Intelligent scheduling method for commodity full-link logistics digital management and control

Abstract

The application discloses an intelligent scheduling method for bulk commodity full-link logistics digital management and control, which comprises the following steps: after collecting port berth operation data, vehicle positioning data and ship arrival information, calling a berth vehicle flow coupling optimization solving algorithm to cooperatively decide ship berthing order and vehicle path, and outputting a berth allocation and vehicle guiding scheme; then generating a node logistics capacity time sequence table and a vehicle scheduling plan by a logistics capacity time sequence balancing scheduling algorithm; then starting a logistics node congestion situation judgment model to calculate a congestion pressure index and mark a level; calling a logistics energy consumption global statistics intelligent analysis system to extract surrounding vehicle energy consumption rates for a congestion node, and generating a detour energy consumption compensation scheme; and finally, fusing all the results to output scheduling instructions by a Pareto optimization iteration. The method integrates berth resource allocation, capacity balancing deployment, congestion early prediction and energy consumption optimization into a unified decision chain, effectively improving the collaborative efficiency and resource utilization rate of bulk commodity logistics full-link.

Description

Technical Field

[0001] This invention relates to the field of bulk commodity logistics management technology, and in particular to an intelligent scheduling method for digital management and control of the entire logistics chain of bulk commodities. Background Technology

[0002] Commodity trade is massive in scale, with supply chains covering multiple levels of logistics nodes, including ports, railway hubs, warehousing parks, and end-processing enterprises. The flow of goods involves complex processes such as ship berthing, vehicle allocation, multimodal transport connections, and warehousing and transshipment. The operational plans of each stage are interconnected, and transportation energy costs account for a continuously rising proportion of total operating expenses, while carbon emission reduction regulations are becoming increasingly stringent. Against this backdrop, logistics scheduling systems need to simultaneously address multiple challenges, including competition for berth resources, uneven spatial and temporal distribution of transport capacity, sudden congestion at key nodes, and dynamic statistics of energy consumption across the entire region. Traditional scheduling models relying on manual experience or single-dimensional optimization are no longer adequate for the large-scale, highly dynamic, and constrained operational needs of commodity supply chains.

[0003] Currently, the industry commonly uses port operating systems to schedule ship berthing sequences and quay crane operation plans, transportation management systems to record vehicle entry and exit times and allocate capacity based on fixed shifts, and warehouse management systems to monitor inventory levels at each node and set capacity warning thresholds. Some systems also incorporate independent energy consumption monitoring modules to collect vehicle fuel or electricity consumption data for post-event statistics. These subsystems are typically loosely integrated; berth scheduling outputs are manually imported into the transportation scheduling module, and the transportation plan is then transmitted to the warehouse management end. Energy consumption data is only summarized and reported at the end of the day or month. Data flow between different stages in the scheduling decision-making process relies on pre-set interface forms or manual intervention, lacking a unified digital management framework to integrate berth resources, capacity allocation, congestion prediction, and energy consumption statistics into a single optimization system.

[0004] Existing technologies have two main drawbacks. First, the independent operation of each subsystem results in a lack of dynamic feedback between berth allocation schemes and capacity scheduling plans. Fluctuations in actual ship berthing times directly lead to ineffective vehicle waiting or berth idleness, but existing systems cannot couple and adjust these schemes in real time. Overall operational efficiency is constrained by the time lag effect of the sequential decision-making chain. Second, existing technologies are limited to reactive, post-congestion response, i.e., replanning some vehicle routes after congestion occurs. Furthermore, the impact of increased energy consumption on total cost is not considered during this replanning, often resulting in detours sacrificing fuel economy for reduced travel time. This fails to create a balanced and optimized scheduling strategy that balances congestion avoidance and energy consumption control. Summary of the Invention

[0005] In order to overcome the shortcomings and deficiencies of existing technologies, this invention provides an intelligent scheduling method for digital management and control of the entire logistics chain of bulk commodities.

[0006] The technical solution adopted in this invention is an intelligent scheduling method for digital management and control of the entire logistics chain of bulk commodities, including the following steps: S1, collecting port berth operation time-series data, real-time location data of transport vehicles, in-transit inventory, and ship arrival forecast information, and forming a multi-source input feature set through data cleaning and structuring; S2, calling the berth-vehicle flow coupling optimization algorithm, using the ship berthing sequence and the route of collection and distribution vehicles as decision variables, constructing a dual-objective coupling function that minimizes the average ship waiting time in port and the average vehicle waiting time, and outputting a berth allocation scheme and a vehicle guidance path set; S3, inputting the berth allocation scheme and the vehicle guidance path set into the logistics capacity time-series equalization scheduling algorithm, with the objective of minimizing the deviation between the available capacity and the amount of goods to be transported in each transportation period, generating... S4. Based on the capacity deployment timeline and vehicle scheduling plan for each transportation node, initiate the logistics node congestion situation assessment model, calculate the instantaneous carrying capacity and throughput pressure index of each logistics node within the prediction time window, mark congested nodes exceeding the threshold, and output congestion level labels; S5. For the marked congested nodes, call the logistics energy consumption full-domain statistical intelligent analysis system to extract the real-time energy consumption rate and idling time of transportation vehicles around the congested nodes, and generate detour energy consumption compensation schemes in combination with route replanning strategies; S6. Integrate berth allocation schemes, vehicle guidance path sets, capacity deployment timelines, vehicle scheduling plans, congestion level labels, and detour energy consumption compensation schemes to construct a multi-constraint feasible solution space, and output the final scheduling instruction set through Pareto optimal iteration.

[0007] Furthermore, the expression for the parking space traffic flow coupling optimization solution algorithm is as follows: ,in For the assembly of ships waiting to berth, For the collection and distribution of vehicles, For ship indexing, For vehicle indexing, This is a weighting factor for the time a ship spends in port. This is a weighting coefficient for vehicle waiting time. For ships The estimated operation time after allocation to a berth. For vehicles Corresponding ships The waiting time for loading and unloading goods, As a binary decision variable, when the vehicle Serving ships The value is 1 if the time is right, otherwise it is 0; the coupling constraint formula for ship waiting time and vehicle waiting time is: ,in For ships The planned unloading volume, For vehicles The corresponding average loading and unloading rate of the quay crane, For vehicles From the entrance to the ship The time taken to travel within the parking lot for the corresponding berth. For vehicles Total length of the on-site path from the entrance to the berth. For path integral variables, For vehicles In path location The instantaneous speed at that location, and and satisfy For vehicles The moment of arrival at the port entrance.

[0008] Furthermore, the expression for the logistics capacity time-series balanced scheduling algorithm is as follows: ,in For discretized time indexing, For the set of transportation nodes, For node indexing, For nodes During the period Available transport capacity For nodes During the period The total amount of goods awaiting shipment;

[0009] The optimized formula for capacity deployment timing is: ,in To predict the duration in advance, For integration time variable, To the node Input the set of upstream nodes of the transport capacity. For upstream node indexing, For time period From upstream node Dispatch to node The increase in transport capacity, For the node The set of downstream nodes that output transportation capacity. For downstream node indexing, For time period From node Distribute to downstream nodes The reduction in transport capacity For the updated capacity capacity in the next time period, For the cumulative deviation, when Increase capacity deployment when Reduce capacity deployment.

[0010] Furthermore, the expression for the logistics node congestion situation assessment model is as follows: ,in For logistics node indexing, This is the end time of the current prediction window. For nodes At any moment Instantaneous cargo density, For nodes Maximum safe packing density, For a moment node The rate of arrival of goods, For a moment node The rate of goods departure, It is a positive constant. As a congestion pressure index, when Mark the node when the threshold is exceeded This is a congested node.

[0011] Furthermore, the multi-constraint feasible solution space includes berth time window constraints, vehicle range constraints, node maximum throughput constraints, and carbon emission quota constraints. The Pareto selection iteration adopts a non-dominated sorting genetic algorithm, with a population size of 150, a crossover probability of 0.85, a mutation probability of 0.12, a maximum number of iterations of 200, and the number of elite individuals retained being 15% of the population size. The final scheduling instruction set is distributed to the port central control system, vehicle-mounted terminals, and node gate control systems via wireless data transmission terminals. The wireless data transmission terminals operate in the 430MHz frequency band and use spread spectrum modulation. The vehicle-mounted terminals have built-in GPS / BeiDou dual-mode positioning modules and inertial navigation units. The node gate control systems are equipped with inductive loops and infrared vehicle separators.

[0012] Furthermore, the calling process of the berth traffic coupling optimization solution algorithm in S2 includes: S21, generating an initial berth occupancy time sequence table based on the expected arrival time of the ships, and allocating available berths to each ship in order of operation priority; S22, extracting the target quay crane number and operation time window corresponding to each ship based on the berth allocation result, and constructing a ternary correlation matrix of ship-quay crane-vehicle; S23, calculating the shortest path and expected travel time of each collection and distribution vehicle from the entrance to the target berth based on the ternary correlation matrix, and updating the cumulative waiting time of each path using dynamic programming; S24, using the weighted sum of the ship's time in port and the vehicle's waiting time as the total cost, and reducing the total cost by exchanging the ship berthing order or vehicle path allocation through neighborhood search, iterating until the total cost no longer decreases.

[0013] Furthermore, the process of generating the capacity deployment time sequence table and vehicle scheduling plan in the logistics capacity time sequence balancing scheduling algorithm in S3 includes: S31, dividing the entire day's transportation period into equally spaced discrete time slots, and counting the tonnage of goods to be transported and the number of available transport vehicles at each transportation node in each time slot; S32, calculating the capacity shortage or surplus of each node in each time slot, and constructing a directed graph of capacity allocation with shortage nodes as the demand side and surplus nodes as the supply side; S33, using the minimum cost flow algorithm on the directed graph to determine the quantity and path of cross-node capacity allocation, and updating the capacity capacity sequence of each node in each time period; S34, allocating the departure time window and vehicle grouping number of each node according to the updated capacity capacity sequence, and generating a vehicle scheduling plan table in units of time slots.

[0014] Furthermore, the process of outputting congestion node marking and congestion level labels in the logistics node congestion situation assessment model in S4 includes: S41, extracting real-time inbound and outbound flow data and current stockpiled cargo volume for each logistics node within the prediction time window, and calculating the instantaneous carrying capacity; S42, multiplying the instantaneous carrying capacity by the inbound and outbound flow ratio to obtain the congestion pressure index, and comparing the index with two preset thresholds of mild and severe to determine the congestion level; S43, for nodes determined to be congested, tracing back the input flow composition of each upstream branch, and identifying the upstream branch with the largest contribution flow as the congestion source; S44, associating the congestion level label and congestion source information with the geographical coordinates of the corresponding node, and generating a congestion situation snapshot with location identification.

[0015] Furthermore, the process of generating the detour energy consumption compensation scheme in S5 includes: S51, extracting the unit mileage energy consumption rate and current idling time of all vehicles on the road within a preset radius around the congestion node from the logistics energy consumption full-domain statistical intelligent analysis system; S52, for each vehicle affected by congestion, calculating the difference between the expected total energy consumption of following the original route and the expected total energy consumption of the detour candidate route; S53, selecting the detour candidate route with the largest energy consumption difference as the recommended detour route for the vehicle, and calculating the energy consumption compensation amount under this route; S54, summarizing the recommended detour routes and energy consumption compensation amounts of all vehicles affected by congestion to form a set of detour energy consumption compensation schemes for congestion nodes.

[0016] An intelligent scheduling method for digital management and control of the entire logistics chain of bulk commodities is proposed. This method is implemented through an intelligent scheduling platform for digital management and control of the entire logistics chain of bulk commodities. It includes: a berth-vehicle flow coupling optimization solution module, which receives port berth operation time-series data and real-time vehicle location data, and outputs berth allocation and vehicle guidance schemes using ship berthing sequence and vehicle routes as decision variables; a logistics capacity time-series balanced scheduling module, connected to the berth-vehicle flow coupling optimization solution module, which receives berth allocation and vehicle guidance schemes and generates a capacity deployment time-series table and vehicle scheduling plan for each transportation node; and a logistics node congestion situation analysis module, connected to the logistics capacity time-series balanced scheduling module, which receives the capacity deployment time-series table and vehicle scheduling plan, calculates the congestion pressure index of each node, and outputs congestion status. The system includes: a level label and congestion source location information; a logistics energy consumption full-domain statistical intelligent analysis module, connected to the logistics node congestion situation judgment module, used to receive congestion level labels and extract real-time energy consumption data of vehicles around congested nodes, and generate detour energy consumption compensation schemes; a multi-constraint feasible solution space construction module, connected to the berth traffic flow coupling optimization solution module, the logistics capacity time-series equilibrium scheduling module, the logistics node congestion situation judgment module, and the logistics energy consumption full-domain statistical intelligent analysis module, used to integrate all output results and apply constraints to construct a feasible solution set; and a Pareto optimal iterative output module, connected to the multi-constraint feasible solution space construction module, used to perform non-dominated sorting and optimal iteration on the feasible solution set, and distribute the final scheduling instructions to the port central control system, vehicle terminals, and node gate control systems.

[0017] Beneficial Effects: This invention proposes an intelligent scheduling method for the digital management and control of the entire logistics chain of bulk commodities. Through a berth-vehicle flow coupling optimization algorithm, it coordinates the decision-making of ship berthing sequence and collection / distribution vehicle routes under the same objective function, enabling dynamic linkage between berth operations and vehicle arrival timing. This overcomes the shortcomings of existing technologies where a sequential decision chain leads to a disconnect between berth plans and capacity plans, and actual berthing fluctuations cause vehicles to wait in vain or berths to remain idle. Furthermore, the logistics capacity time-series balancing scheduling algorithm allocates resources across nodes based on capacity shortages and surpluses at different times, ensuring a close match between capacity deployment and cargo demand in the time dimension, avoiding empty runs or backlogs caused by fixed schedules. The logistics node congestion situation assessment model calculates the congestion pressure of each node in advance. The system uses a force index and assigns levels to trigger route replanning before congestion occurs, overturning the traditional reactive evacuation model. Simultaneously, a comprehensive logistics energy consumption statistics and intelligent analysis system quantitatively compares the increased energy consumption of detour routes with idling energy consumption, selecting the optimal detour strategy based on overall energy consumption. This effectively controls the increase in fuel or electricity consumption while avoiding congestion. These two measures work together to address the pain point of existing technologies failing to achieve a balance between congestion response and energy consumption management. The multi-constraint feasible solution space and Pareto optimal iteration integrate berth allocation, capacity scheduling, congestion handling, and energy compensation into a unified optimization framework, improving the overall operational smoothness and resource utilization efficiency of the entire bulk commodity logistics chain while reducing the energy consumption cost per unit of cargo. Attached Figure Description

[0018] Figure 1 This is a flowchart illustrating the overall process of the method of the present invention.

[0019] Figure 2 This is a flowchart of method step S2 of the present invention;

[0020] Figure 3 This is a flowchart of method step S3 of the present invention;

[0021] Figure 4 This is a flowchart of method step S4 of the present invention;

[0022] Figure 5 This is a flowchart of step S5 of the method of the present invention;

[0023] Figure 6 This is a visualization of the platform interface of the present invention;

[0024] Figure 7 This is a diagram showing the execution of the platform backend code of the present invention. Detailed Implementation

[0025] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. The application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0026] like Figure 1 As shown, the intelligent scheduling method for digital management and control of the entire logistics chain of bulk commodities includes the following steps: S1, collecting port berth operation time-series data, real-time location data of transport vehicles, in-transit inventory, and ship arrival forecast information, and forming a multi-source input feature set after data cleaning and structuring; S2, calling the berth-vehicle flow coupling optimization algorithm, using the ship berthing sequence and the route of collection and distribution vehicles as decision variables, constructing a dual-objective coupling function that minimizes the average ship port time and the average vehicle waiting time, and outputting the berth allocation scheme and vehicle guidance path set; S3, inputting the berth allocation scheme and vehicle guidance path set into the logistics capacity time-series equalization scheduling algorithm, with the objective of minimizing the deviation between the available capacity and the amount of goods to be transported in each transportation period, generating each transportation node. The system generates a capacity deployment schedule and vehicle scheduling plan; S4, based on the capacity deployment schedule and vehicle scheduling plan, it initiates a logistics node congestion situation assessment model, calculates the instantaneous carrying capacity and throughput pressure index of each logistics node within the prediction window, marks congested nodes exceeding the threshold, and outputs congestion level labels; S5, for the marked congested nodes, it calls the logistics energy consumption full-domain statistical intelligent analysis system to extract the real-time energy consumption rate and idling time of transport vehicles around the congested nodes, and generates detour energy consumption compensation schemes in combination with route replanning strategies; S6, it integrates berth allocation schemes, vehicle guidance path sets, capacity deployment schedules, vehicle scheduling plans, congestion level labels, and detour energy consumption compensation schemes to construct a multi-constraint feasible solution space, and outputs the final scheduling instruction set through Pareto optimization iteration.

[0027] The implementation process of step S1 is as follows: When collecting port berth operation time-series data, the actual berthing time, departure time, and corresponding berth number of each ship are obtained through laser berthing instruments and automatic identification system base stations deployed at the front of the berth. The collection frequency is set to record the status change once every 30 seconds. At the same time, the cargo tonnage, length, and operation type of the ships predicted to arrive in the next 4 hours are retrieved from the port database. Real-time positioning data of transport vehicles are obtained through the Beidou positioning terminal installed on each collection and distribution truck. The positioning reporting interval is set to 10 seconds. The data fields include longitude, latitude, instantaneous speed, and driving direction angle. The inventory data in transit is extracted from the inventory management system of each storage node. The current stacking volume and weight of each category are obtained according to the synchronization cycle of once every 5 minutes. The ship arrival forecast information is obtained by connecting to the ship dynamic message system of the maritime department, including the estimated time of arrival at the pilotage anchorage and the estimated berthing window period. When the above four types of raw data enter the data cleaning stage, abnormal trajectory points with positioning drift exceeding 50 meters are removed, duplicate timestamp conflict records are filtered out, and the missing ship cargo volume field is filled with the median of the cargo volume of the three most recent historical voyages. The cleaned structured data is uniformly converted into a multi-source input feature set with timestamp as the primary key for subsequent scheduling steps.

[0028] The implementation process of step S2 is as follows: After importing the multi-source input feature set output from S1 into the berth traffic flow coupling optimization solution algorithm, firstly, extract the remaining operation time of berthed ships in the previous 30 minutes and the number of vehicles queuing at the port entrance in the previous 15 minutes as the initial state parameters of the algorithm; use the berthing order of ships as the first-level decision variable, and the number of quay cranes allocated to each ship and the route selection of collection and distribution vehicles as the second-level decision variables to construct a dual-objective optimization function. The objective of average ship time in port includes two parts: berthing waiting time and quay crane operation time. The objective of average vehicle waiting time includes two parts: entrance queuing time and in-yard driving time. The dual-objective function is solved using a fast non-dominated sorting method with an elitist strategy. The population size is set to... The algorithm consists of 200 individuals, each encoded as a joint chromosome of ship berthing sequence and vehicle path allocation. The crossover operator uses sequential crossover, and the mutation operator uses exchange mutation. The upper limit of the iteration generation is set to 300 generations. Every 50 iterations, the current Pareto front solution set is output. The solution with the minimum weighted sum of ship waiting time and vehicle waiting time is selected as the final output. This weighting coefficient is dynamically adjusted according to the priority given by the port operator; the default weight for ship time is 0.6, and the weight for vehicle time is 0.4. The final output berth allocation scheme includes the berth number assigned to each ship and the expected start and end times of operations. The vehicle guidance path set includes the optimal driving route node sequence from the entrance to the target berth for each vehicle entering the port area.

[0029] The implementation process of step S3 is as follows: After inputting the berth allocation scheme and vehicle guidance path set output from S2 into the logistics capacity time-series balancing scheduling algorithm, the entire 24 hours are first divided into 96 equally spaced time periods, each lasting 15 minutes. The total tonnage of goods awaiting transport and the number of available transport vehicles at each transport node within each time period are calculated. The available transport vehicles are converted into transport capacity tonnage based on a rated load of 30 tons per vehicle. When calculating the capacity deficit at each node in each time period, if the tonnage of goods awaiting transport exceeds 1.2 times the available transport capacity, it is marked as a deficit node; if the available transport capacity exceeds 1.5 times the tonnage of goods awaiting transport, it is marked as a surplus node. The threshold for deficit and surplus is dynamically adjusted based on the average historical capacity utilization rate of the previous 7 days. When constructing the directed graph for capacity allocation, the deficit node is used as the sink node and the surplus node as the source node, and the weights of the directed arcs between nodes are set. The road transport time between the two nodes is set to 30 minutes plus the transfer operation time. The continuous shortest path augmentation algorithm is used to solve the minimum cost maximum flow problem to obtain the number of vehicles transferred from surplus nodes to shortage nodes and the transfer route in each time period. When updating the capacity sequence of each node, the number of vehicles transferred to the target node is added to the available capacity of the target node in the next time period, and the number of vehicles transferred out is deducted from the available capacity of the source node in the current time period. When generating the capacity deployment time sequence table, the number of vehicles to be deployed and the corresponding vehicle number are marked for each node in 96 time periods. When generating the vehicle scheduling plan, the departure time, the order of nodes passed through and the expected arrival time window of each node are assigned to each vehicle according to the capacity deployment time sequence table to ensure that the interval between two adjacent transport tasks of the same vehicle is not less than 30 minutes of loading, unloading and rest buffer time.

[0030] The implementation process of step S4 is as follows: After inputting the capacity deployment time sequence table and vehicle scheduling plan generated in S3 into the logistics node congestion situation analysis model, firstly, determine the prediction time window length as the next 60 minutes from the current moment, and divide the time window into 12 sub-intervals with a step size of 5 minutes; extract the stacked cargo volume and maximum safe stacked volume of each logistics node at the current moment, calculate the instantaneous carrying capacity, and simultaneously collect the total number of vehicles entering and leaving the node in the past 15 minutes as the arrival rate and departure rate, respectively, with the unit of arrival rate and departure rate being vehicles per minute; multiply the instantaneous carrying capacity by the ratio of the arrival rate divided by the sum of the departure rate and the constant 0.01 to obtain the congestion pressure index of the node. A force index value greater than 1.8 is considered mild congestion, and greater than 2.5 is considered severe congestion. These thresholds are dynamically adjusted according to the node type. Specifically, the mild threshold for port-front nodes is set to 2.0, and the severe threshold is set to 3.0, while the mild threshold for inland warehousing nodes is set to 1.5, and the severe threshold is set to 2.2. For nodes identified as congested, the input flow percentage of all their upstream branches in the past 30 minutes is traced back, and branches with a percentage exceeding 40% are marked as congestion sources. After generating a congestion level label and source information for each congested node, the results are associated with the node's latitude and longitude coordinates to form a congestion situation snapshot with spatial location attributes. This snapshot is updated every 5 minutes and cached in local memory for subsequent scheduling steps.

[0031] The implementation process of step S5 is as follows: Based on the congestion level label and congestion source location information output by S4, real-time energy consumption data of all on-the-road transport vehicles within a 5-kilometer radius of the congestion node are extracted from the intelligent analysis system for the whole-domain statistics of logistics energy consumption. The extraction range includes vehicles that are currently in operation and whose geographical distance from the congestion node does not exceed 5 kilometers. The energy consumption data includes the average instantaneous fuel consumption rate or electricity consumption rate of each vehicle in the past 10 minutes and the continuous idling time. The fuel consumption rate is in liters per hour, and the electricity consumption rate is in kilowatt-hours per hour. For each vehicle affected by congestion, the estimated passage time required to pass through the congested area along the original driving route and the corresponding total energy consumption of idling and low-speed driving are first calculated. Then, all feasible detour candidate paths from the vehicle's current location to the next planned node are enumerated. The number of detour candidate routes is limited to no more than 5 to avoid computational overload. For each detour candidate route, its total travel distance and estimated travel time are calculated, and then multiplied by the vehicle's historical average energy consumption rate to obtain the expected total energy consumption of the detour. The expected total energy consumption of each detour candidate route is subtracted from the expected total energy consumption of the original route to obtain the energy consumption difference. The detour route with the smallest negative energy consumption difference is selected as the recommended detour route for the vehicle. If the energy consumption difference of all detour routes is positive, the original route is maintained. The recommended detour routes and corresponding energy consumption compensation amounts of all vehicles affected by congestion are summarized. The energy consumption compensation amount is the positive difference between the total energy consumption of the original route and the total energy consumption of the recommended detour route. The summarized results are compiled into a set of detour energy consumption compensation schemes for the current congested node. Each record in the scheme set includes the vehicle number, the recommended route node sequence, and the compensation energy consumption value.

[0032] The implementation process of step S6 is as follows: The berth allocation scheme and vehicle guidance path set output by S2, the capacity deployment timetable and vehicle scheduling plan output by S3, the congestion level label output by S4, and the detour energy consumption compensation scheme set output by S5 are simultaneously loaded into the scheduling fusion engine. First, a feasible solution space including four types of constraints is constructed. The first type is the berth time window constraint, which requires that the end time of each vessel's operation is no later than its planned departure time plus a 30-minute buffer time. The second type is the vehicle range constraint, which requires that the total mileage of each vehicle throughout the day does not exceed 90% of the maximum range of that vehicle type. The third type is the node throughput constraint, which requires that the cumulative cargo volume handled by each node in any 15-minute period does not exceed 85% of its designed throughput capacity. The fourth type is the carbon emission quota constraint, which requires that the carbon emission converted from the total energy consumption of all vehicles on the day does not exceed The daily quota limit is set; an initial population of 150 individuals is generated within the feasible solution space, with each individual's code including the combination of all decision variables. A non-dominated sorting genetic algorithm is used for multi-objective optimization, where the objective functions include minimizing the total time ships spend in port, minimizing the total waiting time for vehicles, minimizing the total duration of node congestion, and minimizing total energy consumption. The crossover probability is set to 0.85, the mutation probability is set to 0.12, the number of elite individuals retained is 15% of the population size, and the maximum number of iterations is set to 200 generations. Every 50 iterations, the current non-dominated solution set is output to a temporary storage area. After the iteration terminates, the solution with the largest congestion distance is selected from the final Pareto front solution set as the optimal solution. The berth allocation, vehicle route, capacity deployment, scheduling plan, and detour scheme corresponding to the optimal solution are merged to generate the final scheduling instruction set.

[0033] Preferably, the expression for the parking space traffic flow coupling optimization solution algorithm is: ,in For the assembly of ships waiting to berth, For the collection and distribution of vehicles, For ship indexing, For vehicle indexing, This is a weighting factor for the time a ship spends in port. This is a weighting coefficient for vehicle waiting time. For ships The estimated operation time after allocation to a berth. For vehicles Corresponding ships The waiting time for loading and unloading goods, As a binary decision variable, when the vehicle Serving ships The value is 1 if the time is right, otherwise it is 0; the coupling constraint formula for ship waiting time and vehicle waiting time is: ,in For ships The planned unloading volume, For vehicles The corresponding average loading and unloading rate of the quay crane, For vehicles From the entrance to the ship The time taken to travel within the parking lot for the corresponding berth. For vehicles Total length of the on-site path from the entrance to the berth. For path integral variables, For vehicles In path location The instantaneous speed at that location, and and satisfy For vehicles The moment of arrival at the port entrance.

[0034] Specifically, the total cost expression of the berth traffic coupling optimization algorithm is based on the Cartesian product of the ship set and the vehicle set. For each ship-vehicle combination, the weighted sum of the ship's port time and the vehicle's waiting time is calculated and multiplied by a binary decision variable. This binary decision variable takes a value of 1 to indicate that the vehicle serves the ship, and 0 to indicate that it does not serve the ship. The weight coefficient for the ship's port time is 0.6, and the weight coefficient for the vehicle's waiting time is 0.4. The sum of the coefficients is 1 to maintain dimensional consistency. The logic behind this expression stems from the strong coupling relationship between berth scheduling and vehicle routing in the time dimension. That is, the ship berthing plan directly determines the time window for vehicles to perform loading and unloading operations. Therefore, the two are jointly optimized in the same cost function rather than solved independently. The ship's port time is further decomposed into the pure operating time obtained by dividing the planned unloading volume by the average loading and unloading rate of the quay crane, plus the time spent by the vehicle traveling from the entrance to the berth within the yard. The unloading volume is in tons. The loading and unloading rate is measured in tons per hour. The travel time within the yard is obtained by dividing the path length by the instantaneous speed and integrating along the path. This integral form is established because the vehicle speed within the yard is not constant and varies continuously due to the influence of curves, slopes, and other vehicles on port roads. Using integration can accurately accumulate the time increment on each path element. The derivation relationship between the two expressions is that the travel time within the yard is directly added to the ship's time in port. The ship's time in port, in turn, affects the vehicle's waiting time through the waiting time calculation formula. The waiting time is the positive part of the ship's time in port minus the vehicle's arrival time at the entrance. This positive operation ensures the physical rationality of the non-negative waiting time. The implementation of this algorithm is to calculate the joint cost of all ships and vehicles based on the current berth allocation scheme in each iteration, recalculate the cost after changing the ship berthing order or vehicle path allocation through neighborhood search, and select the update scheme with decreasing cost until convergence.

[0035] Preferably, the expression for the logistics capacity time-series balanced scheduling algorithm is: ,in For discretized time indexing, For the set of transportation nodes, For node indexing, For nodes During the period Available transport capacity For nodes During the period The total amount of goods awaiting shipment;

[0036] The optimized formula for capacity deployment timing is: ,in To predict the duration in advance, For integration time variable, To the node Input the set of upstream nodes of the transport capacity. For upstream node indexing, For time period From upstream node Dispatch to node The increase in transport capacity, For the node The set of downstream nodes that output transportation capacity. For downstream node indexing, For time period From node Distribute to downstream nodes The reduction in transport capacity For the updated capacity capacity in the next time period, For the cumulative deviation, when Increase capacity deployment when Reduce capacity deployment.

[0037] Specifically, the total deviation expression of the logistics capacity time-series balancing scheduling algorithm is the sum of the absolute values ​​of the difference between the capacity and the total amount of goods to be transported at each discrete time slot for all transportation nodes. This expression is based on the fundamental principle that the degree of mismatch between supply and demand directly reflects the quality of scheduling balance. The absolute value calculation ensures that both positive surpluses and negative deficits are equally included in the total deviation without offsetting each other. The length of the discrete time slot is set to 15 minutes, with a total of 96 time slots throughout the day. The units for both capacity and total amount of goods to be transported are tons to maintain dimensional consistency. The integral expression for the cumulative deviation integrates the difference between capacity and total amount of goods to be transported over time from the current moment to a preset future duration. This integral form is established because the capacity supply and demand deviation changes continuously within the prediction window rather than only existing at discrete sampling points. The integral can accumulate each time element. The instantaneous deviation is calculated by taking an integral step of 1 second to approximate the accuracy of continuous integration. A positive integral result indicates that there is excess capacity in the future period and subsequent deployment needs to be reduced, while a negative result indicates that there is a capacity shortage and deployment needs to be increased. The capacity update formula expresses the capacity of the next period as the current period capacity plus the sum of the capacity increments from upstream nodes minus the sum of the capacity reductions output to downstream nodes. The derivation relationship between this expression and the integral expression is that the cumulative deviation determines the direction and magnitude of capacity increase or decrease, and the capacity update formula actually performs capacity adjustment based on this magnitude. The two formulas are connected by the sign and magnitude of the cumulative deviation. The adjusted capacity is then used as the input for the next round of deviation calculation to form a recursion. The implementation of this algorithm is to calculate the deviation of each node in a rolling manner according to a 15-minute step and update the capacity until it covers all 96 time slots of the day.

[0038] Preferably, the expression for the logistics node congestion situation assessment model is: ,in For logistics node indexing, This is the end time of the current prediction window. For nodes At any moment Instantaneous cargo density, For nodes Maximum safe packing density, For a moment node The rate of arrival of goods, For a moment node The rate of goods departure, It is a positive constant. As a congestion pressure index, when Mark the node when the threshold is exceeded This is a congested node.

[0039] Specifically, the expression for the logistics node congestion assessment model multiplies the instantaneous carrying capacity by the ratio of the arrival rate to the departure rate plus a constant. This expression is based on the core mechanism that congestion levels are simultaneously affected by both inventory levels and flow imbalances. The instantaneous carrying capacity reflects the degree to which the node's current inventory approaches its safe limit, while the ratio of the arrival rate to the departure rate reflects the net flow input intensity. Multiplying the two results in nodes with high inventory and strong net input receiving the highest congestion pressure index. The instantaneous cargo stacking density is calculated by dividing the current stacked cargo volume of the node by the node's effective storage area. The maximum safe stacking density is set to a maximum of 2.5 tons per square meter, based on the requirements of fire separation distances and fire lane widths. The data collection time window for arrival and departure rates is set to the past 15 minutes, with the rate unit being tons per minute to maintain dimensional consistency. A constant of 0.01 is used to avoid division by zero when the departure rate is zero. Anomaly; This formula does not include a separate derivation prerequisite formula. Its construction is based on the physical laws of congestion evolution, namely, the more goods piled up in a node, the smaller the remaining buffer space. At the same time, when the entry rate is consistently higher than the departure rate, the accumulated goods will accumulate faster. These two factors jointly determine the speed of congestion deterioration. The use of a product form makes the two factors have a multiplicative synergistic effect rather than a simple superposition. When either factor approaches zero, the congestion pressure index approaches zero. When both factors rise simultaneously, the index grows rapidly. This nonlinear response is more in line with the actual characteristics of sudden congestion. The implementation of this model is to calculate the congestion pressure index of all nodes every 5 minutes and compare the calculation results with preset thresholds. The threshold for mild congestion at port front nodes is 2.0 and the threshold for severe congestion is 3.0. The threshold for mild congestion at inland storage nodes is 1.5 and the threshold for severe congestion is 2.2. Once the threshold is exceeded, a congestion flag is immediately triggered and a level label is generated for subsequent scheduling steps.

[0040] Preferably, the multi-constraint feasible solution space includes berth time window constraints, vehicle range constraints, node maximum throughput constraints, and carbon emission quota constraints. The Pareto selection iteration adopts a non-dominated sorting genetic algorithm, with a population size of 150, a crossover probability of 0.85, a mutation probability of 0.12, a maximum number of iterations of 200, and the number of elite individuals retained being 15% of the population size. The final scheduling instruction set is distributed to the port central control system, vehicle-mounted terminals, and node gate control systems via wireless data transmission terminals. The wireless data transmission terminals operate in the 430MHz frequency band and use spread spectrum modulation. The vehicle-mounted terminals have built-in GPS / BeiDou dual-mode positioning modules and inertial navigation units. The node gate control systems are equipped with inductive loops and infrared vehicle separators.

[0041] Preferred, such as Figure 2As shown, the calling process of the berth traffic coupling optimization solution algorithm in S2 includes: S21, generating an initial berth occupancy time sequence table based on the expected arrival time of the ships, and allocating available berths to each ship in order of operation priority; S22, extracting the target quay crane number and operation time window corresponding to each ship based on the berth allocation result, and constructing a ternary correlation matrix of ship-quay crane-vehicle; S23, calculating the shortest path and expected travel time of each collection and distribution vehicle from the entrance to the target berth based on the ternary correlation matrix, and updating the cumulative waiting time of each path using dynamic programming; S24, using the weighted sum of the ship's time in port and the vehicle's waiting time as the total cost, and reducing the total cost by exchanging the ship berthing order or vehicle path allocation through neighborhood search, iterating until the total cost no longer decreases.

[0042] Specifically, the implementation process of the berth traffic coupling optimization algorithm is as follows: In S21, based on the estimated arrival time of the ships, all ships waiting to berth are arranged in order of their estimated arrival time to generate an initial berth occupancy time sequence table. The estimated arrival time is accurate to the minute. For ships with the same estimated arrival time, they are sorted from largest to smallest cargo to determine priority. Available berths with a length not less than the total length of the ship are allocated to each ship in turn. In S22, based on the berth allocation results in S21, the berth number allocated to each ship and the list of quay crane numbers for that berth are extracted. A ternary correlation matrix of ship-quay crane-vehicle is constructed by combining the ship's planned operation time. The row index of the matrix is ​​the ship number, the column index is the vehicle number, and the matrix element value is the sum of the quay crane number corresponding to the ship and the target berth number of the vehicle. Combinatorial encoding; In S23, based on the ternary association matrix, the Dijkstra algorithm is used to calculate the shortest path from the port entrance to the target berth for each transport vehicle. The path weight is the historical average travel time of each segment during peak hours. Then, the estimated travel time is obtained by dividing the path length by the average vehicle speed during that period. The cumulative waiting time of all vehicles on each path is updated recursively. In S24, the estimated travel time of each vehicle in S23 is added to the entrance queuing time to obtain the total waiting time. The total cost is obtained by weighted summation with the berth operation time of the corresponding ship. The simulated annealing algorithm is used for neighborhood search. Swapping the berthing order of two ships or swapping the path assignments of two vehicles are considered neighborhood actions. The initial temperature is set to 100, the cooling coefficient is 0.95, and the termination temperature is set to 1. Iteration continues until the total cost stops decreasing in 50 consecutive neighborhood actions.

[0043] Preferred, such as Figure 3As shown, the process of generating the capacity deployment time sequence table and vehicle scheduling plan in the logistics capacity time sequence balancing scheduling algorithm in S3 includes: S31, dividing the entire day's transportation period into equally spaced discrete time slots, and counting the tonnage of goods to be transported and the number of available transport vehicles at each transportation node in each time slot; S32, calculating the capacity shortage or surplus of each node in each time slot, and constructing a directed graph of capacity allocation with the shortage node as the demand side and the surplus node as the supply side; S33, using the minimum cost flow algorithm on the directed graph to determine the quantity and path of cross-node capacity allocation, and updating the capacity capacity sequence of each node in each time period; S34, allocating the departure time window and vehicle grouping number of each node according to the updated capacity capacity sequence, and generating a vehicle scheduling plan table in units of time slots.

[0044] Specifically, the implementation method of the logistics capacity time-series balancing scheduling algorithm for generating the capacity deployment time sequence table and vehicle scheduling plan is as follows: In S31, the entire 24 hours are divided into 96 equally spaced discrete time slots with 15-minute intervals. The tonnage of cargo to be transported at each transportation node in each time slot is counted from the port operation system and various warehouse management systems. At the same time, the number of available vehicles is calculated based on the number of available vehicles registered at each node multiplied by the rated load of 30 tons per vehicle. In S32, the difference between the tonnage of cargo to be transported and the available capacity is calculated for each node in each time slot. When the difference is positive, it is marked as a node with a shortage and the shortage tonnage is recorded. When the difference is negative, it is marked as a node with a surplus and the surplus tonnage is recorded. With all nodes with shortages as the demand side and all nodes with surpluses as the supply side, a directed graph for capacity allocation is constructed with the road transportation time between nodes plus the 30-minute transfer operation time as the directed arc weight. In S33, the directed graph is constructed using the continuous shortest... The path augmentation algorithm solves for the minimum cost maximum flow. Each augmentation selects the shortest path from the surplus node to the deficit node for flow allocation. The allocation amount is the minimum remaining capacity value on that path. Augmentation is repeated until the demand of all deficit nodes is met or there is no available surplus capacity. The capacity sequence of each node for each time period is updated according to the augmentation results. The allocated capacity is deducted from the available capacity of the surplus node in the current time period and added to the available capacity of the deficit node in the next time period. In S34, based on the updated capacity sequence, a specific departure time window is allocated to each node in each time slot. The start time of the departure time window is the start time of the time slot plus 5 minutes of vehicle preparation time, and the end time is the end time of the time slot minus 10 minutes of vehicle entry buffer time. The number of vehicle groups is determined according to the number of vehicles allocated to each time slot. The vehicle number is associated with the departure time window to generate a vehicle scheduling plan table in units of time slots.

[0045] Preferred, such as Figure 4As shown, the process of outputting congestion node marking and congestion level labels in the logistics node congestion situation assessment model in S4 includes: S41, extracting real-time inbound and outbound flow data and current stockpiled cargo volume for each logistics node within the prediction time window, and calculating the instantaneous carrying capacity; S42, multiplying the instantaneous carrying capacity by the inbound and outbound flow ratio to obtain the congestion pressure index, and comparing the index with two preset thresholds of mild and severe to determine the congestion level; S43, for nodes determined to be congested, tracing back the input flow composition of each upstream branch, and identifying the upstream branch with the largest contribution flow as the congestion source; S44, associating the congestion level label and congestion source information with the geographical coordinates of the corresponding node, and generating a congestion situation snapshot with location identification.

[0046] Specifically, the implementation method for the congestion node marking and congestion level label output process of the logistics node congestion situation assessment model is as follows: In S41, real-time inbound and outbound flow data within the prediction time window are extracted from the automatic identification system of the inbound and outbound gates of each logistics node. Both inbound and outbound flow rates are calculated by dividing the cumulative tonnage of goods in the past 15 minutes by 15 minutes to obtain the rate value. Simultaneously, the instantaneous carrying capacity rate is calculated by dividing the current stockpiled goods volume by the effective storage area of ​​the node from the inventory management system. In S42, the instantaneous carrying capacity rate obtained in S41 is multiplied by the ratio of the inbound rate to the outbound rate plus a constant of 0.01 to obtain the congestion pressure index. This index is compared with preset mild and severe thresholds. The mild threshold for port-front nodes is 2.0, and the severe threshold is 3.0; the mild threshold for inland warehousing nodes is 1.5, and the severe threshold is 2.2. The congestion pressure index is less than... When the threshold for mild congestion is reached, the system is marked as unobstructed. When the threshold is between mild and severe, the system is marked as mildly congested. When the threshold is greater than severe, the system is marked as severely congested. In S43, for nodes identified as congested, the input flow of each upstream branch within the same prediction window is traced back. The percentage of the input flow of each upstream branch relative to the total input flow of the node is calculated. Branches with a percentage exceeding 40% are identified as the congestion source branches contributing the most traffic. The starting node number and the intermediate path segment numbers of the branch are recorded. In S44, the congestion level label output from S42 and the congestion source branch information output from S43 are associated with the geographic coordinates of the corresponding congested node. The geographic coordinates are accurate to 6 decimal places. All data is packaged to generate a congestion status snapshot with location identification. This snapshot includes the node number, congestion level code, source branch number, and coordinate quadruple information.

[0047] Preferred, such as Figure 5As shown, the process of generating the detour energy consumption compensation scheme in S5 includes: S51, extracting the unit mileage energy consumption rate and current idling time of all vehicles on the road within a preset radius around the congestion node from the logistics energy consumption full-domain statistical intelligent analysis system; S52, for each vehicle affected by congestion, calculating the difference between the expected total energy consumption of following the original route and the expected total energy consumption of the detour candidate route; S53, selecting the detour candidate route with the largest energy consumption difference as the recommended detour route for the vehicle, and calculating the energy consumption compensation amount under this route; S54, summarizing the recommended detour routes and energy consumption compensation amounts of all vehicles affected by congestion to form a set of detour energy consumption compensation schemes for congestion nodes.

[0048] Specifically, the detour energy consumption compensation scheme generation process is implemented as follows: In S51, based on the congestion level label and congestion source location information output from S44, the energy consumption records of all on-the-road transport vehicles within a 5-kilometer radius of the congestion node are queried from the real-time energy consumption database of the logistics energy consumption full-domain statistical intelligent analysis system. The radius distance is calculated as the spherical distance between the vehicle's current latitude and longitude and the latitude and longitude of the congestion node. The energy consumption records are the average instantaneous fuel consumption rate or electricity consumption rate and continuous idling time of each vehicle in the past 10 minutes. The fuel consumption rate is in liters per hour, the electricity consumption rate is in kilowatt-hours per hour, and the idling time is in seconds. In S52, for each vehicle affected by congestion, the estimated passage time through the congested area along the original route is calculated. The time spent in idling is multiplied by the idling energy consumption rate, and the time spent in low-speed driving is multiplied by the low-speed driving energy consumption rate. The two are added together to obtain the original route's estimated passage time. The total energy consumption is calculated, and then all feasible detour candidate paths from the vehicle's current location to the next planned node are enumerated. The number of candidate paths is limited to a maximum of 5. The expected total energy consumption of each detour candidate path is calculated by multiplying the path length by the vehicle's average energy consumption rate under smooth conditions. In S53, the difference between the expected total energy consumption of each detour candidate path and the expected total energy consumption of the original path is calculated. The detour path with the largest negative absolute value is selected as the recommended detour path. If all differences are positive, the original path is maintained and no detour is taken. After the recommended detour path is determined, the positive difference between the expected total energy consumption of the original path and the expected total energy consumption of the recommended detour path is calculated as the energy consumption compensation amount. In S54, the recommended detour paths and corresponding energy consumption compensation amounts of all vehicles affected by congestion are summarized one by one. Each summary record includes the vehicle number, the recommended detour path node sequence, and the compensation energy consumption value. These are compiled into a set of detour energy consumption compensation schemes for the current congested node for use in step S6.

[0049] An intelligent scheduling method for digital management and control of the entire logistics chain of bulk commodities is proposed. This method is implemented through an intelligent scheduling platform for digital management and control of the entire logistics chain of bulk commodities. It includes: a berth-vehicle flow coupling optimization solution module, which receives port berth operation time-series data and real-time vehicle location data, and outputs berth allocation and vehicle guidance schemes using ship berthing sequence and vehicle routes as decision variables; a logistics capacity time-series balanced scheduling module, connected to the berth-vehicle flow coupling optimization solution module, which receives berth allocation and vehicle guidance schemes and generates a capacity deployment time-series table and vehicle scheduling plan for each transportation node; and a logistics node congestion situation analysis module, connected to the logistics capacity time-series balanced scheduling module, which receives the capacity deployment time-series table and vehicle scheduling plan, calculates the congestion pressure index of each node, and outputs congestion status. The system includes: a level label and congestion source location information; a logistics energy consumption full-domain statistical intelligent analysis module, connected to the logistics node congestion situation judgment module, used to receive congestion level labels and extract real-time energy consumption data of vehicles around congested nodes, and generate detour energy consumption compensation schemes; a multi-constraint feasible solution space construction module, connected to the berth traffic flow coupling optimization solution module, the logistics capacity time-series equilibrium scheduling module, the logistics node congestion situation judgment module, and the logistics energy consumption full-domain statistical intelligent analysis module, used to integrate all output results and apply constraints to construct a feasible solution set; and a Pareto optimal iterative output module, connected to the multi-constraint feasible solution space construction module, used to perform non-dominated sorting and optimal iteration on the feasible solution set, and distribute the final scheduling instructions to the port central control system, vehicle terminals, and node gate control systems.

[0050] like Figure 6The diagram shows an overview of the visualized interactive interface of the intelligent scheduling platform for digital management and control of the entire logistics chain of bulk commodities, as described in this invention. This interface serves as the human-computer interaction carrier for full-domain scheduling and control, integrating the output results of various functional modules and the full-chain operational status data. A full-domain operational overview dashboard is set at the top of the interface, centrally displaying core indicators such as the number of ships arriving at port today, the number of ships currently in port, the number of vehicle operations today, the congestion pressure index of key nodes, the total energy consumption today, and the completion rate of the work plan. It also indicates the fluctuation range of each indicator compared to yesterday, enabling dispatchers to quickly grasp the overall operational status of the entire chain. A functional navigation bar is located on the left side of the interface, covering functional entry points such as port central control, berth management, capacity scheduling, vehicle management, node gates, energy consumption management, congestion analysis, optimization solutions, and scheduling instructions. These correspond to the core functional modules described in this invention, such as berth-vehicle flow coupling optimization solutions, logistics capacity time-series balanced scheduling, logistics node congestion situation analysis, and intelligent analysis of logistics energy consumption statistics across the entire chain. The central area of ​​the interface sequentially displays the berth operation timeline, congestion node ranking and level distribution, real-time vehicle scheduling status, energy consumption composition statistical analysis, multi-constraint matrix verification results, and a comparison view of the optimal Pareto front solution. This visually displays berth allocation schemes, capacity deployment timing, congestion level labels, detour energy consumption compensation suggestions, and a set of feasible solutions for multi-objective optimization. At the bottom of the interface, a dispatch command issuance control allows dispatchers to select the overall optimal, energy-priority, or efficiency-priority Pareto solution, generate dispatch commands with a single click, and distribute them to the port central control system, vehicle terminals, and node gate control systems, achieving visualized closed-loop management of the entire dispatch decision-making process.

[0051] like Figure 7The diagram shows the backend software system code execution diagram corresponding to the intelligent scheduling method for digital management and control of the entire logistics chain of bulk commodities according to the present invention. The system is developed based on Java language and built on the Spring Boot framework, providing a stable software running platform for the various scheduling algorithms and management logics described in the present invention. The left side of the diagram shows the standardized directory structure of the system project, including engineering components such as the controller layer, service layer, data transmission object layer, and configuration files. The core scheduling controller, DispatchController, serves as the external request entry point of the system, injecting services such as parking space traffic flow optimization, capacity balancing scheduling, congestion analysis, energy consumption analysis, feasible solution space construction, and Pareto optimization, forming a one-to-one mapping relationship with the six functional modules described in the claims of the present invention. The code editing area showcases the core scheduling method logic of the scheduling controller. This method receives scheduling requests carrying port identifiers, berth operation sequences, real-time vehicle location data, and a transportation task list. It sequentially calls the feasible solution space construction unit to complete multi-constraint verification, coordinates various business services to perform berth-vehicle flow coupling solutions and capacity timing scheduling calculations, generates a multi-objective optimal scheduling scheme through the Pareto optimization unit, and finally returns a response result containing a unique scheme identifier, a set of instructions, and the target system's response. The console output at the bottom of the diagram shows that the system application has started successfully, indicating that the backend system can stably support all the scheduling logic of this invention. Through standardized interfaces, it connects to the front-end visualization platform and various terminal devices, achieving fully automated operation from multi-source data input and algorithm iteration to scheduling instruction distribution.

[0052] This intelligent scheduling method for digital management and control of the entire logistics chain of bulk commodities includes a berth-vehicle flow coupling optimization algorithm that places ship berthing sequence and collection and distribution vehicle routes within the same decision-making framework. This enables berth allocation results to drive real-time updates of vehicle guidance routes, eliminating information gaps between berth planning and capacity scheduling. The logistics capacity time-series balanced scheduling algorithm dynamically adjusts the deployment volume based on the capacity shortage at each node and time period, avoiding capacity idleness or backlog caused by fixed shifts. The logistics node congestion situation assessment model identifies risk nodes in advance by calculating the congestion pressure index, transforming passive evacuation into proactive pre-control. The logistics energy consumption full-domain statistical intelligent analysis system quantitatively compares the energy consumption of detour schemes and idling waiting, ensuring that route replanning takes into account both traffic efficiency and energy economy. The cascading of these modules forms a full-chain linkage mechanism from berth to node to transportation terminal, effectively improving the agility of overall scheduling response and the accuracy of resource matching.

[0053] This method addresses the disconnect between berth allocation and capacity planning in traditional sequential decision-making chains. It employs a nested iterative structure of a berth-vehicle flow coupling algorithm and a capacity time-series balancing algorithm, enabling real-time feedback of ship berthing time fluctuations to vehicle arrival schedules. This avoids the simultaneous occurrence of ineffective vehicle waiting and berth idleness, thus resolving operational efficiency losses caused by decision-making delays. Furthermore, addressing the limitations of existing technologies that only address congestion post-congestion evacuation without considering energy costs, this method places congestion situation assessment between capacity deployment and route planning. It pre-marks high-risk nodes and simultaneously calls upon an energy consumption statistics system to evaluate the energy consumption increment of different detour routes. Congestion avoidance measures and energy compensation schemes are output simultaneously, allowing the scheduling strategy to reduce traffic congestion while maintaining transportation energy consumption within a reasonable range. This overcomes the inherent insufficiency of the separation between congestion response and energy consumption management.

[0054] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "set," "install," "connect," "link," and "fix" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal communication between two components. Those skilled in the art will understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0055] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various equivalent changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. An intelligent scheduling method for digital management and control of the entire logistics chain of bulk commodities, characterized in that, Includes the following steps: S1 collects port berth operation time series data, real-time location data of transport vehicles, in-transit inventory and ship arrival forecast information, and forms a multi-source input feature set after data cleaning and structuring; S2 calls the berth traffic flow coupling optimization solution algorithm, which uses the ship berthing order and the collection and distribution vehicle route as decision variables to construct a dual objective coupling function that minimizes the average ship in port time and the average vehicle waiting time, and outputs the berth allocation scheme and vehicle guidance route set. S3 inputs the berth allocation scheme and vehicle guidance path set into the logistics capacity time-series balancing scheduling algorithm. With the goal of minimizing the deviation between the available capacity and the amount of goods to be transported in each transportation period, it generates a capacity deployment time sequence table and vehicle scheduling plan for each transportation node. S4, based on the capacity deployment schedule and vehicle scheduling plan, starts the logistics node congestion situation analysis model, calculates the instantaneous carrying capacity and throughput pressure index of each logistics node within the prediction window, marks congested nodes exceeding the threshold and outputs congestion level labels. S5, for the marked congested nodes, calls the logistics energy consumption full-domain statistical intelligent analysis system to extract the real-time energy consumption rate and idling time of transport vehicles around the congested nodes, and generates a detour energy consumption compensation scheme in combination with the route replanning strategy. S6 integrates berth allocation schemes, vehicle guidance path sets, capacity deployment timetables, vehicle scheduling plans, congestion level labels, and detour energy consumption compensation schemes to construct a multi-constraint feasible solution space, and outputs the final scheduling instruction set through Pareto optimization iteration.

2. The intelligent scheduling method for digital management and control of the entire logistics chain of bulk commodities according to claim 1, characterized in that, The expression for the optimization algorithm for parking space traffic coupling is: ,in For the assembly of ships waiting to berth, For the collection and distribution of vehicles, For ship indexing, For vehicle indexing, This is a weighting factor for the time a ship spends in port. This is a weighting coefficient for vehicle waiting time. For ships The estimated operation time after allocation to a berth. For vehicles Corresponding ships The waiting time for loading and unloading goods, As a binary decision variable, when the vehicle Serving ships The value is 1 if the time is right, otherwise it is 0; the coupling constraint formula for ship waiting time and vehicle waiting time is: ,in For ships The planned unloading volume, For vehicles The corresponding average loading and unloading rate of the quay crane, For vehicles From the entrance to the ship The time taken to travel within the parking lot for the corresponding berth. For vehicles Total length of the on-site path from the entrance to the berth. For path integral variables, For vehicles In path location The instantaneous speed at that location, and and satisfy For vehicles The moment of arrival at the port entrance.

3. The intelligent scheduling method for digital management and control of the entire logistics chain of bulk commodities according to claim 1, characterized in that, The expression for the logistics capacity time-series balanced scheduling algorithm is: ,in For discretized time indexing, For the set of transportation nodes, For node indexing, For nodes During the period Available transport capacity For nodes During the period The total amount of goods awaiting shipment; The optimized formula for capacity deployment timing is: ,in To predict the duration in advance, For integration time variable, To the node Input the set of upstream nodes of the transport capacity. For upstream node indexing, For time period From upstream node Dispatch to node The increase in transport capacity, For the node The set of downstream nodes that output transportation capacity. For downstream node indexing, For time period From node Distribute to downstream nodes The reduction in transport capacity For the updated capacity capacity in the next time period, For the cumulative deviation, when Increase capacity deployment when Reduce transport capacity as needed.

4. The intelligent scheduling method for digital management and control of the entire logistics chain of bulk commodities according to claim 1, characterized in that, The expression for the logistics node congestion situation assessment model is: ,in For logistics node indexing, This is the end time of the current prediction window. For nodes At any moment Instantaneous cargo density, For nodes Maximum safe packing density, For a moment node The rate of arrival of goods, For a moment node The rate of goods departure, It is a positive constant. As a congestion pressure index, when Mark the node when the threshold is exceeded This is a congested node.

5. The intelligent scheduling method for digital management and control of the entire logistics chain of bulk commodities according to claim 1, characterized in that, The multi-constraint feasible solution space includes berth time window constraints, vehicle range constraints, node maximum throughput constraints, and carbon emission quota constraints. The Pareto selection iteration adopts a non-dominated sorting genetic algorithm with a population size of 150, a crossover probability of 0.85, a mutation probability of 0.12, a maximum number of iterations of 200, and retaining 15% of the population as elite individuals. The final scheduling instruction set is distributed to the port central control system, vehicle-mounted terminals, and node gate control systems via wireless data transmission terminals. The wireless data transmission terminals operate in the 430MHz frequency band and use spread spectrum modulation. The vehicle-mounted terminals have built-in GPS / BeiDou dual-mode positioning modules and inertial navigation units. The node gate control systems are equipped with inductive loops and infrared vehicle separators.

6. The intelligent scheduling method for digital management and control of the entire logistics chain of bulk commodities according to claim 1, characterized in that, The calling process of the parking space traffic flow coupling optimization solution algorithm in S2 includes: An initial berth occupancy time sequence table is generated based on the estimated arrival time of ships. Ships are sorted according to their operation priority and then allocated available berths in sequence. Based on the berth allocation results, the target quay crane number and operation time window corresponding to each ship are extracted, and a ternary correlation matrix of ship-quay crane-vehicle is constructed. Based on the ternary correlation matrix, the shortest path and estimated travel time of each collection and distribution vehicle from the entrance to the target berth are calculated, and the cumulative waiting time of each path is updated using dynamic programming. The total cost is calculated by weighting the ship's time in port and the vehicle's waiting time. The total cost is then reduced by exchanging ship berthing order or vehicle route allocation through neighborhood search. This process is iterated until the total cost no longer decreases.

7. The intelligent scheduling method for digital management and control of the entire logistics chain of bulk commodities according to claim 1, characterized in that, The process of generating the capacity deployment time series table and vehicle scheduling plan in the logistics capacity time series balancing scheduling algorithm in S3 includes: The entire day's transportation period is divided into equally spaced discrete time slots, and the tonnage of goods to be transported and the number of available transport vehicles at each transportation node within each time slot are counted. Calculate the capacity deficit or surplus of each node in each time slot, and construct a directed graph of capacity allocation with deficit nodes as the demand side and surplus nodes as the supply side. Use the minimum cost flow algorithm on the directed graph to determine the quantity and path of cross-node capacity allocation, and update the capacity sequence of each node for each time period. Based on the updated capacity sequence, allocate the departure time window and vehicle grouping number of each node, and generate a vehicle scheduling plan in time slots.

8. The intelligent scheduling method for digital management and control of the entire logistics chain of bulk commodities according to claim 1, characterized in that, The process of outputting congestion node marking and congestion level labels in the S4 logistics node congestion situation assessment model includes: Extract real-time inbound and outbound flow data and current stocked cargo volume for each logistics node within the prediction window to calculate the instantaneous carrying capacity; multiply the instantaneous carrying capacity by the inbound and outbound flow ratio to obtain the congestion pressure index; compare the index with two preset thresholds, mild and severe, to determine the congestion level. For nodes identified as congested, the input traffic composition of their upstream branches is traced back, and the upstream branch contributing the most traffic is identified as the source of congestion. The congestion level label and the information on the source of congestion are associated with the geographical coordinates of the corresponding node to generate a snapshot of the congestion situation with location identifiers.

9. The intelligent scheduling method for digital management and control of the entire logistics chain of bulk commodities according to claim 1, characterized in that, The generation process of the detour energy consumption compensation scheme in S5 includes: Extract the unit mileage energy consumption rate and current idling time of all vehicles on the road within a preset radius around the congestion node from the logistics energy consumption full-domain statistical intelligent analysis system; For each vehicle affected by congestion, calculate the difference between its expected total energy consumption by following the original route and the expected total energy consumption by taking the alternative candidate route. The candidate detour route with the largest energy consumption difference is selected as the recommended detour route for the vehicle, and the energy consumption compensation amount under this route is calculated. The recommended detour routes and energy consumption compensation amounts for all vehicles affected by congestion are compiled to form a set of detour energy consumption compensation schemes for congested nodes.

10. The intelligent scheduling method for digital management and control of the entire logistics chain of bulk commodities according to any one of claims 1-9, characterized in that, This method is implemented through an intelligent scheduling platform for digital management and control of the entire logistics chain of bulk commodities. It includes: a berth-vehicle flow coupling optimization solution module, which receives port berth operation time-series data and real-time vehicle location data, and outputs berth allocation and vehicle guidance schemes using ship berthing sequence and vehicle route as decision variables; a logistics capacity time-series balanced scheduling module, connected to the berth-vehicle flow coupling optimization solution module, which receives berth allocation and vehicle guidance schemes and generates a capacity deployment time-series table and vehicle scheduling plan for each transportation node; and a logistics node congestion situation analysis module, connected to the logistics capacity time-series balanced scheduling module, which receives the capacity deployment time-series table and vehicle scheduling plan, calculates the congestion pressure index for each node, and outputs congestion level labels and congestion source location information. The system includes: a logistics energy consumption full-domain statistical intelligent analysis module, connected to the logistics node congestion situation assessment module, used to receive congestion level labels and extract real-time energy consumption data of vehicles around congested nodes, generating detour energy consumption compensation schemes; a multi-constraint feasible solution space construction module, connected to the berth traffic flow coupling optimization solution module, the logistics capacity time-series equilibrium scheduling module, the logistics node congestion situation assessment module, and the logistics energy consumption full-domain statistical intelligent analysis module, used to integrate all output results and apply constraints to construct a feasible solution set; and a Pareto optimal iteration output module, connected to the multi-constraint feasible solution space construction module, used to perform non-dominated sorting and optimal iteration on the feasible solution set, and distribute the final scheduling instructions to the port central control system, vehicle terminals, and node gate control systems.

Images