Intelligent warehouse logistics system based on Internet of Things

Abstract

The invention relates to an intelligent warehouse logistics system based on the Internet of Things, and the system comprises a sensing layer which is used for collecting and obtaining the data of each sub-warehouse of an intelligent warehouse; the data processing layer is used for performing cleaning, standardization processing and feature extraction on the data acquired by the sensing layer and outputting sub-bin feature data; the demand portrait layer is used for generating a sub-warehouse three-dimensional demand label based on the sub-warehouse feature data; the logistics system customization model layer comprises a classification model and a recommendation model; the classification model divides the sub-bins into preset sub-bin types based on the three-dimensional demand labels; the recommendation model outputs a customized logistics system configuration scheme based on the sub-warehouse type and the weight value of the three-dimensional demand label; and the execution layer is used for deploying the logistics system according to the customized logistics system configuration scheme and monitoring the operation data of the sub-bins in real time so as to dynamically optimize the customized logistics system configuration scheme. The method has the effect of effectively improving the logistics efficiency of the intelligent sub-warehouse.

Description

Technical Field

[0001] This application relates to the field of warehousing and logistics, and in particular to an intelligent warehousing and logistics system based on the Internet of Things. Background Technology

[0002] With the rapid development of IoT and AI technologies, intelligent warehousing and logistics systems have become the core infrastructure of modern supply chains. Traditional warehousing systems typically adopt standardized configuration models, improving basic operational efficiency through unified automated equipment (such as AGVs and automated storage and retrieval systems) and general management systems. However, in industries such as e-commerce, FMCG, and fresh food, to meet the needs of nationwide distribution, multi-category storage, and immediacy, a multi-level sub-warehouse architecture consisting of regional central warehouses, city warehouses, and forward warehouses is commonly used. In this large-scale distributed warehousing scenario, the functional requirements of the logistics system vary significantly among the various sub-warehouses due to differences in their roles (such as sorting, storage, cold chain, and reverse processing), cargo characteristics (such as fresh produce, standardized products, and hazardous materials), and geographical environments (such as high altitude, low temperature, and coastal areas).

[0003] At this point, the existing standardized and unified model of traditional intelligent warehousing systems is still adopted. This means that a common set of equipment configurations (such as AGVs, sorting machines, and shelves), management algorithms (such as path planning and inventory allocation), and system modules are used to cover all sub-warehouses. This approach is prone to problems such as equipment mismatch with demand, functional redundancy and deficiency, insufficient environmental adaptability, and lack of dynamic optimization mechanisms, due to the failure to consider the individual needs of sub-warehouses. As a result, each intelligent sub-warehouse is prone to low logistics efficiency.

[0004] In response to the aforementioned technical problems, there is an urgent need for an intelligent warehousing and logistics system based on the Internet of Things. Summary of the Invention

[0005] To address the aforementioned issues, this application provides an intelligent warehousing and logistics system based on the Internet of Things.

[0006] This application provides an intelligent warehousing and logistics system based on the Internet of Things, which adopts the following technical solution: An IoT-based intelligent warehousing and logistics system includes: The perception layer is used to collect basic attribute data, operational data, cargo characteristic data, and environmental parameter data of each sub-warehouse in the smart warehouse. The data processing layer, which is communicatively connected to the perception layer, is used to clean, standardize, and extract features from the data collected by the perception layer, and output sub-warehouse feature data. The demand profiling layer is connected to the data processing layer and is used to generate three-dimensional demand labels for sub-warehouses based on sub-warehouse feature data, including functional demand dimension, cargo demand dimension and environmental demand dimension. Each dimension includes at least one label item, and each label item is associated with a weight value. The logistics system customization model layer, which is communicatively connected to the demand profiling layer, includes a classification model and a recommendation model. The classification model divides sub-warehouses into preset sub-warehouse types based on the three-dimensional demand tags. The recommendation model outputs a customized logistics system configuration scheme based on the sub-warehouse types and the weight values ​​of the three-dimensional demand tags. The customized logistics system configuration scheme includes at least one combination of logistics equipment, at least one optimization algorithm module, and at least one system function module and installation, deployment, and adaptation scheme. The execution layer communicates with the customized model layer of the logistics system and is used to deploy the logistics system according to the customized logistics system configuration scheme, and monitor the operation data of the sub-warehouses in real time to dynamically optimize the customized logistics system configuration scheme.

[0007] Preferably, the sensing layer includes: The basic attribute collection unit is used to collect the geographical location information, surrounding traffic data, and building parameter information of the sub-warehouse, and package them to generate the basic attribute data of the sub-warehouse. The operation data acquisition unit is used to collect data on the average daily order volume, sorting frequency, storage cycle and peak inbound and outbound periods of the sub-warehouse, and package them to generate the sub-warehouse's responsibility operation data. The cargo feature acquisition unit, including an RFID reader and a visual recognition device, is used to collect cargo category, size, and temperature and humidity requirements, and to package and generate cargo feature data for the sub-warehouse. The environmental parameter acquisition unit includes temperature and humidity sensors and dust sensors, which are used to collect temperature, humidity and dust concentration, and package them to generate environmental parameter data for the sub-compartment.

[0008] Preferably, the step of cleaning, standardizing, and extracting features from the data collected by the perception layer to output sub-warehouse feature data specifically includes the following steps: A1. Data Cleaning: Clean the raw data collected by the perception layer, remove outliers and redundant data caused by repeated reporting by the device, and fill in any missing data. A2. Data Standardization: Normalize the cleaned data to obtain standardized data; A3. Data Feature Extraction: Extracting features from standardized data to obtain feature vectors for constructing demand profiles. The feature vectors include static features, dynamic features, and environmental features. A4. Data Feature Integration: The extracted static features, dynamic features, and environmental features are vectorized and concatenated to form sub-warehouse feature data that represents the full range of characteristics of the sub-warehouse, and then output to the demand profile layer.

[0009] Preferably, the step of extracting features from standardized data to obtain feature vectors for constructing demand profiles specifically includes the following steps: B1. Static Feature Extraction: Extract structured features of sub-warehouses from basic attribute data, such as the ratio of area to floor height and the traffic convenience index.

[0010] B2. Dynamic Feature Extraction: Extract time-series features from job operation data, including but not limited to: order volume fluctuation coefficient calculated based on order data from the previous logistics cycle, and sorting efficiency calculated based on sorting time.

[0011] B3. Environmental Feature Extraction: Extract statistical features from environmental parameter data, including but not limited to: the variance of temperature and humidity data, the daily average maximum value of dust concentration, and the cumulative duration of ultraviolet radiation exceeding the standard in the past week.

[0012] Preferably, the requirement profiling layer generates a three-dimensional requirement label for the sub-warehouse based on sub-warehouse feature data, which includes functional requirement dimensions, cargo requirement dimensions, and environmental requirement dimensions. This specifically includes the following steps: C1. Generating Functional Requirement Dimensions: Based on sub-warehouse feature data, sorting efficiency labels and weights, storage density labels and weights, and reverse processing labels and weights are generated for the functional requirement dimensions through a pre-set requirement profile model; the requirement profile model is a machine learning model obtained through iterative training of historical data. C2. Generate cargo demand dimensions: Based on the sub-warehouse feature data, generate cold chain demand labels and weights, explosion-proof demand labels and weights, and flexible processing labels and weights for cargo demand dimensions through a pre-set demand profile model; C3. Generation of environmental requirements dimension: Based on the sub-compartment feature data, the low temperature tolerance label and weight, salt spray protection label and weight, and low pressure adaptation label and weight are generated through a pre-set requirement profile model. C4. Normalization: Normalize the weight values ​​of each label item in the functional requirements dimension, the goods requirements dimension, and the environmental requirements dimension to ensure that the sum of the weight values ​​of each label item in each dimension is 1. C5. Combine and generate 3D demand labels: Combine the normalized weight values ​​with the corresponding label items to generate the 3D demand labels for the sub-warehouse, and output them to the logistics system customized model layer.

[0013] Preferably, the classification model is a multi-classification model based on gradient boosting tree, the recommendation model is a collaborative filtering recommendation model, and the preset sub-warehouse types include sorting type, storage type, cold chain type, and hazardous materials type.

[0014] Preferably, the classification model, based on the three-dimensional demand labels, divides sub-warehouses into preset sub-warehouse types, specifically including the following steps: D1. Feature vectorization: Combine the weight values ​​of each dimension in the three-dimensional demand label into a multi-dimensional numerical feature vector; D2. Type Probability Prediction: Input the multidimensional numerical feature vector into the pre-trained classification model to obtain the probability value of the sub-warehouse belonging to each preset sub-warehouse type; D3. Type determination: Select the sub-warehouse type with the highest probability value as the final classification result of the sub-warehouse; and when the highest probability value is lower than the preset confidence threshold, the sub-warehouse is marked as a pending type and sent to the management personnel for manual review.

[0015] Preferably, the recommendation model outputs a customized logistics system configuration scheme based on the sub-warehouse type and the weight values ​​of the three-dimensional demand labels, specifically including the following steps: E1. Initial screening of candidate solutions: The recommendation model generates candidate logistics system configuration solutions that are suitable for the sub-warehouse type based on the sub-warehouse type matching and screening. E2. Multi-dimensional matching calculation: The recommendation model calculates the comprehensive matching score of each candidate logistics system configuration scheme using a pre-set comprehensive matching score calculation formula; the comprehensive matching score calculation formula is as follows: ; Where S is the overall matching score, , , These are the importance coefficients for the functional requirements, cargo requirements, and environmental requirements dimensions pre-set for the corresponding sub-warehouse types. , and These represent the weight values ​​of the i-th functional requirement label, the j-th goods requirement label, and the k-th environmental requirement label in the sub-warehouse 3D requirement labels generated by the requirement profile layer, respectively. , and These represent the adaptability scores of each candidate logistics system configuration scheme for the i-th functional requirement, j-th cargo requirement, and k-th environmental requirement tag item, respectively; m, n, and p represent the total number of tag items in the three-dimensional requirement tags for the energy requirement dimension, cargo requirement dimension, and environmental requirement dimension, respectively.

[0016] E3. Optimal Solution Generation: Select the candidate logistics system configuration scheme with the highest comprehensive matching score as the recommended customized logistics system configuration scheme.

[0017] Preferably, the execution layer includes: The solution deployment unit is used to load the hardware and software modules of the customized logistics system configuration solution, and to complete the system integration and function activation. The dynamic monitoring unit is used to monitor the operation data of the sub-warehouse in real time. The operation data includes equipment operating status, order processing efficiency, energy consumption data and environmental parameters. An optimized triggering unit is used to trigger the logistics system customization model layer to recalculate and output a new customized logistics system configuration scheme when the deviation between the monitored pre-set key performance indicators and the expected targets continues to exceed a preset threshold.

[0018] Preferably, it also includes a central coordination layer, which connects to the execution layers of each sub-warehouse and is used for cross-sub-warehouse data synchronization and resource scheduling.

[0019] In summary, this application includes at least one of the following beneficial technical effects: 1. By covering all dimensions of data on sub-warehouse basic attributes, operations, goods, and environment through the perception layer, and setting three-dimensional demand labels after data processing, the personalized needs of each sub-warehouse in terms of function, goods, and environment are accurately quantified, achieving precise matching of sub-warehouse logistics needs and overcoming the drawbacks of standardized configuration. At the same time, through the linkage of classification and recommendation dual models, it can not only automatically identify the sub-warehouse type, but also output a complete customized solution covering hardware equipment, algorithms, and functional modules, realizing a leap from single-link optimization to system-level customization. Furthermore, relying on the real-time monitoring and feedback mechanism of the execution layer, the system can continuously optimize the configuration based on actual operating data, forming a closed loop of perception-decision-execution-optimization, which greatly improves the long-term adaptability and stability of the system and effectively improves the efficiency of intelligent sub-warehouse logistics. 2. By adopting a demand profiling model, three-dimensional demand labels and weights are automatically generated, completely eliminating the reliance on human experience and ensuring the objectivity, consistency, and quantifiability of demand profiles, laying a solid foundation for precise customization of logistics solutions. At the same time, normalization processing makes the weights of each dimension equal, eliminating the difference in units of measurement and making the label weights of different sub-warehouses comparable. This helps to accurately and efficiently draw demand profiles based on the actual situation of sub-warehouses, making it easier to efficiently customize logistics solutions for sub-warehouses. 3. The system first filters a large number of candidate solutions by sub-warehouse type, quickly narrowing down the matching range, avoiding unnecessary waste of computing resources, and significantly improving system response speed and recommendation efficiency. Then, by using a comprehensive matching degree calculation formula, the system comprehensively considers the three core requirements of function, goods, and environment and their relative importance, and conducts a refined quantitative evaluation of the candidate solutions. This scientifically ensures that the comprehensive suitability of the final recommended solution is optimal, thereby effectively improving the logistics efficiency of the intelligent sub-warehouse. Attached Figure Description

[0020] Figure 1 This is a system block diagram of an IoT-based intelligent warehousing and logistics system according to an embodiment of this application; Figure 2 This is a flowchart of the data processing method in the embodiments of this application; Figure 3 This is a flowchart of the feature extraction method in data processing in the embodiments of this application; Figure 4 This is a flowchart of the method for generating three-dimensional demand labels in the embodiments of this application; Figure 5 This is a flowchart of the method for classifying sub-compartments in this application embodiment; Figure 6 This is a flowchart of the matching and generation method in the embodiments of this application.

[0021] Explanation of reference numerals in the attached diagram: 1. Perception layer; 11. Basic attribute acquisition unit; 12. Operation data acquisition unit; 13. Cargo feature acquisition unit; 14. Environmental parameter acquisition unit; 2. Data processing layer; 3. Demand profiling layer; 4. Logistics system customized model layer; 5. Execution layer; 51. Solution deployment unit; 52. Dynamic monitoring unit; 53. Optimization triggering unit; 6. Central coordination layer. Detailed Implementation

[0022] The following is in conjunction with the appendix Figures 1-6 This application will be described in further detail.

[0023] This application discloses an intelligent warehousing and logistics system based on the Internet of Things (IoT). (Refer to...) Figure 1 An IoT-based intelligent warehousing and logistics system includes: Perception layer 1 is used to collect basic attribute data, duty operation data, cargo characteristic data and environmental parameter data of each sub-warehouse of the smart warehouse; Data processing layer 2, which is communicatively connected to the perception layer 1, is used to clean, standardize and extract features from the data collected by the perception layer 1, and output sub-warehouse feature data. The demand profiling layer 3 is communicatively connected to the data processing layer 2 and is used to generate three-dimensional demand labels for the sub-warehouse based on the sub-warehouse feature data, including functional demand dimension, cargo demand dimension and environmental demand dimension. Each dimension includes at least one label item, and each label item is associated with a weight value. The logistics system customization model layer 4 is communicatively connected to the demand profiling layer 3, and includes a classification model and a recommendation model. The classification model divides sub-warehouses into preset sub-warehouse types based on the three-dimensional demand tags. The recommendation model outputs a customized logistics system configuration scheme based on the sub-warehouse type and the weight values ​​of the three-dimensional demand tags. The customized logistics system configuration scheme includes at least one combination of logistics equipment, at least one optimization algorithm module, and at least one system function module and installation, deployment, and adaptation scheme. For example, the logistics equipment combination includes AGVs, sorting machines, and automated storage and retrieval systems; the optimization algorithm module includes path optimization algorithms and inventory allocation algorithms; and the system function module includes a temperature control traceability module and an anomaly warning module. The execution layer 5, communicating with the customized model layer 4 of the logistics system, is used to deploy the logistics system according to the customized logistics system configuration scheme and monitor the operation data of the sub-warehouses in real time to dynamically optimize the customized logistics system configuration scheme. Through the perception layer 1, which covers all dimensions of data including basic attributes, operations, goods, and environment of the sub-warehouses, three-dimensional demand labels are set after data processing to accurately quantify the personalized needs of each sub-warehouse in terms of function, goods, and environment, achieving precise matching of sub-warehouse logistics needs and overcoming the drawbacks of standardized configuration. Simultaneously, through the linkage of classification and recommendation models, it can not only automatically identify the sub-warehouse type but also output a complete customized scheme covering hardware equipment, algorithms, and functional modules, achieving a leap from single-stage optimization to system-level customization. Furthermore, relying on the real-time monitoring and feedback mechanism of the execution layer 5, the system can continuously optimize the configuration based on actual operating data, forming a closed loop of perception-decision-execution-optimization, greatly improving the long-term adaptability and stability of the system and effectively improving the efficiency of intelligent sub-warehouse logistics.

[0024] Reference Figure 1 It also includes a central coordination layer 6, which connects to the execution layers 5 of each sub-warehouse for cross-sub-warehouse data synchronization and resource scheduling. By setting up the central coordination layer 6, the data and resource barriers between sub-warehouses are broken down, which helps to achieve unified scheduling and optimal allocation of resources such as inventory, manpower, and transportation capacity, significantly improving the overall efficiency and emergency response capabilities of the entire warehousing network. At the same time, through centralized coordination and management, the failure or fluctuation of a single sub-warehouse can be compensated by the system's global resource allocation, enhancing the stability of the overall network and facilitating the rapid access and integrated management of new sub-warehouses.

[0025] Reference Figure 1 The perception layer 1 includes: The basic attribute acquisition unit 11 is used to collect the geographical location information, surrounding traffic data and building parameter information of the sub-warehouse, and package them to generate the basic attribute data of the sub-warehouse. The operation data acquisition unit 12 is used to collect data on the average daily order volume, sorting frequency, storage cycle and peak inbound and outbound periods of the sub-warehouse, and package them to generate the responsibility operation data of the sub-warehouse. The cargo feature acquisition unit 13 includes an RFID reader and a visual recognition device, used to collect cargo category, size and temperature and humidity requirements, and package and generate cargo feature data for the sub-warehouse. The environmental parameter acquisition unit 14 includes temperature and humidity sensors and dust sensors, used to collect temperature, humidity, and dust concentration, and package them to generate environmental parameter data for the sub-warehouse. Through the configuration of each unit, the basic attribute data, operational data, cargo characteristic data, and environmental parameter data of the sub-warehouse are accurately collected. First, the inherent conditions of the sub-warehouse are anchored, focusing on geographical location, transportation, and building parameters, providing core basic data for subsequent equipment layout and transportation connections, avoiding blind configuration that deviates from the inherent attributes of the sub-warehouse. Then, dynamic data such as average daily orders and sorting frequency are accurately captured, clearly reflecting the sub-warehouse's operational load and peak patterns, providing operational basis for subsequent algorithm scheduling (such as equipment expansion during peak periods) and efficiency optimization. Simultaneously, combined with RFID (efficient cargo identification) and visual recognition (accurate size / category measurement), the system comprehensively collects cargo category, size, and temperature and humidity requirements, directly matching cargo adaptation requirements (such as temperature and humidity corresponding to cold chain equipment), reducing the adaptation deviation between cargo and the system. Finally, multiple environmental sensors are integrated to achieve real-time monitoring of key parameters such as temperature, humidity, and dust, providing direct data support for the storage of special cargo and the configuration of environmentally adaptable equipment, effectively ensuring cargo safety and a stable operating environment.

[0026] Reference Figure 2 The data processing layer 2 cleans, standardizes, and extracts features from the data collected by the perception layer 1, and outputs sub-warehouse feature data, specifically including the following steps: A1. Data cleaning: Clean the raw data collected by the perception layer 1, remove outliers and redundant data caused by repeated reporting by the device, and fill in the missing data. A2. Data Standardization: Normalize the cleaned data to obtain standardized data; A3. Data Feature Extraction: Extracting features from standardized data to obtain feature vectors for constructing demand profiles. The feature vectors include static features, dynamic features, and environmental features. A4. Data Feature Integration: The extracted static, dynamic, and environmental features are concatenated into vectors to form sub-warehouse feature data that represents the comprehensive characteristics of the sub-warehouse, which is then output to the demand profiling layer 3. Through a systematic data cleaning and standardization process, noise is effectively removed and missing data is repaired, transforming multi-source heterogeneous raw data into high-quality, computable standard data, providing reliable input for subsequent intelligent models. Static, dynamic, and environmental features are intelligently extracted from the standardized data to deeply explore the sub-warehouse's operational patterns, business models, and environmental constraints hidden beneath the raw data, providing highly abstract and information-rich feature vectors for accurate demand profiling. Finally, feature vectors from different dimensions are concatenated and integrated to form standardized feature data that comprehensively describes the sub-warehouse, providing a structured and unified input for demand profiling layer 3, greatly improving the accuracy of demand profiling.

[0027] Reference Figure 3The above-mentioned feature extraction of standardized data to obtain feature vectors for constructing demand profiles specifically includes the following steps: B1. Static Feature Extraction: Extract structured features of sub-warehouses from basic attribute data, such as the ratio of area to floor height and the traffic convenience index.

[0028] B2. Dynamic Feature Extraction: Extract time-series features from job operation data, including but not limited to: order volume fluctuation coefficient calculated based on order data from the previous logistics cycle, and sorting efficiency calculated based on sorting time.

[0029] B3. Environmental Feature Extraction: Statistical features are extracted from environmental parameter data, including but not limited to: the variance of temperature and humidity data, the daily average maximum value of dust concentration, and the cumulative duration of ultraviolet radiation exceeding the standard in the past week. Through the above steps, the most core feature indicators (such as structural ratio, fluctuation coefficient, and environmental variance) are extracted from attribute, operational, and environmental data respectively, directly corresponding to the three major demand dimensions of function, goods, and environment, providing the most representative input for generating high-precision demand labels. Among them, static feature extraction anchors the inherent adaptability of sub-warehouses, dynamic feature extraction captures the operational rhythm pattern, and environmental feature extraction quantifies the environmental risk dimension. The simultaneous extraction of static (space), dynamic (efficiency), and environmental (risk) features fully covers the physical constraints, operational efficiency, and external challenges of sub-warehouses, avoiding the profile bias caused by single-dimensional features, and further improving the accuracy of demand profiles.

[0030] Reference Figure 4 The above-mentioned requirement profiling layer 3 generates a three-dimensional requirement label for the sub-warehouse based on the sub-warehouse feature data, which includes functional requirement dimensions, cargo requirement dimensions, and environmental requirement dimensions. Specifically, this includes the following steps: C1. Generating Functional Requirement Dimensions: Based on sub-warehouse feature data, sorting efficiency labels and weights, storage density labels and weights, and reverse processing labels and weights are generated for the functional requirement dimensions through a pre-set requirement profile model; the requirement profile model is a machine learning model obtained through iterative training of historical data. C2. Generate cargo demand dimensions: Based on the sub-warehouse feature data, generate cold chain demand labels and weights, explosion-proof demand labels and weights, and flexible processing labels and weights for cargo demand dimensions through a pre-set demand profile model; C3. Generation of environmental requirements dimension: Based on the sub-compartment feature data, the low temperature tolerance label and weight, salt spray protection label and weight, and low pressure adaptation label and weight are generated through a pre-set requirement profile model. C4. Normalization: Normalize the weight values ​​of each label item in the functional requirements dimension, the goods requirements dimension, and the environmental requirements dimension to ensure that the sum of the weight values ​​of each label item in each dimension is 1. C5. Generate 3D Demand Labels: Combine the normalized weight values ​​with the corresponding label items to generate the 3D demand labels for the sub-warehouse, and output them to the logistics system's customized model layer 4. Through the above steps, the demand profiling model automatically generates 3D demand labels and weights, completely eliminating reliance on manual experience and ensuring the objectivity, consistency, and quantifiability of the demand profile, laying a solid foundation for precise customization of logistics solutions. Simultaneously, normalization ensures that the weights of each dimension are summed to one, eliminating dimensional differences and making the label weights of different sub-warehouses comparable. This helps to accurately and efficiently draw demand profiles based on the actual situation of each sub-warehouse, facilitating efficient customization of logistics solutions for each sub-warehouse.

[0031] Furthermore, the classification model of the logistics system's customized model layer 4 is a multi-classification model based on gradient boosting trees, and the recommendation model is a collaborative filtering recommendation model. The preset sub-warehouse types include sorting, storage, cold chain, and hazardous materials. The gradient boosting tree model can efficiently and accurately process the multi-dimensional features (static / dynamic / environmental) of sub-warehouses, precisely classifying them into four core types: sorting, storage, cold chain, and hazardous materials. The classification logic aligns with actual operational scenarios, effectively narrowing the subsequent recommendation range and avoiding interference from irrelevant configurations. The collaborative filtering model can draw on the adaptation experience of similar sub-warehouses, deeply exploring the implicit correlation between historical solutions and sub-warehouse needs. Combining the three-dimensional demand label weights of sub-warehouses (such as the "low-temperature tolerance weight" for cold chain warehouses), it outputs solutions that conform to both type commonalities (such as cold chain requiring low-temperature equipment) and individual differences (such as cold chain warehouses with high sorting requirements being equipped with temperature-controlled sorting machines), resulting in better adaptability. By setting up a two-tier architecture, efficiency and accuracy are balanced. The process of classifying first and then recommending reduces the computational redundancy of the recommendation model (no need to match non-target type devices), improves the efficiency of solution generation, and avoids the generalization bias of a single model by focusing on classification and refining recommendations, ensuring the accuracy of configuration. It intelligently matches the most suitable combination of devices, algorithms and functional modules for various sub-warehouses, realizing true personalized customization.

[0032] Among them, reference Figure 5 The above classification model, based on the three-dimensional demand labels, divides sub-warehouses into preset sub-warehouse types, specifically including the following steps: D1. Feature vectorization: Combine the weight values ​​of each dimension in the three-dimensional demand label into a multi-dimensional numerical feature vector; D2. Type Probability Prediction: Input the multidimensional numerical feature vector into the pre-trained classification model to obtain the probability value of the sub-warehouse belonging to each preset sub-warehouse type; D3. Type Determination: The sub-warehouse type with the highest probability value is selected as the final classification result for that sub-warehouse. If the highest probability value is lower than a preset confidence threshold, the sub-warehouse is marked as a pending type and sent to management for manual review. The weight values ​​of the three-dimensional requirement labels are combined into a multi-dimensional numerical vector, transforming qualitative requirements into quantitative data. The classification model then outputs the probability values ​​of each sub-warehouse belonging to its respective type, rather than directly classifying them. This clearly reflects the type matching degree, providing valuable decision-making reference information for uncertain scenarios and enhancing the system's practicality and flexibility in handling complex real-world scenarios. Simultaneously, the introduction of a confidence threshold-triggered manual review mechanism provides reliable error prevention and correction for low-confidence predictions, forming a collaborative decision-making model with machine judgment as the primary method and human intervention as a backup, greatly improving the overall reliability of the classification results.

[0033] Reference Figure 6 The above recommendation model, based on the sub-warehouse type and the weight values ​​of the three-dimensional demand labels, outputs a customized logistics system configuration scheme, specifically including the following steps: E1. Initial screening of candidate solutions: The recommendation model generates candidate logistics system configuration solutions that are suitable for the sub-warehouse type based on the sub-warehouse type matching and screening. E2. Multi-dimensional matching calculation: The recommendation model calculates the comprehensive matching score of each candidate logistics system configuration scheme using a pre-set comprehensive matching score calculation formula; the comprehensive matching score calculation formula is as follows: ; Where S is the overall matching score, , , These are the importance coefficients for the functional requirements, cargo requirements, and environmental requirements dimensions pre-set for the corresponding sub-warehouse types. , and These represent the weight values ​​of the i-th functional requirement label, the j-th goods requirement label, and the k-th environmental requirement label in the sub-warehouse 3D requirement labels generated by the requirement profile layer, respectively. , and These represent the adaptability scores of each candidate logistics system configuration scheme for the i-th functional requirement, j-th cargo requirement, and k-th environmental requirement tag item, respectively; m, n, and p represent the total number of tag items in the three-dimensional requirement tags for the energy requirement dimension, cargo requirement dimension, and environmental requirement dimension, respectively.

[0034] E3. Optimal Solution Generation: The candidate logistics system configuration with the highest comprehensive matching score is selected as the recommended customized logistics system configuration. Through the above steps, a massive pool of candidate solutions is initially screened based on sub-warehouse type, quickly narrowing the matching range, avoiding unnecessary waste of computational resources, and significantly improving system response speed and recommendation efficiency. Then, using the comprehensive matching score calculation formula, the three core requirements of function, goods, and environment, along with their relative importance, are comprehensively considered to conduct a refined quantitative evaluation of the candidate solutions, scientifically ensuring that the final recommended solution has the optimal comprehensive suitability, effectively improving the logistics efficiency of intelligent sub-warehouses.

[0035] Reference Figure 1 Execution layer 5 includes: The solution deployment unit 51 is used to load the hardware and software modules of the customized logistics system configuration solution, and to complete the system integration and function activation. The dynamic monitoring unit 52 is used to monitor the operation data of the sub-warehouse in real time. The operation data includes equipment operating status, order processing efficiency, energy consumption data and environmental parameters. The optimization trigger unit 53 is used to trigger the logistics system customization model layer 4 to recalculate and output a new customized logistics system configuration scheme when the deviation between the monitored pre-set key performance indicators and the expected targets continuously exceeds a preset threshold. The scheme deployment unit 51 achieves standardized operation of the entire process of hardware loading, software configuration, and system integration, ensuring that the customized scheme can be deployed and implemented efficiently and accurately in the physical sub-warehouse, eliminating errors in the implementation stage. Through the collaborative work of dynamic monitoring and the optimization trigger unit 53, a complete closed loop from data perception to decision optimization is constructed, enabling the system to continuously self-adjust and optimize based on the actual operating status and the deviation of key performance indicators as the basis for optimization triggering, which helps to improve the system's adaptability, universality, and service life.

[0036] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit the scope of protection of the invention. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on these embodiments, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art can still combine, add, delete, or otherwise adjust the features of the various embodiments of the present invention according to the circumstances without conflict or creative effort, thereby obtaining different technical solutions that do not fundamentally depart from the concept of the present invention. These technical solutions also fall within the scope of protection of the present invention.

Claims

1. An intelligent warehousing and logistics system based on the Internet of Things, characterized in that, include: The perception layer (1) is used to collect basic attribute data, duty operation data, cargo characteristic data and environmental parameter data of each sub-warehouse of the smart warehouse; The data processing layer (2) is communicatively connected to the perception layer (1) and is used to clean, standardize and extract features from the data collected by the perception layer (1) and output sub-warehouse feature data. The demand profile layer (3) is connected to the data processing layer (2) and is used to generate a three-dimensional demand label for the sub-warehouse based on the sub-warehouse feature data, which includes functional demand dimension, cargo demand dimension and environmental demand dimension. Each dimension includes at least one label item, and each label item is associated with a weight value. The logistics system customization model layer (4) is connected to the demand profiling layer (3) and includes a classification model and a recommendation model; The classification model divides sub-warehouses into preset sub-warehouse types based on the three-dimensional demand labels; the recommendation model outputs a customized logistics system configuration scheme based on the weight values ​​of the sub-warehouse types and the three-dimensional demand labels; the customized logistics system configuration scheme includes at least one combination of logistics equipment, at least one optimization algorithm module, and at least one system function module and installation, deployment and adaptation scheme. The execution layer (5) is connected to the logistics system customization model layer (4) and is used to deploy the logistics system according to the customized logistics system configuration scheme and monitor the operation data of the sub-warehouse in real time to dynamically optimize the customized logistics system configuration scheme.

2. The intelligent warehousing and logistics system based on the Internet of Things according to claim 1, characterized in that, The sensing layer (1) includes: The basic attribute collection unit (11) is used to collect the geographical location information, surrounding traffic data and building parameter information of the sub-warehouse, and package them to generate the basic attribute data of the sub-warehouse. The operation data acquisition unit (12) is used to collect the daily average order volume, sorting frequency, storage cycle and peak inbound and outbound time data of the sub-warehouse, and package them to generate the sub-warehouse's duty operation data; The cargo feature collection unit (13) includes an RFID reader and a visual recognition device, used to collect cargo category, size and temperature and humidity requirements, and package and generate cargo feature data for the sub-warehouse; The environmental parameter acquisition unit (14) includes a temperature and humidity sensor and a dust sensor, which are used to collect temperature, humidity and dust concentration and package them to generate environmental parameter data of the sub-compartment.

3. The intelligent warehousing and logistics system based on the Internet of Things according to claim 1, characterized in that, The process of cleaning, standardizing, and extracting features from the data collected by the perception layer (1) to output sub-warehouse feature data specifically includes the following steps: A1. Data cleaning: Clean the raw data collected by the perception layer (1), remove outliers and redundant data caused by repeated reporting by the device, and fill in the missing data. A2. Data Standardization: Normalize the cleaned data to obtain standardized data; A3. Data Feature Extraction: Feature extraction is performed on standardized data to obtain feature vectors for constructing demand profiles. The feature vectors include static features, dynamic features, and environmental features. A4. Data feature integration: The extracted static features, dynamic features and environmental features are vectorized and concatenated to form sub-warehouse feature data that represents the full range of characteristics of the sub-warehouse, and output to the demand profile layer (3).

4. The intelligent warehousing and logistics system based on the Internet of Things according to claim 3, characterized in that, The process of extracting features from standardized data to obtain feature vectors for constructing demand profiles specifically includes the following steps: B1. Static Feature Extraction: Extract structured features of sub-warehouses from basic attribute data, such as the ratio of area to floor height and the traffic convenience index; B2. Dynamic Feature Extraction: Extract time-series features from job operation data, including but not limited to: order volume fluctuation coefficient calculated based on order data from the previous logistics cycle, and sorting efficiency calculated based on sorting time. B3. Environmental Feature Extraction: Extract statistical features from environmental parameter data, including but not limited to: the variance of temperature and humidity data, the daily average maximum value of dust concentration, and the cumulative duration of ultraviolet radiation exceeding the standard in the past week.

5. The intelligent warehousing and logistics system based on the Internet of Things according to claim 1, characterized in that, The requirement profiling layer (3) generates three-dimensional requirement labels for sub-warehouses based on sub-warehouse feature data, including functional requirement dimensions, cargo requirement dimensions, and environmental requirement dimensions. Specifically, this includes the following steps: C1. Generating Functional Requirement Dimensions: Based on sub-warehouse feature data, sorting efficiency labels and weights, storage density labels and weights, and reverse processing labels and weights are generated for the functional requirement dimensions through a pre-set requirement profile model; the requirement profile model is a machine learning model obtained through iterative training of historical data. C2. Generate cargo demand dimensions: Based on the sub-warehouse feature data, generate cold chain demand labels and weights, explosion-proof demand labels and weights, and flexible processing labels and weights for cargo demand dimensions through a pre-set demand profile model; C3. Generation of environmental requirements dimension: Based on the sub-compartment feature data, the low temperature tolerance label and weight, salt spray protection label and weight, and low pressure adaptation label and weight are generated through a pre-set requirement profile model. C4. Normalization: Normalize the weight values ​​of each label item in the functional requirements dimension, the goods requirements dimension, and the environmental requirements dimension to ensure that the sum of the weight values ​​of each label item in each dimension is 1. C5. Combine and generate three-dimensional demand labels: Combine the normalized weight values ​​with the corresponding label items to generate the three-dimensional demand labels for the sub-warehouse and output them to the logistics system customized model layer (4).

6. The intelligent warehousing and logistics system based on the Internet of Things according to claim 1, characterized in that: The classification model is a multi-classification model based on gradient boosting trees, the recommendation model is a collaborative filtering recommendation model, and the preset sub-warehouse types include sorting type, storage type, cold chain type, and hazardous materials type.

7. The intelligent warehousing and logistics system based on the Internet of Things according to claim 1, characterized in that, The classification model divides sub-warehouses into preset sub-warehouse types based on the three-dimensional demand labels, specifically including the following steps: D1. Feature vectorization: Combine the weight values ​​of each dimension in the three-dimensional demand label into a multi-dimensional numerical feature vector; D2. Type Probability Prediction: Input the multidimensional numerical feature vector into the pre-trained classification model to obtain the probability value of the sub-warehouse belonging to each preset sub-warehouse type; D3. Type determination: Select the sub-warehouse type with the highest probability value as the final classification result of the sub-warehouse; and when the highest probability value is lower than the preset confidence threshold, the sub-warehouse is marked as a pending type and sent to the management personnel for manual review.

8. The intelligent warehousing and logistics system based on the Internet of Things according to claim 1, characterized in that, The recommendation model, based on the sub-warehouse type and the weight values ​​of the three-dimensional demand labels, outputs a customized logistics system configuration scheme, specifically including the following steps: E1. Initial screening of candidate solutions: The recommendation model generates candidate logistics system configuration solutions that are suitable for the sub-warehouse type based on the sub-warehouse type matching and screening. E2. Multi-dimensional matching calculation: The recommendation model calculates the comprehensive matching score of each candidate logistics system configuration scheme using a pre-set comprehensive matching score calculation formula; the comprehensive matching score calculation formula is as follows: ； Where S is the overall matching score, , , These are the importance coefficients for the functional requirements, cargo requirements, and environmental requirements dimensions pre-set for the corresponding sub-warehouse types. , and These represent the weight values ​​of the i-th functional requirement label, the j-th goods requirement label, and the k-th environmental requirement label in the sub-warehouse 3D requirement labels generated by the requirement profile layer, respectively. , and These represent the adaptability scores of each candidate logistics system configuration scheme for the i-th functional requirement, j-th cargo requirement, and k-th environmental requirement tag item, respectively; m, n, and p represent the total number of tag items in the three-dimensional requirement tags for the energy requirement dimension, cargo requirement dimension, and environmental requirement dimension, respectively. E3. Optimal Solution Generation: Select the candidate logistics system configuration scheme with the highest comprehensive matching score as the recommended customized logistics system configuration scheme.

9. The intelligent warehousing and logistics system based on the Internet of Things according to claim 1, characterized in that, The execution layer (5) includes: The solution deployment unit (51) is used to load the hardware and software modules of the customized logistics system configuration solution and to complete the system integration and function activation. The dynamic monitoring unit (52) is used to monitor the operation data of the sub-warehouse in real time. The operation data includes equipment operating status, order processing efficiency, energy consumption data and environmental parameters. The optimization trigger unit (53) is used to trigger the logistics system customization model layer (4) to recalculate and output a new customized logistics system configuration scheme when the deviation between the monitored pre-set key performance indicators and the expected target continues to exceed the preset threshold.

10. The intelligent warehousing and logistics system based on the Internet of Things according to claim 1, characterized in that: It also includes a central coordination layer (6), which connects to the execution layer (5) of each sub-warehouse and is used for cross-sub-warehouse data synchronization and resource scheduling.

Images