash value": "abc123...", "Current block hash value": "def456..."} The effect of using blockchain technology to record this data is significant. First, the immutability of blockchain ensures the authenticity of all allocation records, avoiding the problem of human tampering or data loss. Second, through distributed storage, each blockchain node saves a complete copy of the data, and even if some nodes fail, the recorded data can still be fully restored. Third, the transparency of blockchain records allows the detailed process of each inventory allocation task to be audited in real time, and the relevant parties can query the execution details and results of specific tasks at any time. For example, when an inventory allocation dispute occurs, the system can provide conclusive evidence by querying blockchain records, thereby quickly resolving the problem.

[0076] From an application perspective, blockchain records can effectively improve the management efficiency and trust of inventory allocation. In supply chain cooperation scenarios, different partners can access the blockchain to view allocation records, reducing reliance on traditional reconciliation processes. For example, in an inventory system managed by multiple parties, suppliers, distributors, and retailers can all obtain the status of allocation tasks in real time, avoiding delays or misunderstandings caused by information asymmetry. At the same time, the transparency of blockchain records can improve the execution efficiency of allocation tasks, because all parties can clearly understand the execution schedule and resource flow of the task.

[0077] Preferably, the transparent inventory status data includes the current inventory status of the dynamic inventory, the buffer inventory and the strategic inventory, the execution result of the inventory resource allocation task and the real-time feedback information of the task completion rate.

[0078] In the present invention, transparent inventory status data is the core link to achieve real-time monitoring and efficient decision-making. The generation and application of transparent inventory status data relies on the system's comprehensive perception of dynamic inventory, buffer inventory and strategic inventory, as well as real-time feedback and result recording of inventory resource allocation tasks. The comprehensiveness and real-time nature of these data ensure the intelligent and precise inventory management.

[0079] The content of transparent inventory status data includes three key parts: the current inventory status of dynamic inventory, buffer inventory and strategic inventory, the execution results of inventory resource allocation tasks, and real-time feedback information of task completion rate. The current inventory status of dynamic inventory mainly reflects the quantity of available resources in the short term. For example, if the current inventory of the dynamic inventory layer is 200 units and the real-time demand is 180 units, the system records the inventory balance as 20 units through transparent inventory status data and updates it dynamically. If the demand fluctuates greatly and the inventory is insufficient, the system guides the allocation of resources from buffer inventory or strategic inventory to dynamic inventory through real-time feedback to meet short-term demand.

[0080] The current inventory status of the buffer stock is used to reflect the elasticity of medium-term resource reserves. For example, the inventory status data of the buffer stock layer includes the current inventory volume (such as 500 units), the number of deployments (such as 3 deployments in this cycle), and the deployment direction (such as mainly to dynamic inventory). Through these data, the system can evaluate whether the buffer stock layer needs to replenish resources or adjust the deployment frequency. The current inventory status of the strategic inventory mainly serves long-term planning. For example, the inventory status data of the strategic inventory records the expected demand within 30 days (such as 2,000 units) and the current inventory satisfaction rate (such as 80%). These data provide support for the long-term stability of the supply chain.

[0081] The second part of the transparent inventory status data is the execution result of the inventory resource allocation task, which includes the source inventory layer, target inventory layer, allocation quantity and allocation completion status of the allocation task. For example, in an allocation task, the dynamic inventory layer requests 50 units of resources from the buffer inventory layer, and actually completes the allocation of 40 units. The execution result is recorded as follows: Task ID: T002, Source stock layer: buffer stock, Target inventory layer: dynamic inventory, Requested quantity: 50 units, Actual quantity deployed: 40 units, Completion Status: Partially completed.

[0082] The execution results are transmitted to the management system and relevant parties through transparent inventory status data, ensuring that all inventory allocation behaviors are clearly visible.

[0083] The third part of the transparent inventory status data is the real-time feedback information of the task completion rate, which is calculated by comparing the task request quantity and the actual allocation quantity. For example, if the requested allocation quantity of the above task is 50 units and the actual allocation quantity is 40 units, the task completion rate is 80%. Real-time feedback information also includes the execution time and delay of the allocation task. For example, if the estimated time of the allocation task is 10 minutes and the actual time is 12 minutes, the system will record the delay time as 2 minutes and trigger path optimization or task priority adjustment through the feedback mechanism.

[0084] Transparent inventory status data is processed and displayed in real time through a cloud data platform, using visualization tools such as charts, digital dashboards, and geographic location maps to provide managers with a clear view of inventory status and allocation. For example, on the dashboard of a warehouse center, managers can simultaneously see the current status of dynamic inventory (inventory volume, demand volume, allocation frequency, etc.), the replenishment plan for buffer inventory, and the long-term inventory satisfaction rate of strategic inventory. Transparent inventory status data can also provide input for system automated decision-making, such as task priority adjustment triggered by completion rate below a certain threshold or allocation path optimization triggered by dynamic inventory shortage.

[0085] like Figure 5 As shown, a system for implementing the cloud computing-based commodity inventory management method described in the claims comprises: The multimodal data acquisition module is used to collect commodity dynamic information, environmental status data and commodity status data. Among them, commodity dynamic information is collected through RFID tags, environmental status data is collected through temperature and humidity sensors, and commodity status data is collected through pressure sensors and high-definition cameras; the multimodal data acquisition module obtains commodity dynamic information, environmental status data and commodity status data through RFID tags, temperature and humidity sensors, pressure sensors and high-definition cameras, providing comprehensive real-time data support for the system.

[0086] The data preprocessing module is communicatively connected with the multimodal data acquisition module, and is used to clean the collected data, remove redundancy and screen outliers, standardize the format of the screened data, and use the biometric learning model to extract environmental state feature data and commodity state feature data, and generate enhanced data in combination with the dynamic weight optimization algorithm; the data is cleaned, screened and standardized by the data preprocessing module, and the key feature data of the environment and commodities are extracted in combination with the biometric learning model, and the dynamic weight optimization algorithm is used to generate enhanced data, which lays a high-quality data foundation for subsequent analysis.

[0087] The dynamic scene graph construction and deduction module is connected to the data preprocessing module in communication, and is used to construct a dynamic scene graph based on the enhanced data, wherein the dynamic scene graph is composed of scene nodes and scene edges, the scene nodes represent standard scenes, extreme scenes and abnormal scenes, and the scene edges represent the resource flow path and its adjustment cost between the scene nodes; the dynamic scene graph construction and deduction module further uses the distributed simulation system to simulate the resource flow of the scene nodes in the dynamic scene graph, and adjusts the scene weight distribution and scene edge weight in the dynamic scene graph in combination with the quantum heuristic optimization algorithm to generate scene prediction data; in the dynamic scene graph construction and deduction module, the system constructs a dynamic scene graph based on the enhanced data, and defines the scene nodes and scene edges to represent the resource flow path and its adjustment cost. Combined with the distributed simulation system to simulate the resource flow, and adjust the scene weight and path weight through the quantum heuristic optimization algorithm, accurate scene prediction data is generated.

[0088] The hierarchical inventory optimization module is communicatively connected with the dynamic scenario graph construction and deduction module, and is used to construct a hierarchical inventory structure based on the scenario prediction data, wherein the hierarchical inventory structure includes dynamic inventory, buffer inventory and strategic inventory; realizes dynamic allocation of resources among dynamic inventory, buffer inventory and strategic inventory through a multi-agent collaborative optimization algorithm, and generates inventory optimization data; the hierarchical inventory optimization module constructs a hierarchical structure of dynamic inventory, buffer inventory and strategic inventory based on the scenario prediction data, realizes dynamic allocation of resources between inventory layers through a multi-agent collaborative optimization algorithm, and generates inventory optimization data, thereby achieving efficient management of short-term, medium-term and long-term demand.

[0089] The inventory resource execution module is in communication with the hierarchical inventory optimization module, and is used to execute the inventory resource allocation task based on the inventory optimization data, and use blockchain technology to record the execution process of the inventory resource allocation task, and generate transparent inventory status data including the current inventory status and task execution results. The inventory resource execution module executes the resource allocation task based on the optimization data, and uses blockchain technology to record the entire process, and generates transparent inventory status data to ensure the traceability and management transparency of the allocation process.

[0090] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems or computer program products. Therefore, the present application may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware.

[0091] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.

Claims

1. A commodity inventory management method based on cloud computing, characterized in that: The following steps are involved: Collect product dynamic information, environmental status data and product status data, pre-process the collected data, and generate enhanced data; Based on the enhanced data, a dynamic scene graph is constructed, the scene nodes in the dynamic scene graph are deduced using a distributed simulation system, and the scene prediction data is generated in combination with a quantum heuristic optimization algorithm; Based on the scenario prediction data, a hierarchical inventory structure is constructed, which includes dynamic inventory, buffer inventory and strategic inventory. The dynamic allocation of resources among dynamic inventory, buffer inventory and strategic inventory is realized through a multi-agent collaborative optimization algorithm, and inventory optimization data is generated; Based on inventory optimization data, inventory resource allocation tasks are executed, and the execution process of inventory resource allocation tasks is recorded to generate transparent inventory status data.

2. The commodity inventory management method based on cloud computing according to claim 1, characterized in that: The dynamic information of the commodity is collected through an RFID tag, which records the unique identification, current location and inventory change information of the commodity; the environmental status data is collected through a temperature and humidity sensor, including the temperature and humidity of the storage environment; the commodity status data is collected through a pressure sensor and a high-definition camera, the pressure sensor is used to detect the pressure state of the commodity surface, and the high-definition camera is used to obtain the appearance image data of the commodity.

3. The commodity inventory management method based on cloud computing according to claim 1, characterized in that: The commodity dynamic information, environmental status data and commodity status data are cleaned, redundant data are removed and outlier screening is performed. The outlier screening includes marking the temperature values ​​in the environmental status data that are lower than -10 degrees Celsius or higher than 50 degrees Celsius as invalid data; the screened data are formatted in a standardized manner, and the environmental status feature data and commodity status feature data are extracted based on the biometric learning model. The environmental status feature data and the commodity status feature data are prioritized in combination with a dynamic weight optimization algorithm to generate enhanced data.

4. The commodity inventory management method based on cloud computing according to claim 1, characterized in that: The dynamic scene graph is composed of scene nodes and scene edges, wherein the scene nodes represent standard scenes, extreme scenes and abnormal scenes, and the scene edges represent the resource flow paths between scene nodes and their adjustment costs, and the adjustment costs are calculated according to the following formula: ; in, Represents the node from the scene To scene node resource adjustment costs; Represents a scene node The weight of the scene node Priority; Indicates that the resource is from the scene node Adjust to scene node the time required; Indicates that the resource is from the scene node Adjust to scene node consumption.

5. The commodity inventory management method based on cloud computing according to claim 4, characterized in that: A distributed simulation system is used to simulate the resource flow of each scene node in the dynamic scene graph to generate initial resource allocation data for the scene nodes. A quantum heuristic optimization algorithm is used to dynamically adjust the scene weight distribution in the dynamic scene graph, and the scene edge weights of the dynamic scene graph are recalculated to generate scene prediction data, wherein the scene prediction data includes the optimization weight, resource demand and resource fluctuation range of each scene node.

6. The commodity inventory management method based on cloud computing according to claim 1, characterized in that: The hierarchical inventory structure includes dynamic inventory, buffer inventory and strategic inventory, wherein the dynamic inventory is used to meet short-term demand within 1 to 2 days, the buffer inventory is used to meet fluctuating demand within 3 to 7 days, and the strategic inventory is used to support long-term demand planning within 30 days; the dynamic inventory, buffer inventory and strategic inventory are dynamically connected through a resource allocation path to form a resource flow network.

7. The commodity inventory management method based on cloud computing according to claim 6, characterized in that: Dynamic allocation of resources among the dynamic inventory, the buffer inventory and the strategic inventory is achieved through a multi-agent collaborative optimization algorithm. Each inventory layer acts as an independent agent node. When executing resource allocation, the allocation demand is calculated by sharing the current inventory status and resource demand information. The resource allocation path and resource allocation strategy are adjusted according to the feedback mechanism between agents, and inventory optimization data is generated according to scenario feedback.

8. The commodity inventory management method based on cloud computing according to claim 1, characterized in that: During the execution of the inventory resource allocation task, the blockchain technology is used to record the execution process of the inventory resource allocation task, and the record content includes the source inventory layer, target inventory layer, allocation quantity, execution time and task completion status of the resource allocation; The blockchain records are stored in the form of transaction blocks, each of which includes a unique identifier and specific data for task execution.

9. The commodity inventory management method based on cloud computing according to claim 8, characterized in that: The transparent inventory status data includes the current inventory status of the dynamic inventory, the buffer inventory and the strategic inventory, the execution result of the inventory resource allocation task and the real-time feedback information of the task completion rate.

10. A system for implementing the commodity inventory management method based on cloud computing according to any one of claims 1 to 9, characterized in that: include: Multimodal data collection module, used to collect commodity dynamic information, environmental status data and commodity status data, where commodity dynamic information is collected through RFID tags, environmental status data is collected through temperature and humidity sensors, and commodity status data is collected through pressure sensors and high-definition cameras; A data preprocessing module, which is in communication with the multimodal data acquisition module, is used to clean the collected data, remove redundancy and filter outliers, standardize the format of the filtered data, extract environmental state feature data and commodity state feature data using a biometric learning model, and generate enhanced data in combination with a dynamic weight optimization algorithm; A dynamic scene graph construction and deduction module is communicatively connected to the data preprocessing module, and is used to construct a dynamic scene graph based on the enhanced data, wherein the dynamic scene graph is composed of scene nodes and scene edges, the scene nodes represent standard scenes, extreme scenes and abnormal scenes, and the scene edges represent resource flow paths and adjustment costs between scene nodes; the dynamic scene graph construction and deduction module further uses a distributed simulation system to simulate resource flow of scene nodes in the dynamic scene graph, and adjusts the scene weight distribution and scene edge weight in the dynamic scene graph in combination with a quantum heuristic optimization algorithm to generate scene prediction data; A hierarchical inventory optimization module is communicatively connected to the dynamic scenario graph construction and deduction module, and is used to construct a hierarchical inventory structure based on the scenario prediction data, wherein the hierarchical inventory structure includes dynamic inventory, buffer inventory and strategic inventory; and realizes dynamic allocation of resources among dynamic inventory, buffer inventory and strategic inventory through a multi-agent collaborative optimization algorithm to generate inventory optimization data; The inventory resource execution module is communicatively connected with the hierarchical inventory optimization module, and is used to execute the inventory resource allocation task based on the inventory optimization data, and use blockchain technology to record the execution process of the inventory resource allocation task, and generate transparent inventory status data including the current inventory status and task execution results.

Images