Intelligent logistics and supply chain collaborative management method based on Internet of Things

Abstract

The invention relates to the technical field of supply chain management, and discloses a smart logistics and supply chain collaborative management method based on the Internet of Things, which comprises the following steps: 1, in logistics transportation and storage links, collecting logistics-related multi-source data by using Internet of Things equipment arranged on a transportation vehicle, a storage facility and a distribution terminal, and storing the logistics-related multi-source data in the logistics transportation and storage links; the collected data are sent to an edge computing node for preprocessing through a wireless communication network; and 2, performing formatting conversion and anomaly detection on the received multi-source data at an edge computing node. The supply chain scheduling model of reinforcement learning optimization is introduced, the transportation and inventory strategies can be adaptively adjusted in a complex logistics network, dynamic optimization is achieved, compared with a traditional logistics scheduling method based on a static rule, the limitation of fixed threshold setting is broken through, the problem that the response speed is low under sudden orders and inventory fluctuation is solved, and the logistics scheduling method is suitable for large-scale popularization and application. The flexibility of the supply chain is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of supply chain management, and specifically provides a method for collaborative management of intelligent logistics and supply chain based on the Internet of Things. Background Art

[0002] In recent years, with the rapid development of e-commerce, intelligent manufacturing, and global supply chains, the complexity of logistics and supply chain management has been continuously increasing. To improve operational efficiency and reduce transportation and warehousing costs, enterprises have gradually introduced technologies such as the Internet of Things, big data analysis, and artificial intelligence to achieve intelligent logistics scheduling and supply chain collaborative management. However, existing logistics management systems still have many deficiencies in terms of data accuracy, scheduling optimization, and dynamic adaptability, and cannot meet the requirements of efficient, intelligent, and flexible management of modern logistics networks.

[0003] Currently, logistics systems widely rely on Internet of Things technology to collect transportation, warehousing, and order data in real time through devices such as GPS, RFID, and temperature and humidity sensors. However, due to problems such as sensor measurement errors, data loss, and signal interference, the data collected by the system is uncertain, resulting in deviations in decision-making for path planning and inventory management.

[0004] Under the traditional logistics management mode, the scheduling system relies on preset rules for decision-making, such as replenishing goods at fixed time intervals or managing inventory based on historical average order volumes. While the scheduling method based on static rules can play a certain role under stable demand, it lacks flexibility in the face of large fluctuations in order demand and frequent emergencies.

[0005] Existing logistics optimization algorithms are mainly based on linear programming, heuristic algorithms, or rule engines, which can improve transportation efficiency and reduce costs to a certain extent, but have limitations in the face of complex and changing supply chain environments.

[0006] In view of the above problems, technical personnel in this field provide a method for collaborative management of intelligent logistics and supply chain based on the Internet of Things to solve the above-mentioned problems. Summary of the Invention

[0007] Aiming at the deficiencies of the prior art, the present invention provides a method for collaborative management of intelligent logistics and supply chain based on the Internet of Things to solve the problems raised in the above background art.

[0008] To achieve the above objectives, the present invention is realized through the following technical solutions: A method for collaborative management of intelligent logistics and supply chain based on the Internet of Things includes the following steps:

[0009] Step 1: In the logistics transportation and warehousing links, use the Internet of Things devices deployed on transportation vehicles, warehousing facilities, and distribution terminals to collect multi-source data related to logistics, and send the collected data to the edge computing node for preprocessing through a wireless communication network;

[0010] Step 2: At the edge computing node, perform formatting conversion and anomaly detection on the received multi-source data, and perform interpolation processing on the missing values in combination with historical data;

[0011] Step 3: Based on the preprocessed data, use the state estimation method to perform fusion calculation on the key state variables of the logistics system to obtain the inventory status, transportation status, and order matching situation, so as to generate structured data;

[0012] Step 4: On the basis of data fusion, according to the constraint conditions in the logistics network, use the path planning method to calculate the optimal transportation path, optimize the vehicle scheduling plan in combination with the supply chain scheduling model, and then transmit the optimized scheduling plan to the intelligent decision-making system for execution;

[0013] Step 5: On the basis of the optimized transportation path and scheduling plan, use the intelligent decision-making system to respond to emergencies in real time, and dynamically adjust the transportation strategy, inventory replenishment plan, and warehousing management plan based on historical data and real-time monitoring information;

[0014] Step 6: Execute the logistics scheduling and inventory management strategies generated by the intelligent decision-making system at each node of the supply chain, and collect the execution data in real time through the Internet of Things devices, feedback it to the edge computing node for analysis, verify the decision-making effect, and perform iterative adjustment on the optimization model according to the feedback results.

[0015] Preferably, the data preprocessing in the second step adopts the state correction method based on error covariance estimation to improve the accuracy of the collected data, and the data state correction calculation is as follows:

[0016] X t =X t-1 +K t (Z t -HX t-1 ),

[0017] where, X t is the state variable at time t, X t-1 is the state variable at time t-1, K t is the gain matrix, Z t is the observed data, and H is the observation matrix.

[0018] Preferably, the state estimation method in the third step adopts a multi-dimensional Markov state transition model to predict the future state of the logistics system, and the state transition equation is as follows:

[0019]

[0020] where P(s t+1 |s t ) is the total probability of transitioning from state s t to state s t+1 .

[0021] s t is the state of the logistics system at time t, s t+1 is the predicted state of the logistics system at time t + 1, a t is the decision-making action taken at time t, and A is the set of all possible actions,

[0022] P(s t+1 |s t , a t ) is the conditional probability of transitioning from state s t to state s t after taking action a t+1 .

[0023] π(a t |s t ) is the probability of selecting action a t in state s t under the current policy.

[0024] Preferably, the path planning method in the fourth step adopts a dynamic path optimization model based on the optimal logistics cost, and constructs the path cost function of the transportation network as follows:

[0025]

[0026] where C path is the transportation path cost, N is the number of nodes on the path, d i is the distance of the i-th path segment, v i is the average speed of the path segment, α is the carbon emission cost coefficient, and e i is the carbon emission of the path segment.

[0027] Preferably, the supply chain scheduling model in the fourth step adopts a reinforcement learning optimization method, and defines the immediate reward function as follows:

[0028] R(s t , a t ) = -(C transport (s t , a t ) + C storage (s t , a t ))

[0029] where R(st , a t ) is the reward obtained for executing action a t in state s t .

[0030] C transport (s t , a t ) is the logistics transportation cost.

[0031] C storage (s t , a t ) is the inventory storage cost.

[0032] Preferably, the intelligent decision-making system in the fifth step uses a deep Q-network for decision optimization, and the loss function is defined as follows:

[0033]

[0034] where L(θ) is the loss function of the deep Q-network, θ is the neural network parameter, E is the mathematical expectation, r t is the current reward, γ is the discount factor.

[0035] is the maximum Q value obtained by taking the best action a t+1 in the next state s ′ , θ - is the target network parameter.

[0036] Q(s t , a t ; θ) is the Q value calculated for taking action a t in the current state s t .

[0037] Preferably, when the intelligent decision-making system in the fifth step dynamically adjusts the inventory replenishment plan, it adopts an inventory level adjustment strategy based on optimal control, and the dynamic change equation of the inventory state is as follows:

[0038] I t+1 = I t + O t - D t ,

[0039] where I t+1 is the inventory level at time t + 1, I t is the inventory level at time t, O t is the replenishment quantity, and D t is the order demand.

[0040] Preferably, after the execution data in the sixth step is fed back to the edge computing node, a policy optimization method based on long-term return estimation is used to iteratively adjust the optimization model. The formula for the long-term return is as follows:

[0041]

[0042] where G t is the long-term return accumulated from time t, and γ k is the discount factor.

[0043] R(s t+k , a t+k ) is the immediate reward obtained by taking action a t+k at time t + k in state s t+k .

[0044] Preferably, during the iterative adjustment process of the optimization model, an experience replay mechanism is used to store historical decision data, and the model parameters are updated using the method of mini-batch sampling to reduce data correlation and improve learning stability.

[0045] Preferably, the intelligent logistics and supply chain collaborative management method is applicable to logistics scenarios such as urban distribution, long-distance transportation, and warehouse management, and different optimization strategies are adopted to improve the adaptability of logistics supply chain management.

[0046] In the urban distribution scenario, the vehicle route planning parameters and scheduling priorities are dynamically adjusted according to the urban road traffic conditions and order density.

[0047] In the long-distance transportation scenario, the main line transportation route and vehicle configuration plan are optimized in combination with transportation costs, vehicle load, and energy consumption.

[0048] In the warehouse management scenario, based on inventory demand forecasting and warehouse space utilization rate, the inventory replenishment threshold and warehouse layout strategy are adjusted to improve storage efficiency and order fulfillment speed.

[0049] The present invention provides an intelligent logistics and supply chain collaborative management method based on the Internet of Things, having the following beneficial effects:

[0050] 1. The present invention adopts a state correction method based on error covariance estimation to improve the accuracy of logistics data, reduce the influence of sensor noise on decision-making, and solve the problem of decision deviation caused by data error accumulation compared with the method of calculating paths and schedules relying on raw sensor data in the prior art, enabling accurate planning of transportation paths.

[0051] 2. The supply chain scheduling model optimized by the present invention through reinforcement learning can adaptively adjust transportation and inventory strategies in complex logistics networks to achieve dynamic optimization. Compared with traditional logistics scheduling methods based on static rules, it breaks through the limitation of fixed threshold setting, solves the problem of slow response speed under sudden orders and inventory fluctuations, and improves the flexibility of the supply chain.

[0052] 3. The present invention uses a deep Q-network for intelligent decision-making optimization, which can perform iterative learning based on historical data and real-time feedback, enabling the logistics system to continuously improve scheduling efficiency during long-term operation. Compared with traditional methods based on linear optimization, it avoids the limitations of manual parameter setting and solves the deficiency that scheduling schemes cannot adapt to actual needs in a complex environment with multiple variables. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 is a flowchart of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0054] To enable those skilled in the art to understand the solution of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the scope of protection of the present invention.

[0055] The present invention will be described in detail below with reference to the accompanying drawings:

[0056] Embodiment:

[0057] Please refer to the attached Figure 1 , the embodiment of the present invention provides an Internet of Things-based intelligent logistics and supply chain collaborative management method, including the following steps:

[0058] First step, in the logistics transportation and warehousing links, use Internet of Things devices deployed on transportation vehicles, warehousing facilities, and distribution terminals to collect multi-source data related to logistics, and send the collected data to the edge computing node through a wireless communication network for preprocessing;

[0059] Second step, at the edge computing node, perform formatting conversion and anomaly detection on the received multi-source data, and perform interpolation processing on missing values in combination with historical data;

[0060] Third step, based on the preprocessed data, use a state estimation method to perform fusion calculation on the key state variables of the logistics system to obtain the inventory state, transportation state, and order matching situation, so as to generate structured data;

[0061] Step 4: On the basis of data fusion, according to the constraints in the logistics network, use the path planning method to calculate the optimal transportation path, optimize the vehicle scheduling plan in combination with the supply chain scheduling model, and then transmit the optimized scheduling plan to the intelligent decision-making system for execution;

[0062] Step 5: On the basis of the optimized transportation path and scheduling plan, use the intelligent decision-making system to respond to emergencies in real time, and dynamically adjust the transportation strategy, inventory replenishment plan, and warehousing management plan based on historical data and real-time monitoring information;

[0063] Step 6: Execute the logistics scheduling and inventory management strategies generated by the intelligent decision-making system at each node of the supply chain, and collect the execution data in real time through Internet of Things devices, and feedback it to the edge computing node for analysis to verify the decision-making effect, and iteratively adjust the optimization model according to the feedback results.

[0064] Benefits of the first step of Internet of Things devices collecting multi-source logistics data: By deploying Internet of Things devices in transportation vehicles, warehousing facilities, and distribution terminals, real-time and accurate data collection can be achieved, avoiding the lag and errors of manual records. The collected data is transmitted to the edge computing node through the wireless communication network, improving the timeliness of the data and ensuring the real-time and high efficiency of the logistics system.

[0065] Benefits of the second step of data preprocessing: Through format conversion, data from different sources can be processed under the same framework, improving data consistency and availability. The anomaly detection technology can eliminate error data and avoid wrong decisions caused by incorrect sensor data. Interpolation processing of missing values in combination with historical data reduces the impact of data missing on optimization decisions and improves data integrity.

[0066] Benefits of the third step of logistics status estimation and data fusion: By using the state estimation method, multi-source sensor data can be fused to provide accurate inventory status, transportation status, and order matching. The generated structured data can provide reliable input for subsequent path planning and scheduling optimization, improving the intelligence level of the logistics system.

[0067] Benefits of the fourth step of calculating the optimal transportation path and supply chain scheduling optimization: Based on the constraints of the logistics network, calculate the optimal transportation path, reduce transportation time and costs, and improve distribution efficiency. Optimize the vehicle scheduling plan in combination with the supply chain scheduling model, make the allocation of transport capacity resources reasonable, reduce the empty load rate, and improve operation efficiency.

[0068] Benefits of the fifth step of dynamic adjustment by the intelligent decision-making system: It can respond to emergencies in real time, ensuring the stability and robustness of the logistics system. Dynamically adjust the transportation strategy, inventory replenishment plan, and warehousing management plan through historical data + real-time monitoring information, improving the flexibility and adaptability of the supply chain.

[0069] Benefits of the sixth-step feedback mechanism and optimization model iteration: Real-time collection of execution data through IoT devices to verify the effectiveness of intelligent decisions and ensure the reliability of optimization solutions. The feedback data is used for iterative adjustment of the optimization model, enabling the logistics scheduling system to continuously learn and optimize itself, and improving the intelligent level of long-term operation.

[0070] In the data preprocessing in the second step, a state correction method based on error covariance estimation is adopted to improve the accuracy of the collected data. The data state correction calculation is as follows:

[0071] X t = X t-1 + K t (Z t - HX t-1 ),

[0072] where X t is the state variable at time t, X t-1 is the state variable at time t-1, K t is the gain matrix, Z t is the observed data, and H is the observation matrix.

[0073] Using the error covariance estimation method to correct the data can effectively reduce the influence of sensor noise. For example, there are deviations in the vehicle position data measured by GPS devices. Reliable position information can be obtained after state correction to ensure accurate scheduling decisions.

[0074] This method relies on current observed data and combines past state information for correction, so that even if the data of a certain sensor is lost or abnormal, historical data can still be used for compensation to improve the stability of logistics state estimation.

[0075] The error covariance estimation method can be used in inventory level correction, transportation route optimization, and order matching to enhance data consistency and enable subsequent optimization algorithms to calculate based on reliable data.

[0076] By dynamically adjusting the correction amplitude through the Kalman gain matrix, the system can continuously optimize the estimation results according to real-time data, improving the real-time response ability of supply chain scheduling.

[0077] In the state estimation method in the third step, a multi-dimensional Markov state transition model is adopted to predict the future state of the logistics system. The state transition equation is as follows:

[0078]

[0079] where P(s t+1 |s t ) is the total probability of the state s t transitioning to the state s t+1 .

[0080] s t is the state of the logistics system at time t, s t+1 is the predicted state of the logistics system at time t+1, a t is the decision-making action taken at time t, and A is the set of all available actions

[0081] P(s t+1 |s t ,a t ) is the conditional probability of transitioning to state s t after taking action a t in state s t+1 .

[0082] π(a t |s t ) is the probability of selecting action a t in state s t under the current policy

[0083] The multi-dimensional Markov state transition model can predict the future system state based on historical states and decision-making behaviors. Compared with traditional rule-based prediction methods, this method can accurately simulate the dynamic changes of the logistics system, making scheduling and inventory management precise

[0084] This model considers the current state of the logistics system and combines the impact of different decision-making actions on state transitions, enabling the system to evaluate the long-term benefits of different strategies. For example, selecting different replenishment times or delivery routes will affect future inventory levels and transportation efficiency. This method can calculate the optimal transition paths of different strategies, improving the quality of decision-making

[0085] This method can adapt to various logistics scenarios, including urban distribution, long-distance transportation, and warehouse management. The state variables and action sets can be dynamically adjusted in different scenarios, enabling the model to have generalization ability. For example, in urban distribution, the state variables include order density and traffic conditions, while in warehouse management, they include inventory levels and replenishment frequencies

[0086] By modeling the probability of action selection in different states through a policy function, the system can gradually learn the optimal policy and achieve intelligent scheduling. For example, the system can dynamically adjust the warehouse replenishment policy based on historical data, avoiding inventory shortages or surpluses and improving the response speed of the supply chain

[0087] The path planning method in the fourth step adopts a dynamic path optimization model based on the optimal logistics cost, and constructs the path cost function of the transportation network as follows

[0088]

[0089] where C pathLet \(C\) be the cost of the transportation path, \(N\) be the number of nodes on the path, and \(d\) i be the distance of the \(i\)-th path segment, and \(v\) i be the average speed of the path segment, \(\alpha\) be the carbon emission cost coefficient, and \(e\) i be the carbon emission of the path segment.

[0090] This path optimization model comprehensively considers the distance and average speed of the path, can dynamically calculate the transportation time of different paths, preferentially selects the path with the shortest total transportation time, reduces order delivery delays, and improves logistics efficiency.

[0091] Traditional path planning focuses on transportation time or distance, while this model introduces a carbon emission cost coefficient. While optimizing transportation time, it can reduce carbon emissions, making the logistics system environmentally friendly and in line with the concept of modern green supply chain management.

[0092] This model dynamically adjusts the optimal path according to real-time traffic conditions, order demands, and environmental factors, avoiding congestion or resource waste caused by traditional fixed path planning, and improving the flexibility and adaptability of the supply chain.

[0093] This path optimization model can be applied to urban distribution, long-distance transportation, and transfer between warehouses, and has strong generalization ability. For example, in urban distribution, this method can select congested roads that avoid peak hours, while in long-distance transportation, it can preferentially select highways to reduce the total transportation time.

[0094] The supply chain scheduling model in the fourth step adopts a reinforcement learning optimization method, and defines the immediate reward function as follows:

[0095] \(R(s\) t ,a\) t ) = -(C\) transport (s\) t ,a\) t ) + C\) storage (s\) t ,a\) t )),

[0096] where \(R(s\) t ,a\) t ) is the reward obtained by executing action \(a\) t in state \(s\) t ,

[0097] C\) transport (s\) t ,a\) t ) is the logistics transportation cost,

[0098] C\) storage (s\) t ,a\) t ) is the inventory storage cost.

[0099] Through the reinforcement learning method, the system can continuously optimize the decision-making strategy based on real-time data and historical experience, gradually improving the scheduling ability of the supply chain. Different from traditional rule-based methods, reinforcement learning can adaptively adjust the strategy according to the actual situation, enabling the system to make optimal decisions in a changing environment.

[0100] The immediate reward function takes into account both transportation costs and inventory storage costs. By balancing the two, the system can minimize costs while ensuring the best state of logistics efficiency and inventory management. Multi-objective optimization can ensure efficient resource utilization in a dynamic environment.

[0101] The reinforcement learning model has dynamic adaptability and can adjust the scheduling strategy in real time in the face of order demand fluctuations, inventory changes, and emergencies. The system can make immediate responses according to different situations, reducing delays and improving the flexibility of the supply chain.

[0102] By defining the immediate reward function, the system optimizes the benefits of the current decision and can achieve global optimality through long-term learning strategies. For example, adjusting the replenishment frequency or scheduling path can significantly reduce logistics costs and inventory backlogs in the long term, improving the overall efficiency of the supply chain.

[0103] The intelligent decision-making system in the fifth step uses a deep Q-network for decision optimization, and the loss function is defined as follows:

[0104]

[0105] Where L(θ) is the loss function of the deep Q-network, θ is the neural network parameter, E is the mathematical expectation, r t is the current reward, γ is the discount factor,

[0106] is the maximum Q value obtained by taking the best action a t+1 in the next state s ′ , θ - is the target network parameter,

[0107] Q(s t , a t ; θ) is the Q value calculated by taking action a t in the current state s t .

[0108] The deep Q-network combines reinforcement learning and deep neural networks, enabling the system to autonomously learn the optimal decision-making strategy without the need for manual setting of complex rules, improving the intelligence level of logistics scheduling and supply chain management.

[0109] The discount factor in the loss function enables the system to consider both current and future benefits when making decisions, thereby optimizing the long-term operational efficiency of the supply chain. For example, in inventory replenishment decisions, the deep Q-network can balance the short-term cost of immediate replenishment and the impact on long-term supply chain stability.

[0110] The deep Q-network adopts target network parameters and updates them with the current network parameters, which helps reduce the volatility of Q-value estimation, improve the stability of system training, and avoid drastic oscillations in the strategy.

[0111] The deep Q-network is applied to various complex supply chain scenarios, such as multi-warehouse inventory management, urban distribution scheduling, and long-distance transportation optimization. By processing multi-dimensional input data through a deep neural network, the deep Q-network can make better decisions in a high-dimensional complex state space and improve the operating efficiency of the logistics system.

[0112] When the intelligent decision-making system in the fifth step dynamically adjusts the inventory replenishment plan, it adopts an inventory level adjustment strategy based on optimal control. The dynamic change equation of the inventory state is as follows:

[0113] I t+1 =I t +O t -D t ,

[0114] where, I t+1 is the inventory level at time t + 1, I t is the inventory level at time t, O t is the replenishment quantity, and D t is the order demand quantity.

[0115] This method adopts an inventory level adjustment strategy based on optimal control, which can dynamically adjust the replenishment quantity and optimize inventory management according to order demand. Compared with traditional fixed replenishment rules, this method is intelligent, efficient, and flexible, can adapt to market demand changes, optimize the inventory level, reduce inventory costs, and improve the overall operational efficiency of the supply chain.

[0116] After the execution data in the sixth step is fed back to the edge computing node, a strategy optimization method based on long-term return estimation is adopted to iteratively adjust the optimization model. The long-term return calculation formula is as follows:

[0117]

[0118] where, G t is the long-term return accumulated starting from time t, γ k is the discount factor,

[0119] R(s t+k ,a t+k ) is at time t + k, state s t+kTake action a below t+k The immediate reward obtained.

[0120] This method optimizes the strategy based on the long-term return estimation, uses the discount factor to balance the short-term and long-term benefits, dynamically adjusts the strategy in complex supply chain scenarios, and improves the global benefits and adaptability of decision-making. Compared with traditional static optimization methods, this solution has dynamic response capabilities, stability, and wide applicability, and can effectively improve the intelligent level and operation efficiency of the supply chain.

[0121] During the iterative adjustment process of the optimization model, the experience replay mechanism is adopted to store historical decision-making data, and the mini-batch sampling method is used to update the model parameters to reduce data correlation and improve learning stability.

[0122] The intelligent logistics and supply chain collaborative management method is applicable to logistics scenarios such as urban distribution, long-distance transportation, and warehouse management, and different optimization strategies are adopted to improve the adaptability of logistics supply chain management;

[0123] In the urban distribution scenario, according to the urban road traffic status and order density, dynamically adjust the vehicle route planning parameters and scheduling priorities;

[0124] In the long-distance transportation scenario, combine the transportation cost, vehicle load, and energy consumption conditions to optimize the main line transportation route and vehicle configuration plan;

[0125] In the warehouse management scenario, based on the inventory demand forecast and warehouse space utilization rate, adjust the inventory replenishment threshold and warehouse layout strategy to improve the storage efficiency and order fulfillment speed.

[0126] Although the embodiments of the present invention have been shown and described, for those of ordinary skill in the art, it can be understood that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principle and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for collaborative management of intelligent logistics and supply chain based on the Internet of Things, characterized in that, The method includes the following steps: In the first step, in the logistics transportation and warehousing links, use the Internet of Things devices deployed on transportation vehicles, warehousing facilities, and distribution terminals to collect multi-source data related to logistics, and send the collected data to the edge computing node for preprocessing through a wireless communication network; In the second step, at the edge computing node, perform formatting conversion and anomaly detection on the received multi-source data, and perform interpolation processing on the missing values in combination with historical data; In the third step, based on the preprocessed data, use the state estimation method to perform fusion calculation on the key state variables of the logistics system to obtain the inventory status, transportation status, and order matching situation, so as to generate structured data; In the fourth step, on the basis of data fusion, according to the constraint conditions in the logistics network, use the path planning method to calculate the optimal transportation path, optimize the vehicle scheduling plan in combination with the supply chain scheduling model, and then transmit the optimized scheduling plan to the intelligent decision-making system for execution; In the fifth step, on the basis of the optimized transportation path and scheduling plan, use the intelligent decision-making system to respond to emergencies in real time, and dynamically adjust the transportation strategy, inventory replenishment plan, and warehousing management plan based on historical data and real-time monitoring information; In the sixth step, execute the logistics scheduling and inventory management strategies generated by the intelligent decision-making system at each node of the supply chain, and collect the execution data in real time through the Internet of Things devices, feedback it to the edge computing node for analysis, verify the decision-making effect, and perform iterative adjustment on the optimization model according to the feedback results.

2. The intelligent logistics and supply chain collaborative management method based on the Internet of Things according to claim 1, wherein, The data preprocessing in the second step adopts the state correction method based on error covariance estimation to improve the accuracy of the collected data. The data state correction calculation is as follows: X t = X t-1 + K t (Z t - HX t-1 ) where, X t is the state variable at time t, X t-1 is the state variable at time t-1, K t is the gain matrix, Z t is the observed data, and H is the observation matrix.

3. The intelligent logistics and supply chain collaborative management method based on the Internet of Things according to claim 2, wherein, The state estimation method in the third step adopts the multi-dimensional Markov state transition model to predict the future state of the logistics system. The state transition equation is as follows: Among them, P(s t+1 |s t ) is the total probability that state s t transfers to state s t+1 . s t is the state of the logistics system at time t, s t+1 is the predicted state of the logistics system at time t + 1, a t is the decision-making action taken at time t, and A is the set of all possible actions P(s t+1 |s t ,a t ) is the conditional probability of transitioning to state s t after taking action a t to state s t+1 , π(a t |s t ) is the probability of selecting action a t in state s t under the current policy.

4. A method for collaborative management of intelligent logistics and supply chain based on the Internet of Things according to claim 3, characterized in that, The path planning method in the fourth step adopts the dynamic path optimization model based on the optimal logistics cost, and constructs the path cost function of the transportation network as follows: Among them, C path is the transportation path cost, N is the number of nodes on the path, d i is the distance of the i-th path segment, v i is the average speed of the path segment, α is the carbon emission cost coefficient, e i is the carbon emission of the path segment.

5. The collaborative management method of intelligent logistics and supply chain based on the Internet of Things according to claim 4, characterized in that The supply chain scheduling model in the fourth step adopts the reinforcement learning optimization method, and defines the immediate reward function as follows: R(s t ,a t ) = -(C transport (s t ,a t ) + C storage (s t ,a t )) where R(s t , a t ) is the reward obtained by executing action a t in state s t . C transport (s t ,a t ) is the logistics transportation cost, C storage (s t ,a t ) is the inventory storage cost.

6. A method for collaborative management of intelligent logistics and supply chain based on the Internet of Things according to claim 5, characterized in that, The intelligent decision-making system in the fifth step adopts the deep Q network for decision optimization. The loss function is defined as follows: Among them, \(L(\theta)\) is the loss function of the deep Q-network, \(\theta\) is the neural network parameter, \(E\) is the mathematical expectation, \(r\) t is the current reward, and \(\gamma\) is the discount factor. To take the best action a t+1 in the next state s ′ the maximum Q value obtained, θ - is the target network parameter, Q(s t ,a t ; θ) is the Q-value calculated for the current state s t when taking action a t .

7. A method for collaborative management of intelligent logistics and supply chain based on the Internet of Things according to claim 6, characterized in that, When the intelligent decision-making system in the fifth step dynamically adjusts the inventory replenishment plan, it adopts the inventory level adjustment strategy based on optimal control. The dynamic change equation of the inventory status is as follows: I t+1 = I t + O t - D t , Among them, I t+1 is the inventory level at time t + 1, and I t is the inventory level at time t, O t is the replenishment quantity, and D t is the order demand quantity.

8. A collaborative management method for intelligent logistics and supply chain based on the Internet of Things according to claim 7, characterized in that, After the execution data in the sixth step is fed back to the edge computing node, the strategy optimization method based on long-term return estimation is adopted to perform iterative adjustment on the optimization model. The long-term return calculation formula is as follows: Among them, G t is the long-term return accumulated starting from time t, and γ k is the discount factor. R(s t+k ,a t+k ) is the immediate reward obtained for taking action a t+k in state s t+k at time t + k.

9. A method for collaborative management of intelligent logistics and supply chain based on the Internet of Things according to claim 8, characterized in that, During the iterative adjustment process of the optimization model, an experience replay mechanism is adopted to store historical decision-making data, and a small batch sampling method is used to update the model parameters to reduce data correlation and improve learning stability.

10. A method for collaborative management of intelligent logistics and supply chain based on the Internet of Things according to claim 9, characterized in that, The intelligent logistics and supply chain collaborative management method is applicable to logistics scenarios such as urban distribution, long-distance transportation, and warehousing management, and adopts different optimization strategies to improve the adaptability of logistics supply chain management; In the urban distribution scenario, according to the urban road traffic status and order density, dynamically adjust the vehicle path planning parameters and scheduling priorities; In the long-distance transportation scenario, optimize the main-line transportation route and vehicle configuration plan in combination with transportation costs, vehicle load capacity, and energy consumption. In the warehouse management scenario, based on inventory demand forecasting and warehouse space utilization rate, adjust the inventory replenishment threshold and warehouse layout strategy to improve storage efficiency and order fulfillment speed.

Images