An intelligent warehousing management system and method based on the Internet of Things

Abstract

The application relates to the technical field of intelligent warehousing, and discloses an intelligent warehousing management system and method based on the Internet of Things, which comprises the following steps: real-time monitoring of a cold-chain warehousing environment and states of each intelligent agent, and collection of relevant data; each intelligent agent is responsible for controlling an independent physical device, wherein the types of the independent physical devices include a generator, a motor and a sensor; each intelligent agent evaluates the state thereof and shares state information to enhance a global perspective; and a conflict detection algorithm is used to identify potential strategy conflicts between all pairs of intelligent agents. The application can significantly improve the collaborative efficiency of each subsystem in the cold-chain warehousing, reduce energy consumption and goods loss, and guarantee the continuous and stable operation of the system by means of synchronous sampling of multiple source sensors, modeling based on a conflict function, dynamic influence weight distribution for enhancing a global collaborative perspective, conflict threshold value updating of a Q function iteration, and action optimization driven by a coordination factor to solve multi-intelligent agent strategy conflicts.

Description

Technical Field

[0001] This invention relates to the field of smart warehousing, and more specifically, to a smart warehouse management system and method based on the Internet of Things. Background Technology

[0002] Cold chain warehousing is a specialized warehousing system that uses refrigeration technology and equipment to maintain a low-temperature environment. It is mainly used for the preservation of temperature-sensitive goods such as food and medicine, and is the core link of cold chain logistics.

[0003] In cold chain warehousing, because the internal environment of cold storage is controlled by multiple control devices, when optimizing refrigeration units and independently scheduling and managing AGVs, the frequent opening of cold storage doors can lead to the loss of cold air, increasing the difficulty of warehouse management adjustments. Summary of the Invention

[0004] This invention provides an intelligent warehouse management system and method based on the Internet of Things, solving the technical problems in related technologies.

[0005] This invention provides an intelligent warehouse management method based on the Internet of Things, comprising the following steps:

[0006] S100, Data Collection: Real-time monitoring of the cold chain storage environment and the status of each intelligent agent, and collection of relevant data;

[0007] Each intelligent agent is responsible for controlling an independent physical device, the types of which include generators, motors, and sensors;

[0008] S200, State Assessment and Information Sharing: Each agent assesses its own state and shares state information to enhance the global perspective;

[0009] S300, Conflict Detection: Employs a conflict detection algorithm to identify potential policy conflicts between all pairs of agents;

[0010] S400, Coordination Decision: Through a coordination mechanism, adjust the action choices in strategic conflicts to achieve global optimization;

[0011] S500, Control Execution: Applying the coordinated strategies of each intelligent agent to the actual control system to adjust the equipment operating parameters;

[0012] S600, Feedback Learning and Adjustment: Based on the adjustment of equipment operating parameters and environmental feedback, the policy parameters of each agent are updated, and self-optimization is carried out by using the reward mechanism in reinforcement learning;

[0013] S700, Strategy Deployment and Closed-Loop Management: Apply updated strategy parameters to the actual cold chain storage environment, monitor and control the effect, and form a continuous optimization closed loop.

[0014] Furthermore, S100 specifically includes the following steps:

[0015] S110, multi-source sensor synchronous sampling: collects spatial temperature field through distributed optical fiber temperature measurement network, obtains cargo thermal inertia parameters using RFID shelf reader, and collects cold storage door status and compressor power;

[0016] S120, Agent State Synchronization: Subscribe to the state vectors published by each agent through ROS nodes, record the path planning parameters of the AGV scheduling agent, and collect the control parameters of the chiller unit agent.

[0017] S130, Heterogeneous data preprocessing: sliding window calibration is performed on temperature data, wavelet denoising is performed on power data, and spatiotemporal alignment is performed;

[0018] S140, Metadata Encapsulation and Transmission: The processed data is encapsulated into tuples of a unified format and transmitted to the edge computing node via the OPC UA protocol;

[0019] S150, Output of the data collection phase: The time series dataset is finally generated through tuples in a uniform format.

[0020] Furthermore, S200 specifically includes the following steps:

[0021] S210, Define the conflict function: Each agent calculates the degree of conflict between local and global objectives based on its own state and external input;

[0022] The formula for calculating the conflict function is as follows:

[0023] ;

[0024] in Represents intelligent agents At any moment The conflict value, Represents intelligent agents The current state vector, Represents the global target state vector. Represents intelligent agents The neighborhood group, , This represents the weighting coefficients in the conflict function. >0, >0, ;

[0025] S220, Calculate influence weights: Dynamically allocate the influence weights of agents based on conflict values ​​and historical performance;

[0026] S230, State Information Sharing: Based on influence weights, update the local state and broadcast it to neighboring agents;

[0027] S240, Global Consistency Check: Checks whether the deviations between the states of all agents and the global objective have converged.

[0028] Furthermore, the S300 specifically includes the following steps:

[0029] S310, Parameter Predefined: Initialize the agent action set, define the conflict determination threshold, policy network parameters, learning rate and discount factor;

[0030] S320, Conflict Function Modeling: Define a binary conflict function to quantify the policy conflict between two actions;

[0031] The formula for calculating the bivariate conflict function is as follows:

[0032] ;

[0033] in Represents a binary conflict function. Indicates action Weight of global impact The state-action value function;

[0034] S330, Agent Action Pair Traversal: Traverse all agent action pairs to avoid redundant calculations;

[0035] S340, Dynamic Conflict Threshold Update: The conflict determination threshold is updated iteratively based on the Q function, in sync with the strategy optimization;

[0036] S350, Calculate the total number of conflicts: Calculate the conflict results of all agent pairs and output the total number of conflicts.

[0037] Furthermore, the agent action pairs in S330 are ;

[0038] in , This represents the total number of intelligent agents.

[0039] Furthermore, the S400 specifically includes the following steps:

[0040] S410, Initialize Coordination Factor: Set the coordination factor to balance the weight of its own Q value and external influences;

[0041] S420, Calculate the influence function of other agents' actions: For each non-current agent, quantify the influence value of its actions on the agent;

[0042] S430, combining Q-value and external influences: summing the weighted influences of its own Q-value and other agents to generate an adjusted evaluation value;

[0043] S440, Select the optimal action based on the adjustment value: Determine the new action by maximizing the adjusted evaluation value;

[0044] S450, Conflict-aware Q-value correction: If a new action triggers a conflict, the Q-value update magnitude is reduced proportionally.

[0045] Furthermore, the S500 specifically includes the following steps:

[0046] S510, Strategy Mapping and System Integration: Mapping the coordinated strategy to the physical parameters of the actual control system;

[0047] S520, Real-time Parameter Adaptive Adjustment: Dynamically corrects equipment parameters based on feedback signals to ensure that control objectives align with strategies;

[0048] S530, Conflict Resolution and Stability Verification: Detects conflicts in multi-device collaboration and verifies the validity of parameters through stability criteria.

[0049] Furthermore, the S600 specifically includes the following steps:

[0050] S610, Calculate the target Q value: Construct the target value based on the current state, action, immediate reward, and the maximum expected Q value of the next state;

[0051] S620, calculates timing difference error: quantizes the direction of strategy optimization by the difference between the target value and the current Q value;

[0052] S630, Update Q function parameters: Adjust Q function parameters according to error. This gradually approaches the optimal strategy.

[0053] Furthermore, the S700 specifically includes the following steps:

[0054] S710, Policy Distribution and Device Synchronization: The updated policy parameters are distributed to each device controller to ensure policy synchronization across multiple devices;

[0055] S720, Real-time Control and Status Monitoring: The equipment adjusts operating parameters based on the new strategy and collects status data in real time;

[0056] S730, verification of multi-device collaboration effect: detecting device collaboration conflicts and verifying system stability;

[0057] S740, Performance Feedback and Dynamic Readjustment: Triggers adaptive adjustments and reinforcement learning updates based on real-time performance;

[0058] S750, closed-loop learning cycle: Stores device operation data in the experience pool, periodically triggers the Q function update of S600, forming a "strategy optimization-execution-feedback" closed loop.

[0059] This invention also proposes an IoT-based intelligent warehouse management system, which executes the aforementioned IoT-based intelligent warehouse management method, including:

[0060] Data acquisition and synchronization module: Executes the S100 stage tasks, including multi-source sensor synchronization, agent state synchronization, heterogeneous data processing, metadata encapsulation and time series dataset generation;

[0061] Multi-agent collaborative decision-making module: integrates S200-S400 stages to realize state assessment, conflict detection, and coordinated decision-making;

[0062] Equipment control and execution module: corresponding to the S500 stage, responsible for policy mapping, adaptive parameter adjustment and conflict resolution verification;

[0063] Reinforcement learning feedback optimization module: covering the S600-S700 stage, realizing policy parameter updates and closed-loop deployment;

[0064] Global monitoring and verification module: runs through the process of S100-S700, and performs multi-device collaborative verification, performance feedback and stability monitoring.

[0065] The beneficial effects of this invention are as follows:

[0066] This invention uses multi-source sensor synchronous sampling and enhances the global collaborative perspective based on conflict function modeling and dynamic influence weight allocation. It combines Q-function iterative conflict threshold updates and action optimization driven by coordination factors to resolve multi-agent strategy conflicts. This can significantly improve the collaborative efficiency of various subsystems in cold chain warehousing, reduce energy consumption and cargo loss, and ensure the continuous and stable operation of the system. Attached Figure Description

[0067] Figure 1 This is a flowchart of an intelligent warehouse management method based on the Internet of Things proposed in this invention;

[0068] Figure 2 This is a structural block diagram of an intelligent warehouse management system based on the Internet of Things proposed in this invention.

[0069] In the diagram: 101. Data acquisition and synchronization module; 102. Multi-agent collaborative decision-making module; 103. Equipment control and execution module; 104. Reinforcement learning feedback optimization module; 105. Global monitoring and verification module. Detailed Implementation

[0070] The subject matter described herein will now be discussed with reference to exemplary embodiments. It should be understood that these embodiments are discussed only to enable those skilled in the art to better understand and implement the subject matter described herein, and changes may be made to the function and arrangement of the elements discussed without departing from the scope of this specification. Various processes or components may be omitted, substituted, or added as needed in the examples. Furthermore, features described in some examples may be combined in other examples.

[0071] like Figure 1 As shown, an IoT-based intelligent warehouse management method includes the following steps:

[0072] S100, Data Collection: Real-time monitoring of the cold chain storage environment and the status of each intelligent agent, and collection of relevant data;

[0073] In one embodiment of the present invention, the specific steps are as follows:

[0074] S110, multi-source sensor synchronous sampling: acquires spatial temperature field through a distributed fiber optic temperature measurement network. ( , (Identify sensor numbers); use RFID shelf readers to obtain the thermal inertia parameters of the goods. ( , (For shelf numbering); collect cold storage door status data. (Door opening duration) and compressor power ;

[0075] The calculation formula is as follows:

[0076] ;

[0077] ;

[0078] ;

[0079] in Indicates temperature error. Represents the actual temperature. ,in This represents the variance of temperature measurement noise, where Indicates the first Heat capacity of goods on each shelf, Indicates the power error, where , Denotes the Laplace distribution, where and These are the temperature change and the time interval, respectively. This represents the noise intensity parameter for power measurement, where This represents the compressor reference voltage, where Indicates the set temperature;

[0080] S120, Agent State Synchronization: Subscribing to the state vectors published by each agent through ROS nodes. ( (Identify the agent); record the path planning parameters of the AGV scheduling agent. ; Collect control parameters of the refrigeration unit's intelligent agent ;

[0081] in This indicates the maximum speed of the AGV equipment. This indicates the maximum acceleration of the AGV equipment. Indicates the safe distance of the AGV, where Indicates the bandwidth after the compressor starts and stops;

[0082] State synchronization model:

[0083] ;

[0084] in Indicates the first An intelligent agent in The state of observation at any given moment. This represents the action vector from the previous moment. This represents the reward value at the previous moment. Indicates the first An intelligent agent in Time control parameters, Indicates the first An intelligent agent in The state at any given moment;

[0085] S130, Heterogeneous Data Preprocessing:

[0086] Perform sliding window calibration on temperature data:

[0087] ;

[0088] in Indicates the first The drift coefficient of each sensor, This indicates the calibrated temperature data. Indicates temperature error;

[0089] Perform wavelet noise reduction on the power data:

[0090] ;

[0091] in This indicates the compressor power data after noise reduction. This represents the total number of wavelet decomposition layers (usually 5-8). Represents the filter function. Indicates the wavelet decomposition level;

[0092] Spacetime alignment processing (resolving 5ms-level clock skew):

[0093] ;

[0094] in Indicates the timestamp after synchronization. Represents the original timestamp. Indicates network clock synchronization error. This indicates the amount of data in the aligned data;

[0095] S140, Metadata Encapsulation and Transmission:

[0096] The processed data is encapsulated into tuples of a uniform format:

[0097] ;

[0098] Transmitted to edge computing nodes via OPC UA protocol:

[0099] ;

[0100] in This indicates the encoding length of a single temperature data (fixed at 32 bits). Indicates network bandwidth. Indicates the protocol stack processing latency;

[0101] S150, Output of the data collection phase: The time series dataset is finally generated through tuples in a unified format.

[0102] ;

[0103] in The data collection cycle is designed to meet the real-time requirements of cold chain management. Indicates the initial time of data collection. Indicates the end time of data collection.

[0104] S200, State Assessment and Information Sharing: Each agent assesses its own state and shares state information to enhance the global perspective;

[0105] In one embodiment of the present invention, the specific steps are as follows:

[0106] S210, Define the conflict function: Each agent calculates the degree of conflict between local and global objectives based on its own state and external input;

[0107] The calculation formula is as follows:

[0108] ;

[0109] in Represents intelligent agents At any moment The conflict value, Represents intelligent agents The current state vector, This represents the global target state vector, which is provided by the data collection phase. Represents intelligent agents The neighborhood group, , Represents the weighting coefficients in the conflict function ( >0, >0, must meet the following conditions ).

[0110] S220, Calculate influence weights: Dynamically allocate the influence weights of agents based on conflict values ​​and historical performance;

[0111] The calculation formula is as follows:

[0112] ;

[0113] in Represents intelligent agents At any moment Influence weight, The attenuation coefficient representing the collision value ( ), Indicates the historical performance gain coefficient ( ),in Represents intelligent agents Historical performance rating Represents the total number of agents, of which Represents intelligent agents At any moment The conflict value.

[0114] S230, State Information Sharing: Based on influence weights, update the local state and broadcast it to neighboring agents;

[0115] The calculation formula is as follows:

[0116] ;

[0117] in Represents intelligent agents The updated state vector, Representing neighboring intelligent agents The current state vector, where Represents intelligent agents At any moment Influence weight.

[0118] S240, Global Consistency Check: Checks whether the deviations between the states of all agents and the global objective have converged;

[0119] The calculation formula is as follows:

[0120] ;

[0121] Termination condition: If (threshold) If the condition is met, it is considered converged; otherwise, parameter adjustment is triggered.

[0122] S300, Conflict Detection: Employs a conflict detection algorithm to identify potential policy conflicts between all pairs of agents;

[0123] In one embodiment of the present invention, the specific steps are as follows:

[0124] S310, Parameter predefinition: Initialize the agent action set Define the conflict determination threshold Policy network parameters Learning rate and discount factor ;

[0125] ;

[0126] in This represents the action of the i-th agent. , This represents the total number of intelligent agents.

[0127] S320, Conflict Function Modeling: Defining a Binary Conflict Function This quantifies the strategic contradiction between the two actions.

[0128] The calculation formula is as follows:

[0129] ;

[0130] in Indicates action Weight of global impact This is the state-action value function.

[0131] S330, Agent Action Pair Traversal: Traverse all agent action pairs ,in To avoid double counting, .

[0132] S340, Dynamic Conflict Threshold Update: Iteratively updating the conflict determination threshold based on the Q function. To ensure synchronization with strategy optimization;

[0133] The calculation formula is as follows:

[0134] ;

[0135] in This is the current state. For the next state, As candidate actions, This represents the conflict determination threshold for the next state. This indicates the current conflict determination threshold.

[0136] S350, Calculate the total number of conflicts: Calculate the conflict results of all agent pairs and output the total number of conflicts.

[0137] The calculation formula is as follows:

[0138] ;

[0139] in The total number of collisions is an integer, and its range is... .

[0140] S400, Coordination Decision: Through a coordination mechanism, conflicting action choices are adjusted to achieve global optimization. Its calculation formula is expressed as:

[0141] ;

[0142] in This represents the adjusted action of the i-th agent. As a coordinating factor, The effect of other agents' actions on the current agent.

[0143] S410, Initialize Coordination Factor: Set the coordination factor ( ), used to balance the weight of its own Q value and external influences;

[0144] in This represents the coordination factor (a scalar parameter that needs to be optimized experimentally).

[0145] S420, Calculate the influence function of actions of other agents: for each non-current agent... ( ), quantify its actions For intelligent agents The impact value;

[0146] The calculation formula is as follows:

[0147] ;

[0148] in This represents the conflict value of agent j. This represents the interaction distance (e.g., Euclidean distance) between agents i and j. Indicates the minimum value ( ≠0), This indicates the conflict indicator function (1 if a conflict exists, 0 otherwise).

[0149] S430, combining Q-value and external influences: summing the weighted influences of its own Q-value and other agents to generate an adjusted evaluation value;

[0150] The calculation formula is as follows:

[0151] ;

[0152] in Represents state-based and policy network parameters The value of the action, This represents the global impact term weighted by the coordination factor;

[0153] S440, Select the optimal action based on the adjustment value: Determine the new action by maximizing the adjusted evaluation value;

[0154] The calculation formula is as follows:

[0155] ;

[0156] in Represents the action space;

[0157] S450, Conflict-aware Q-value correction: If a new action causes a conflict, the Q-value update magnitude is reduced proportionally.

[0158] The calculation formula is as follows:

[0159] ;

[0160] in This represents the corrected Q value. Indicates the learning rate. This indicates the current maximum conflict value. Indicates the preset conflict threshold;

[0161] S500, Control Execution: Applying the coordinated strategies of each intelligent agent to the actual control system to adjust the equipment operating parameters;

[0162] In one embodiment of the present invention, the specific steps are as follows:

[0163] S510, Strategy Mapping and System Integration: Mapping coordinated strategies (such as optimization weights and priority rules) to the physical parameters of the actual control system.

[0164] Let the strategy parameters generated in step 3 be... (Dynamic weights) (Priority coefficient), the constraint boundary generated in step 4 is The target value for adjusting the equipment parameters is:

[0165] ;

[0166] in Indicates dynamic weights. , Indicates the priority coefficient. , Indicates the constraint boundary. .

[0167] S520, Real-time Parameter Adaptive Adjustment: Dynamically corrects equipment parameters based on feedback signals to ensure consistency between control objectives and strategies.

[0168] Formula: Let the real-time feedback error be... The historical error mean is Then the parameter correction amount for:

[0169] ;

[0170] Updated parameters:

[0171] ;

[0172] in Indicates the updated parameters. Indicates the learning rate. Indicates the attenuation factor. Indicates instantaneous error. This represents the historical average error.

[0173] S530, Conflict Resolution and Stability Verification: Detects conflicts in multi-device collaboration and verifies the validity of parameters through stability criteria.

[0174] The conflict indicators are as follows:

[0175] ;

[0176] The stability criterion is as follows:

[0177] like and If so, it is determined to be stable;

[0178] in Indicates the conflict weight. , For reference benchmark value, The conflict threshold, This represents the threshold for the rate of change of error.

[0179] S600, Feedback Learning and Adjustment: Based on the adjustment of equipment operating parameters and environmental feedback, the policy parameters of each agent are updated, and self-optimization is carried out by using the reward mechanism in reinforcement learning;

[0180] Its calculation formula is expressed as follows:

[0181] ;

[0182] in As a reward, This is the discount factor.

[0183] S610, Calculate the target Q-value: based on the current state. ,action Instant rewards and the next state Maximum expectation Value, construct target value .

[0184] The calculation formula is as follows:

[0185] ;

[0186] in Indicates the target value. Indicates the discount factor. , Indicates the parameters of the Q function. This indicates an optional action for the next state.

[0187] S620, calculates the timing difference error (TD Error): quantizes the direction of strategy optimization by the difference between the target value and the current Q value.

[0188] Its calculation formula is expressed as follows:

[0189] ;

[0190] in This is the error.

[0191] S630, Parameter Update: Adjusts the parameters of the Q function based on the error. In order to gradually approach the optimal strategy.

[0192] Its calculation formula is expressed as follows:

[0193] ;

[0194] in Indicates the learning rate. ;

[0195] S700, Strategy Deployment and Closed-Loop Management: Apply updated strategy parameters to the cold chain storage environment, monitor and control effects, and form a continuous optimization closed loop;

[0196] In one embodiment of the present invention, the specific steps are as follows:

[0197] S710, Policy Distribution and Device Synchronization: Updated policy parameters (such as optimized ones) will be distributed to the device. function parameters Dynamic weights The policy is distributed to each device controller to ensure synchronization across multiple devices;

[0198] Its calculation formula is expressed as follows:

[0199] ;

[0200] in Indicates the updated parameters. , The total number of devices. This represents the k-th device parameter.

[0201] S720, Real-time Control and Status Monitoring: The equipment adjusts operating parameters (such as power and speed) based on a new strategy and collects status data (such as...) in real time. , ).

[0202] The formula for calculating action selection is as follows:

[0203] ;

[0204] Among them, the aforementioned , , , .

[0205] S730, Verification of Multi-Device Collaboration Effect: Detecting device collaboration conflicts and verifying system stability (using the aforementioned conflict indicators). (and stability criteria).

[0206] The calculation formula is as follows:

[0207] ;

[0208] S740, Performance Feedback and Dynamic Readjustment: Based on real-time performance (such as error) ,award This triggers the aforementioned adaptive adjustment or the aforementioned reinforcement learning update.

[0209] like Then recalculate. ;in This indicates the error threshold.

[0210] S750, Closed-Loop Learning Cycle: Transferring device operating data... Stored in the experience pool, periodically triggering the aforementioned... Function updates form a closed loop of "strategy optimization-execution-feedback".

[0211] It should be noted that each agent is responsible for controlling an independent physical device (e.g., the device numbered k, corresponding to...). ).

[0212] The types of independent physical devices include, but are not limited to, generators, motors, and sensors.

[0213] The agent sends policy parameters (such as...) through the device controller. ), directly drive the device to perform actions , where the state Includes real-time device parameters, rewards Calculated based on equipment performance indicators.

[0214] The multiple agents must meet system-level constraints to avoid conflicts in physical quantities between devices (such as power overload and frequency instability).

[0215] like Figure 2 As shown, the present invention also discloses an IoT-based intelligent warehouse management system, which executes the above-described IoT-based intelligent warehouse management method and includes the following modules:

[0216] Data acquisition and synchronization module 101: Executes the S100 stage tasks, including multi-source sensor synchronization, agent state synchronization, heterogeneous data processing, metadata encapsulation, and time series dataset generation;

[0217] Multi-agent collaborative decision-making module 102: integrates the S200-S400 stages to realize state assessment, conflict detection, and coordinated decision-making;

[0218] Equipment control and execution module 103: corresponding to the S500 stage, responsible for strategy mapping, adaptive parameter adjustment and conflict resolution verification;

[0219] Reinforcement Learning Feedback Optimization Module 104: Covers the S600-S700 stage, realizing policy parameter updates and closed-loop deployment;

[0220] Global monitoring and verification module 105: runs through the process of S100-S700, and performs multi-device collaborative verification, performance feedback and stability monitoring.

[0221] The present invention also discloses a storage medium storing non-transitory computer-readable instructions for performing one or more steps in the aforementioned Internet of Things-based intelligent warehouse management method.

[0222] Computer programs may be stored / distributed on suitable media, such as optical storage media or solid-state media supplied together with or as part of other hardware, but may also be distributed in other forms, such as via the Internet or other wired or wireless telecommunications systems. No reference numerals in the claims should be construed as limiting the scope.

[0223] The embodiments of this example have been described above. However, this example is not limited to the specific implementation methods described above. The specific implementation methods described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms based on the guidance of this example, and all of them are within the protection scope of this example.

Claims

1. A smart warehouse management method based on the Internet of Things, characterized in that, Includes the following steps: S100, Data Collection: Real-time monitoring of the cold chain storage environment and the status of each intelligent agent, and collection of relevant data; Each intelligent agent is responsible for controlling an independent physical device, the types of which include generators, motors, and sensors; S200, State Assessment and Information Sharing: Each agent assesses its own state and shares state information to enhance the global perspective; Specifically, S200 includes the following steps: S210, Define the conflict function: Each agent calculates the degree of conflict between local and global objectives based on its own state and external input; The formula for calculating the conflict function is as follows: Where C i (t) represents the conflict value of agent i at time t, x i (t) represents the current state vector of agent i, x global (t) represents the global target state vector, N i Let α1 and β1 represent the set of neighbors of agent i, and let α1 and β1 represent the weight coefficients in the conflict function, where α1 > 0, β1 > 0, and α1 + β1 = 1. S220, Calculate influence weights: Dynamically allocate the influence weights of agents based on conflict values ​​and historical performance; S230, State Information Sharing: Based on influence weights, update the local state and broadcast it to neighboring agents; S240, Global Consistency Check: Checks whether the deviations between the states of all agents and the global objective have converged; S300, Conflict Detection: Employs a conflict detection algorithm to identify potential policy conflicts between all pairs of agents; S400, Coordination Decision: Through a coordination mechanism, adjust the action choices in strategic conflicts to achieve global optimization; S500, Control Execution: Applying the coordinated strategies of each intelligent agent to the actual control system to adjust the equipment operating parameters; S600, Feedback Learning and Adjustment: Based on the adjustment of equipment operating parameters and environmental feedback, the policy parameters of each agent are updated, and self-optimization is carried out by using the reward mechanism in reinforcement learning; S700, Strategy Deployment and Closed-Loop Management: Apply updated strategy parameters to the actual cold chain storage environment, monitor and control the effect, and form a continuous optimization closed loop.

2. The intelligent warehouse management method based on the Internet of Things according to claim 1, characterized in that, S100 specifically includes the following steps: S110, multi-source sensor synchronous sampling: collects spatial temperature field through distributed optical fiber temperature measurement network, obtains cargo thermal inertia parameters using RFID shelf reader, and collects cold storage door status and compressor power; S120, Agent State Synchronization: Subscribe to the state vectors published by each agent through ROS nodes, record the path planning parameters of the AGV scheduling agent, and collect the control parameters of the chiller unit agent. S130, Heterogeneous data preprocessing: sliding window calibration is performed on temperature data, wavelet denoising is performed on power data, and spatiotemporal alignment is performed; S140, Metadata Encapsulation and Transmission: The processed data is encapsulated into tuples of a unified format and transmitted to the edge computing node via the OPC UA protocol; S150, Output of the data collection phase: The time series dataset is finally generated through tuples in a uniform format.

3. The intelligent warehouse management method based on the Internet of Things according to claim 2, characterized in that, The S300 specifically includes the following steps: S310, Parameter Predefined: Initialize the agent action set, define the conflict determination threshold, policy network parameters, learning rate and discount factor; S320, Conflict Function Modeling: Define a binary conflict function to quantify the policy conflict between two actions; The formula for calculating the bivariate conflict function is as follows: Where conflict(a i a j ) represents a binary conflict function, influence(a j ) represents action a j The weight of the global impact, Q(s) i a i ;θ) is the state-action value function; S330, Agent Action Pair Traversal: Traverse all agent action pairs to avoid redundant calculations; S340, Dynamic Conflict Threshold Update: The conflict determination threshold is updated iteratively based on the Q function, in sync with the strategy optimization; S350, Calculate the total number of conflicts: Calculate the conflict results of all agent pairs and output the total number of conflicts.

4. The intelligent warehouse management method based on the Internet of Things according to claim 3, characterized in that, The agent action pair in S330 is (a i a j ); Where i≠j, N represents the total number of agents.

5. The intelligent warehouse management method based on the Internet of Things according to claim 4, characterized in that, The S400 specifically includes the following steps: S410, Initialize Coordination Factor: Set the coordination factor to balance the weight of its own Q value and external influences; S420, Calculate the influence function of other agents' actions: For each non-current agent, quantify the influence value of its actions on the agent; S430, combining Q-value and external influences: summing the weighted influences of its own Q-value and other agents to generate an adjusted evaluation value; S440, Select the optimal action based on the adjustment value: Determine the new action by maximizing the adjusted evaluation value; S450, Conflict-aware Q-value correction: If a new action triggers a conflict, the Q-value update magnitude is reduced proportionally.

6. The intelligent warehouse management method based on the Internet of Things according to claim 5, characterized in that, The S500 specifically includes the following steps: S510, Strategy Mapping and System Integration: Mapping the coordinated strategy to the physical parameters of the actual control system; S520, Real-time Parameter Adaptive Adjustment: Dynamically corrects equipment parameters based on feedback signals to ensure that control objectives align with strategies; S530, Conflict Resolution and Stability Verification: Detects conflicts in multi-device collaboration and verifies the validity of parameters through stability criteria.

7. The intelligent warehouse management method based on the Internet of Things according to claim 6, characterized in that, The S600 specifically includes the following steps: S610, Calculate the target Q value: Construct the target value based on the current state, action, immediate reward, and the maximum expected Q value of the next state; S620, calculates timing difference error: quantizes the direction of strategy optimization by the difference between the target value and the current Q value; S630, Update Q function parameters: Adjust the parameters θ of the Q function according to the error, and gradually approach the optimal strategy.

8. The intelligent warehouse management method based on the Internet of Things according to claim 7, characterized in that, The S700 specifically includes the following steps: S710, Policy Distribution and Device Synchronization: The updated policy parameters are distributed to each device controller to ensure policy synchronization across multiple devices; S720, Real-time Control and Status Monitoring: The equipment adjusts operating parameters based on the new strategy and collects status data in real time; S730, verification of multi-device collaboration effect: detecting device collaboration conflicts and verifying system stability; S740, Performance Feedback and Dynamic Readjustment: Triggers adaptive adjustments and reinforcement learning updates based on real-time performance; S750, closed-loop learning cycle: Stores device operating data in the experience pool, periodically triggers the Q function update of S600, forming a "strategy optimization-execution-feedback" closed loop.

9. An intelligent warehouse management system based on the Internet of Things, characterized in that, Performing an IoT-based intelligent warehouse management method as described in any one of claims 1-8 includes: Data acquisition and synchronization module: Executes the S100 stage tasks, including multi-source sensor synchronization, agent state synchronization, heterogeneous data processing, metadata encapsulation and time series dataset generation; Multi-agent collaborative decision-making module: integrates S200-S400 stages to realize state assessment, conflict detection, and coordinated decision-making; Equipment control and execution module: corresponding to the S500 stage, responsible for policy mapping, adaptive parameter adjustment and conflict resolution verification; Reinforcement learning feedback optimization module: covering the S600-S700 stage, realizing policy parameter updates and closed-loop deployment; Global monitoring and verification module: runs through the process of S100-S700, and performs multi-device collaborative verification, performance feedback and stability monitoring.

Images