A warehouse management method and system based on multi-dimensional data fusion

Abstract

The application discloses a kind of warehouse management method and system based on multidimensional data fusion, it is related to warehouse management technical field, including data acquisition module, dynamic data fusion module, distributed task scheduling module, adaptive weight distribution module, parameter optimization module and real-time decision module.Integrating multi-source data (such as image, sensor, RFID etc.), using hierarchical data fusion strategy and distributed computing architecture, combining entropy method dynamically adjusting weight, and using improved particle swarm algorithm to optimize system parameters, efficient data processing and accurate real-time decision are realized.The application can significantly improve the intelligent level and operation efficiency of warehouse management system, meet the needs of modern warehouse for efficient and accurate management.

Description

Technical Field

[0001] This invention relates to the fields of warehouse management and information technology, and in particular to a warehouse management method and system based on multi-dimensional data fusion. Background Technology

[0002] Warehouse management systems based on multidimensional data fusion can comprehensively utilize various data sources (such as images, sensors, RFID, etc.) to achieve intelligent and efficient warehouse management. However, existing warehouse management systems have certain limitations in multidimensional data fusion, real-time assurance, and system integration, which affect overall management efficiency and accuracy.

[0003] In existing technologies, a warehouse management system (publication number CN108197882B) collects image information of a designated area in a warehouse through an image acquisition module and an image recognition module, and uses a business recognition module to analyze the movement patterns of inventory, thereby improving warehouse management efficiency and reducing labor costs. However, this solution mainly relies on the acquisition and processing of image information and lacks comprehensive analysis of other types of data such as temperature, humidity, and dynamic changes in inventory, which limits the system's adaptability and decision-making capabilities in complex scenarios. Furthermore, the accuracy and real-time performance of the image recognition module may be affected by factors such as ambient light and occlusion, thus impacting system performance.

[0004] In the prior art, a warehouse management system and method disclosed in CN113034068B obtains the storage status of items in a first receiving section in real time through a first sensor and updates the binding table in real time based on the storage status, facilitating retrieval and handling and improving warehouse management efficiency. However, this solution focuses on the dynamic management of item storage location and fails to fully consider the integrated analysis of multi-dimensional data such as inventory status, environmental parameters, and logistics information, which may lead to insufficient flexibility of the system under diverse warehousing needs. In addition, the binding table update mechanism may experience delays or data conflicts in large-scale warehousing scenarios, which may have a certain impact on the real-time performance and reliability of the system.

[0005] The aforementioned problems indicate that existing warehouse management systems still require further optimization in areas such as multi-dimensional data fusion, real-time performance assurance, adaptability to complex scenarios, and system scalability. Therefore, this invention provides a warehouse management method and system based on multi-dimensional data fusion, aiming to integrate multiple data sources (such as images, sensors, RFID, logistics information, etc.) and combine them with intelligent algorithms to achieve efficient data fusion and real-time decision-making, optimize warehouse management processes, improve the system's intelligence level and operational efficiency, and meet the demands of modern warehousing for efficient and precise management. Summary of the Invention

[0006] This invention provides a warehouse management method and system based on multi-dimensional data fusion. Addressing the shortcomings of existing warehouse management systems in multi-dimensional data fusion, real-time performance assurance, and system integration, it proposes a novel technical solution. Existing systems primarily rely on single data sources (such as images or sensor information), failing to fully integrate multiple data types, resulting in limited adaptability and decision-making capabilities in complex scenarios. Furthermore, the system may experience latency or data conflicts in large-scale warehousing scenarios, affecting real-time performance and reliability. To address these issues, this solution constructs a dynamic data fusion module, introduces a distributed computing architecture, designs an adaptive weight allocation mechanism, optimizes the data processing flow, and combines an improved particle swarm optimization algorithm to achieve parameter optimization, thereby improving the system's intelligence level and operational efficiency.

[0007] This invention provides a warehouse management system based on multi-dimensional data fusion, comprising a data acquisition module, a dynamic data fusion module, a distributed task scheduling module, an adaptive weight allocation module, a parameter optimization module, and a real-time decision-making module. The dynamic data fusion module integrates multiple data sources (such as images, sensors, RFID, and logistics information), designs a hierarchical data fusion strategy, normalizes data from different sources, and establishes correlations based on timestamps and spatial locations. The distributed task scheduling module adopts a distributed computing architecture, decomposing data processing tasks into multiple sub-tasks and allocating them to different computing nodes to ensure processing efficiency in high-concurrency scenarios. The adaptive weight allocation module dynamically adjusts the weights of each data source based on its reliability and real-time performance, designing a weight calculation formula based on entropy to ensure the accuracy of the data fusion results. The parameter optimization module optimizes system parameters using an improved particle swarm optimization algorithm, designs diverse evaluation indicators, and dynamically adjusts inertia and convergence factors to avoid getting trapped in local optima. The real-time decision-making module generates warehouse management instructions based on the fused data, supporting functions such as dynamic inventory adjustment, environmental monitoring, and logistics scheduling.

[0008] This invention provides a warehouse management method based on multi-dimensional data fusion, which includes the following steps: Step S1: Data acquisition; Step S2: Dynamic data fusion; Step S3: Distributed task scheduling; Step S4: Adaptive weight allocation; Step S5: Parameter optimization; Step S6: Real-time decision making.

[0009] In step S1, the data acquisition involves collecting information from multiple data sources within the warehousing environment, including image data, sensor data, RFID data, and logistics information. The image data includes the appearance characteristics and placement of items within the warehouse; the sensor data includes temperature, humidity, light intensity, and air quality parameters; the RFID data includes the unique identifier and location information of each item; and the logistics information includes the entry time, exit time, and transportation route. All collected data is accompanied by timestamps and spatial location information for subsequent data fusion and analysis.

[0010] In step S2, the dynamic data fusion specifically includes the following steps: Step S21: Data normalization processing, normalizing the collected data of different types so that the data values ​​fall within the range of [0,1]; Step S22: Data association, associating different types of data based on timestamps and spatial location information to form a multi-dimensional data matrix; Step S23: Hierarchical data fusion, designing a hierarchical data fusion strategy, first performing preliminary fusion on data of the same type, and then performing secondary fusion on the fusion results of different types to finally generate a comprehensive dataset.

[0011] In step S3, the distributed task scheduling specifically includes the following steps: Step S31: Task decomposition, decomposing the data processing task into multiple sub-tasks, each sub-task corresponding to a data processing unit; Step S32: Task allocation, allocating the sub-tasks to different computing nodes based on the load of the computing nodes and network bandwidth; Step S33: Task execution, each computing node executes the allocated tasks in parallel and uploads the processing results to the central node for aggregation.

[0012] In step S4, the adaptive weight allocation specifically includes the following steps: Step S41: Weight initialization, setting the initial weights of each data source to equal values; Step S42: Weight calculation, calculating the weights of each data source based on the entropy method, using the following formula:

[0013]

[0014] In the formula, w i H is the weight of the i-th data source. i is the entropy value of the i-th data source, and n is the number of data sources; Step S43: Weight update, dynamically adjust the weights according to the real-time performance and reliability of the data sources to ensure the accuracy of the data fusion results.

[0015] In step S5, the parameter optimization specifically includes the following steps: Step S51: Initialize the particle swarm, using system parameters as particle positions, and initialize particle positions based on the search space; Step S52: Diversity assessment, calculating the diversity of the particle swarm using the following formula:

[0016]

[0017] In the formula, D(r) is the diversity of the particle swarm at the r-th iteration, q is the number of particles, m is the parameter dimension, and x... ij (r) is the coordinate of the i-th particle in the j-th dimension. It is the average coordinate of the j-th dimension; Step S53: Adjustment of inertia factor and convergence factor. The inertia factor and convergence factor are dynamically adjusted based on the diversity assessment results. The formula used is as follows:

[0018] ω(r)=ω min +(ω max -c min )·e -λD(r)

[0019] c1(r)=c 1min +(c 1max -c 1min )ω(1-e -μD(r) )

[0020] c2(r)=c 2min +(c 2max -c 2min )·(1-e -vD(r) )

[0021] In the formula, ω(r) is the inertia factor at the r-th iteration, c1(r) and c2(r) are the personal cognitive factor and the social cognitive factor, respectively, and λ, μ, and v are control parameters; Step S54: Position update, adjust the particle position based on the updated inertia factor and convergence factor, using the following formula:

[0022] x i (r+1)=x i (r)+ω(r)·v i (r)+c1(r)·rand()·(p i -x i (r))+c2(r)·rand()·(g

[0023] -x i (r))

[0024] In the formula, x i (r+1) is the position of the i-th particle in the (r+1)-th iteration, v i (r) is the velocity of the particle, p i is the historical best position of the particle, and g is the global best position.

[0025] In step S6, the real-time decision-making is based on the fused data to generate warehouse management instructions, including dynamic inventory adjustment, environmental monitoring, and logistics scheduling. Specifically, when the inventory level in a certain area is detected to be lower than a preset threshold, a replenishment instruction is generated; when environmental parameters exceed the safe range, an alarm instruction is generated; and when logistics information shows abnormal transportation routes, a route optimization instruction is generated.

[0026] The beneficial effects of this invention are:

[0027] This invention enables efficient fusion and real-time decision-making of multi-dimensional data, significantly improving the intelligence level and operational efficiency of the warehouse management system, and meeting the needs of modern warehousing for efficient and precise management. Attached Figure Description

[0028] Figure 1 This is a schematic diagram of the overall structure of the present invention.

[0029] The attached figures are labeled as follows:

[0030] 1. Data acquisition module; 2. Dynamic data fusion module; 3. Distributed task scheduling module; 4. Adaptive weight allocation module; 5. Parameter optimization module; 6. Real-time decision-making module. Detailed Implementation

[0031] The present invention will be further described below with reference to the accompanying drawings and specific embodiments. The illustrative embodiments and descriptions herein are used to explain the present invention, but are not intended to limit the present invention.

[0032] This invention provides a warehouse management method and system based on multi-dimensional data fusion, the specific implementation of which is described in conjunction with the appendix. Figure 1 A detailed explanation of the overall structural diagram is provided below. (Attached) Figure 1 The document illustrates the system's six core modules and their interrelationships, including data acquisition module 1, dynamic data fusion module 2, distributed task scheduling module 3, adaptive weight allocation module 4, parameter optimization module 5, and real-time decision-making module 6. These modules are connected through data flow and control flow to form a complete warehouse management system framework.

[0033] Data acquisition module 1 is the foundation of the entire system, responsible for acquiring information from various data sources within the warehouse environment. These data sources include image data, sensor data, RFID data, and logistics information. Image data is collected by cameras installed in the warehouse to record the appearance and placement of items; sensor data is collected by temperature and humidity sensors, light intensity sensors, and air quality sensors distributed throughout the warehouse to monitor environmental parameters; RFID data is collected by RFID tags attached to items and readers to identify the unique identity and location information of each item; logistics information comes from the warehouse management system, including inbound time, outbound time, and transportation route. All collected data is accompanied by timestamps and spatial location information, which serve as crucial evidence for subsequent data fusion and analysis. Data acquisition module 1 transmits the collected data to dynamic data fusion module 2 via wired or wireless means.

[0034] The dynamic data fusion module 2 receives multi-source data from the data acquisition module 1 and performs normalization, data association, and hierarchical data fusion on it. First, in the data normalization stage, module 2 maps different types of data values ​​to the [0,1] interval to eliminate the impact of data dimension differences on subsequent processing. For example, the original range of temperature and humidity data may be 0 to 100, while the range of image feature data may be 0 to 255. A normalization formula unifies these data to the same numerical range. Second, in the data association stage, module 2 matches and associates data from different sources based on timestamps and spatial location information to form a multi-dimensional data matrix. For example, temperature and humidity data of a certain area at a certain moment, along with image data and RFID data of that area, are integrated into the same data matrix. Finally, in the hierarchical data fusion stage, module 2 designs a hierarchical data fusion strategy, first performing preliminary fusion on data of the same type, and then performing secondary fusion on the fusion results of different types, ultimately generating a comprehensive dataset. These comprehensive datasets are transmitted to the distributed task scheduling module 3 via data stream.

[0035] The distributed task scheduling module 3 adopts a distributed computing architecture, decomposing the comprehensive dataset generated by the dynamic data fusion module 2 into multiple sub-tasks and allocating them to different computing nodes for parallel processing. In the task decomposition phase, module 3 divides the data processing task into several independent sub-tasks based on the dataset size and computational complexity, with each sub-task corresponding to a data processing unit. For example, one sub-task might process temperature and humidity data for a specific area, while another sub-task might process image data for that area. In the task allocation phase, module 3 allocates sub-tasks to the optimal computing node based on the load and network bandwidth of each computing node using a load balancing algorithm. For example, when a computing node has a low load and high network bandwidth, tasks are preferentially allocated to that node. In the task execution phase, each computing node executes the allocated tasks in parallel and uploads the processing results to the central node for aggregation. The aggregated data is then transmitted to the adaptive weight allocation module 4 via a data stream.

[0036] The adaptive weight allocation module 4 dynamically adjusts the weights of each data source based on its reliability and real-time performance, ensuring the accuracy of the data fusion results. During the weight initialization phase, module 4 sets the initial weights of each data source to equal values; for example, the initial weights of image data, sensor data, RFID data, and logistics information are all 0.25. In the weight calculation phase, module 4 calculates the weights of each data source based on the entropy method, using the following formula:

[0037]

[0038] In the formula, w i H is the weight of the i-th data source. i Here, is the entropy value of the i-th data source, and n is the number of data sources. For example, when the entropy value of a data source is low, it indicates that it has a large amount of information and high reliability, so its weight will increase accordingly. During the weight update phase, module 4 dynamically adjusts the weights based on the real-time performance and reliability of the data sources. For example, when the latency of a data source is high or the data quality is poor, its weight will be reduced to minimize its impact on the fusion result. The adjusted weights are transmitted to the parameter optimization module 5 via data stream.

[0039] Parameter optimization module 5 optimizes system parameters using an improved particle swarm optimization algorithm to enhance system efficiency and decision-making accuracy. In the particle swarm initialization phase, module 5 uses system parameters as particle positions, initializing particle positions based on the search space. For example, system parameters might include inventory thresholds, environmental monitoring parameters, and logistics scheduling parameters, whose initial values ​​are randomly distributed within the search space. In the diversity assessment phase, module 5 calculates the diversity of the particle swarm using the following formula:

[0040]

[0041] In the formula, D(r) is the diversity of the particle swarm at the r-th iteration, q is the number of particles, m is the parameter dimension, and x... ij (r) is the coordinate of the i-th particle in the j-th dimension. This is the average coordinate of the j-th dimension. For example, when the diversity of the particle swarm is low, it indicates that the particle distribution is relatively concentrated, which may pose a risk of getting trapped in a local optimum. During the adjustment phase of the inertia factor and convergence factor, Module 5 dynamically adjusts the inertia factor and convergence factor based on the diversity assessment results, using the following formula:

[0042] ω(r)=ω min +(ω max -ω min )·e -λD(r)

[0043] c1(r)=c 1min +(c 1max -c 1min )·(1-e -μD(r) )

[0044] c2(r)=c 2min +(c 2max -c 2min )·(1-e -vD(r) )

[0045] In the formula, ω(r) is the inertia factor at the r-th iteration, c1(r) and c2(r) are the personal cognitive factor and the social cognitive factor, respectively, and λ, μ, and v are control parameters. For example, when the diversity of the particle swarm is high, the inertia factor and convergence factor will be adjusted to larger values ​​to accelerate the convergence speed. In the position update phase, module 5 adjusts the position of the particles based on the updated inertia factor and convergence factor, using the following formula: x i (r+1)=x i (r)+ω(r)·v i (r)+c1(r)·rand()·(p i -x i (r))+c2(r)·rand()·(g

[0046] -x i (r))

[0047] In the formula, x i (r+1) is the position of the i-th particle in the (r+1)-th iteration, v i (r) is the velocity of the particle, p i is the particle's historical optimal position, and g is the global optimal position. The optimized parameters are transmitted to the real-time decision module 6 via data stream.

[0048] The real-time decision-making module 6 generates warehouse management instructions based on the fused data, supporting functions such as dynamic inventory adjustment, environmental monitoring, and logistics scheduling. For example, when the inventory level in a certain area is detected to be lower than a preset threshold, module 6 generates a replenishment instruction and notifies relevant operators or automated equipment to perform replenishment operations through the control system. When environmental parameters exceed safe ranges, module 6 generates an alarm instruction and alerts relevant personnel to take measures through audible and visual alarm devices. When logistics information shows abnormal transportation routes, module 6 generates route optimization instructions and replans the transportation routes through the logistics scheduling system. These instructions are fed back to the data acquisition module 1 through control flow, forming a closed-loop control system, thereby achieving intelligent and efficient warehouse management.

[0049] In the above implementation process, the various modules work closely together through data flow and control flow to jointly complete the warehouse management task. For example, data acquisition module 1 transmits the collected multi-source data to dynamic data fusion module 2, which generates a comprehensive dataset through normalization, data association, and hierarchical data fusion, and then transmits it to distributed task scheduling module 3. Module 3 decomposes the data processing task into multiple sub-tasks and allocates them to different computing nodes for parallel processing. The processing results are transmitted to adaptive weight allocation module 4 through data flow. Module 4 dynamically adjusts the weights based on the reliability and real-time performance of the data sources and transmits the adjusted weights to parameter optimization module 5. Module 5 optimizes the system parameters using an improved particle swarm optimization algorithm and transmits the optimized parameters to real-time decision module 6. Module 6 generates warehouse management instructions based on the fused data and feeds them back to data acquisition module 1 through control flow, forming a complete closed-loop control system. This modular design not only improves the system's scalability and flexibility but also significantly enhances the intelligence level and operational efficiency of warehouse management.

[0050] To enable those skilled in the art to fully understand and implement this invention, the specific implementation principles of this invention are further explained below in conjunction with specific application scenarios.

[0051] In the actual operation of warehouse management, the data acquisition module 1 first completes the task of collecting multi-source data. Taking a large intelligent warehouse as an example, the warehouse is equipped with multiple high-definition cameras, temperature and humidity sensors, RFID readers, and a logistics management system interface. The cameras capture real-time image information of the shelf area to identify the placement status and appearance characteristics of the items; the temperature and humidity sensors are deployed in key locations in the warehouse to continuously monitor changes in environmental parameters; the RFID readers obtain the unique identification and location information of the items by scanning the tags attached to them; and the logistics management system provides dynamic information such as inbound time, outbound time, and transportation route. All collected data is accompanied by accurate timestamps and spatial location information and is transmitted to the dynamic data fusion module 2 via a wireless network.

[0052] After receiving the aforementioned multi-source data, the dynamic data fusion module 2 first performs normalization processing on it. For example, the image feature data collected by the camera ranges from 0 to 255, while the data from the temperature and humidity sensor ranges from 0 to 100. Module 2 uses a normalization formula to map these data uniformly into the [0,1] interval, thereby eliminating the impact of dimensional differences on subsequent processing. Subsequently, in the data association stage, module 2 matches data from different sources based on timestamps and spatial location information. For example, the temperature and humidity data of a certain shelf area at a certain moment is integrated with the image data and RFID data of that area into the same data matrix. Finally, in the hierarchical data fusion stage, module 2 first performs preliminary fusion on data of the same type (such as all temperature and humidity data), and then performs secondary fusion on the fusion results of different types of data to generate a comprehensive dataset. This process ensures the comprehensiveness and consistency of the data, providing a reliable foundation for subsequent analysis.

[0053] The distributed task scheduling module 3 receives the comprehensive dataset from the dynamic data fusion module 2 and decomposes it into multiple sub-tasks. For example, a task to monitor the temperature and humidity of a specific shelf area might be assigned to computing node A, while an image data analysis task for that area might be assigned to computing node B. Module 3 optimizes the task allocation strategy using a load balancing algorithm based on the load of each computing node and network bandwidth. For example, when computing node A has a low load and high network bandwidth, tasks are preferentially assigned to that node. Each computing node executes the assigned tasks in parallel and uploads the processing results to the central node for aggregation. This distributed architecture significantly improves the system's concurrent processing capabilities, making it particularly suitable for large-scale warehousing scenarios.

[0054] The adaptive weight allocation module 4 dynamically adjusts the weights of each data source based on the aggregated data. For example, if an RFID reader experiences high data latency due to signal interference, module 4 will reduce its weight to minimize its impact on the fusion result. During weight calculation, module 4 uses the entropy method to evaluate the information content of each data source. For instance, if a data source has a low entropy value, it indicates a large amount of information and high reliability, and therefore its weight will be increased accordingly. Through this dynamic adjustment mechanism, module 4 ensures the accuracy of the data fusion result while improving the system's adaptability in complex scenarios.

[0055] Parameter optimization module 5 optimizes system parameters using an improved particle swarm optimization algorithm. For example, for key parameters such as inventory thresholds, environmental monitoring parameters, and logistics scheduling parameters, module 5 initializes the positions of the particle swarm and randomly distributes initial values ​​of particles based on the search space. During the diversity assessment phase, module 5 calculates the distribution of the particle swarm. If the particle distribution is found to be too concentrated, there is a risk of getting trapped in local optima. To address this, module 5 dynamically adjusts the inertia factor and convergence factor to accelerate convergence or expand the search range. For example, when the diversity of the particle swarm is high, the inertia factor and convergence factor are adjusted to larger values, thereby improving the global search capability. Finally, module 5 outputs the optimized parameters to support real-time decision-making.

[0056] The real-time decision-making module 6 generates specific warehouse management instructions based on the fused data. For example, when the inventory level in a certain shelf area is detected to be lower than a preset threshold, module 6 generates a replenishment instruction and notifies the automated equipment to perform the replenishment operation through the control system. When the temperature and humidity sensors detect that environmental parameters exceed the safe range, module 6 generates an alarm instruction and alerts relevant personnel to take measures through an audible and visual alarm device. In addition, when logistics information shows abnormal transportation routes, module 6 generates route optimization instructions and replans the transportation routes through the logistics scheduling system. These instructions are fed back to the data acquisition module 1 through the control flow, forming a closed-loop control system, thereby realizing intelligent and efficient warehouse management.

[0057] Through the above steps, this invention achieves efficient fusion of multi-dimensional data and real-time decision-making. For example, in a specific operation, the system successfully identified insufficient inventory in a certain shelf area and promptly generated a replenishment order, preventing operational interruptions due to stockouts. Simultaneously, by monitoring environmental parameters in real time, the system promptly detected and addressed abnormal temperature and humidity issues, ensuring the safety of the storage environment. This modular design not only improves the system's scalability and flexibility but also significantly enhances the intelligence level and operational efficiency of warehouse management, meeting the demands of modern warehousing for efficient and precise management.

[0058] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A warehouse management system based on multi-dimensional data fusion, characterized in that: It includes a data acquisition module (1), a dynamic data fusion module (2), a distributed task scheduling module (3), an adaptive weight allocation module (4), a parameter optimization module (5), and a real-time decision-making module (6). The data acquisition module (1) is used to collect information from multiple data sources in the warehousing environment, including image data, sensor data, RFID data and logistics information; The dynamic data fusion module (2) is used to normalize, associate, and fusion the data source information collected by the data acquisition module (1) to generate a comprehensive dataset. The distributed task scheduling module (3) receives the comprehensive dataset from the dynamic data fusion module (2), decomposes it into multiple sub-tasks, and then allocates the sub-tasks to different computing nodes based on the load of the computing nodes and network bandwidth. Each computing node executes the allocated task in parallel and uploads the processing results to the central node for aggregation. The adaptive weight allocation module (4) dynamically adjusts the weights of each data source based on the aggregated data and the entropy method; The adaptive weight allocation module (4) first sets the initial weights of each data source to equal values; then it calculates the weights of each data source based on the entropy method, using the following formula: In the formula, It is the first The weight of each data source, It is the first The entropy value of each data source. It refers to the number of data sources; Finally, the weights are dynamically adjusted based on the real-time performance and reliability of the data source. The parameter optimization module (5) is used to optimize the system parameters through the improved particle swarm algorithm, design diversity evaluation indicators, dynamically adjust the inertia factor and convergence factor to output the optimized parameters, and provide support for real-time decision-making. The parameter optimization module (5) first uses the system parameters as the position of the particle and initializes the position of the particle based on the search space; Then, the diversity of the particle swarm is calculated using the following formula: In the formula, It is the first The diversity of the particle swarm at the next iteration It is the number of particles. It is a parameter dimension. It is the first The particle in the first dimensional coordinates, It is the first The average coordinates of the dimension; Next, the inertia factor and convergence factor are adjusted dynamically based on the diversity assessment results, using the following formula: In the formula, It is the first The inertia factor at the next iteration and These are personal cognitive factors and social cognitive factors. , and These are control parameters; Finally, the particle positions are adjusted based on the updated inertia and convergence factors, using the following formula: In the formula, It is the first The particle in the first Position at the next iteration It is the velocity of the particle. It is the best position in the history of the particle. It is the globally optimal position; The real-time decision module (6) is used to generate warehouse management instructions based on the optimized parameters and feed them back to the data acquisition module (1) through the control flow.

2. The warehouse management system based on multi-dimensional data fusion according to claim 1, characterized in that: The dynamic data fusion module (2) first maps the collected different types of data source information to the [0, 1] interval; then it matches the different types of data source information based on timestamps and spatial location information to form a multidimensional data matrix; Finally, after initial fusion of data of the same type, the fusion results of different types are fused a second time to generate a comprehensive dataset.

3. A warehouse management system based on multi-dimensional data fusion according to claim 1, characterized in that: The distributed task scheduling module (3) first divides the comprehensive dataset processing task into multiple independent subtasks, each subtask corresponding to a data processing unit; then, based on the load of the computing nodes and network bandwidth, it allocates the subtasks to the optimal computing nodes through a load balancing algorithm; finally, each computing node executes the allocated tasks in parallel and uploads the processing results to the central node for aggregation.

4. A management method for a warehouse management system based on multi-dimensional data fusion as described in any one of claims 1-3, characterized in that: Includes the following steps: S1: Data Acquisition: Collect information from multiple data sources in the warehouse environment, including image data, sensor data, RFID data, and logistics information; S2: Dynamic data fusion: Normalizing, associating, and fusion the collected data; S3: Distributed task scheduling: Decomposes data processing tasks into multiple subtasks and distributes them to different computing nodes; S4: Adaptive weight allocation: Dynamically adjusts the weight of each data source based on the reliability and real-time performance of the data source; S5: Parameter optimization: System parameters are optimized using an improved particle swarm optimization algorithm; S6: Real-time decision-making: Generate warehouse management instructions based on the fused data.

5. The management method of the warehouse management system based on multi-dimensional data fusion according to claim 4, characterized in that: In step S1, the image data includes the appearance features and placement status of the items in the warehouse, the sensor data includes temperature and humidity, light intensity and air quality parameters, the RFID data includes the unique identification code and location information of the items, and the logistics information includes the entry time, exit time and transportation route; all collected data are accompanied by timestamps and spatial location information.

6. The management method of the warehouse management system based on multi-dimensional data fusion according to claim 4, characterized in that: In step S6, the generated warehouse management instructions include inventory dynamic adjustment instructions, environmental monitoring instructions, and logistics scheduling instructions; when the inventory level in a certain area is detected to be lower than a preset threshold, a replenishment instruction is generated; when environmental parameters exceed the safe range, an alarm instruction is generated; when logistics information shows abnormal transportation routes, a route optimization instruction is generated.

Images