Transportation special line carrying optimization method and system based on intelligent scheduling

Abstract

The invention provides a transportation special line carrying optimization method and system based on intelligent scheduling. The method comprises the following steps: constructing a dynamic timeliness graph model and carrying out real-time data acquisition; path planning and task allocation are carried out based on the dynamic timeliness graph model; designing a micro-service architecture based on a distributed architecture; and a self-learning mechanism is designed, automatic adjustment is carried out according to historical tasks and real-time feedback, and the carrying task of the special transportation line is optimized. According to the method, graph theory optimization, a micro-service architecture and an intelligent self-learning algorithm are introduced, so that the problems of low efficiency, low flexibility, poor expansibility and insufficient real-time scheduling capability in a traditional transportation special line scheduling method are effectively solved.

Description

Technical Field

[0001] The present invention belongs to the field of intelligent scheduling, and in particular relates to a transportation line handling optimization method and system based on intelligent scheduling. Background Art

[0002] With the vigorous development of the modern logistics industry, dedicated transport lines have become one of the key links. Dedicated transport systems play an important role in the transportation of various materials and commodities, especially in large-scale, scheduled, and fixed-point cargo transportation. Traditional dedicated transport line scheduling methods usually rely on manual scheduling, empiricism, or scheduling algorithms based on fixed rules. Although these methods guarantee the completion of transportation tasks to a certain extent, they face many limitations and challenges. In particular, in the face of increasingly complex transportation networks, frequently changing transportation tasks, increasing scheduling needs, and the diversity of transportation resources, traditional solutions often find it difficult to cope with these challenges.

[0003] First, the inefficiency and low flexibility of traditional scheduling methods are one of the biggest problems. Existing dedicated transport line scheduling systems are often scheduled in a centralized manner, and most of them adopt static scheduling methods based on rules or heuristic algorithms, which lack flexibility and real-time performance. This results in the inability of the scheduling system to respond quickly and make dynamic adjustments when traffic congestion, sudden weather or vehicle failures occur in the transportation network, thereby increasing the risk of transportation delays. In addition, most current systems fail to fully utilize advanced technologies such as big data, cloud computing and artificial intelligence. Scheduling decisions mainly rely on historical data and qualitative analysis, and are unable to comprehensively optimize and dynamically predict transportation tasks.

[0004] Secondly, the scalability and modularity issues of the system also limit the further development of traditional transportation dedicated line scheduling systems. With the continuous expansion of the scale of the transportation network and the diversification of the number of transportation tasks and resource types, traditional scheduling systems usually adopt a single large-scale centralized architecture. This architecture is difficult to expand flexibly when faced with complex and dynamically changing task requirements. For example, during the peak period of cargo transportation, adding new scheduling resources or adjusting task allocation often requires modifying the entire system, which is not only inefficient but also increases the maintenance cost of the system. In addition, the problem of information islands in transportation scheduling is also relatively serious. It is difficult to achieve efficient coordination of information from different transportation links and different data sources, resulting in scheduling decisions that cannot reflect the current transportation status in real time and comprehensively.

[0005] Third, the limitations of scheduling task optimization. At present, most transportation scheduling systems rely only on static path selection and resource allocation rules, lacking dynamic optimization of complex time and space constraints. For example, in some special cases, the selection of transportation routes not only needs to consider factors such as time and distance, but also must comprehensively consider real-time information such as weather, road conditions, and traffic flow. If the system cannot make adjustments in time, it may lead to waste of resources or transportation delays, seriously affecting transportation efficiency and service quality.

[0006] In response to the above problems, existing technologies generally face the problem of how to improve the efficiency, flexibility and scalability of transportation line scheduling through innovative algorithms, advanced architectures and the coordination of real-time data streams. Although transportation scheduling systems based on artificial intelligence, the Internet of Things, cloud computing and other technologies have gradually begun to emerge in recent years, these systems still have a lot of room for improvement in terms of overall architecture flexibility, intelligent real-time scheduling and optimization capabilities of large-scale transportation networks. Summary of the invention

[0007] The purpose of the present invention is to propose a transport line handling optimization method and system based on intelligent scheduling, which effectively overcomes the problems of low efficiency, low flexibility, poor scalability and insufficient real-time scheduling capability in traditional transport line scheduling methods by introducing graph theory optimization, microservice architecture and intelligent self-learning algorithm.

[0008] In order to achieve the above object, a first aspect of the present invention provides a transportation line handling optimization method based on intelligent scheduling, the method comprising:

[0009] S1. Construct a dynamic time-efficiency graph model and collect real-time data; wherein the structure of the dynamic time-efficiency graph model is as follows: each node represents a key location, including a starting point, an end point or a transfer point, and contains location information and key attributes of the task; each edge represents a transportation path, and the weight of the edge reflects the timeliness of the path; the update mechanism of the dynamic time-efficiency graph model uses a sliding window mechanism to update the weight of the edge;

[0010] S2. Perform path planning and task allocation based on a dynamic time-effectiveness graph model; wherein the objective function of the path planning and task allocation is to minimize the total transportation cost and satisfy the task time window and resource constraints. The total transportation cost includes time cost and resource allocation cost. The objective function of the total transportation cost is:

[0011]

[0012] Where C is the total transportation cost, w ij For the edge (v i ,v j ) represents the path cost, x ij∈{0,1} is a decision variable, indicating that the edge (v i ,v j ) is selected; φ r Assign costs to resources, combining the empty load rate and allocation efficiency of each transportation resource;

[0013] S3. Design a microservice architecture based on a distributed architecture, split functions into different modules, and different microservices are responsible for independent functions. Each microservice will dynamically adjust the execution strategy based on the output of the dynamic timeliness graph model and real-time monitoring data, and in combination with the real-time monitoring data. If a delay or idle resources occur during task execution, the scheduling service of the microservice architecture needs to dynamically adjust the task allocation strategy according to the deviation time of the task; wherein, the functions are split into different modules, including: path planning service, task scheduling service, real-time monitoring service and adaptive scheduling adjustment service;

[0014] The path planning service is used to calculate and dynamically adjust the path selection of the task;

[0015] The task scheduling service is used to schedule tasks according to the path planning results and real-time data;

[0016] The real-time monitoring service is used to monitor the status changes during task execution in real time and provide feedback;

[0017] The adaptive scheduling adjustment service is used to automatically adjust the resource allocation strategy according to real-time feedback and task status;

[0018] S4. Design a self-learning mechanism to automatically adjust according to historical tasks and real-time feedback to optimize the transportation line handling tasks.

[0019] Furthermore, the weights of the edges of the dynamic temporal graph model are calculated as follows:

[0020] w ij =α·D ij +β·T ij +γ·C ij +δ·W ij +λ·f ij

[0021] Among them, D ij is the distance between nodes i and j, T ij is the estimated time of the transportation route, C ij is the impact of traffic flow on the path, W ij is the impact of weather on the path, f ij is an additional path constraint, α, β, γ, δ, and λ are adjustment parameters that adjust the impact of different factors on the weights;

[0022] The sliding window mechanism is used to update the edge weights, which is expressed as:

[0023] Updated ij (t) = w ij (t-Δt)+Δw ij (t)

[0024] Among them, w ij (t) represents the edge weight at time t, Δt is the time interval, Δw ij (t) is the edge weight change caused by real-time data flow.

[0025] Furthermore, the S1 also includes optimizing the sensor data using Kalman filtering, which is expressed as:

[0026]

[0027] in, represents the estimated value of the state (such as position, speed, etc.) at time k, z k is the measured value at time k, K k is the Kalman gain, H k is the measurement matrix, which represents the relationship between measurement and state.

[0028] Furthermore, the total transportation cost is also subject to constraints, including:

[0029] Each task node must be visited exactly once:

[0030]

[0031] Where V is the set of nodes in the graph;

[0032] All tasks must meet the time window constraints:

[0033]

[0034] Where T i is the task start time, T start,i and T end,i The earliest and latest time for the task;

[0035] The load of a transport resource must not exceed its capacity:

[0036]

[0037] where d i is the task demand, C r is the maximum load of resource r.

[0038] Furthermore, the S2 also includes the joint optimization of path planning and task allocation, specifically:

[0039] According to the task priority and time window constraints, an initial access sequence is generated, where nodes with high priority and tight time windows are accessed first;

[0040] Use the dynamic time-effective graph model for path planning, and for each task node v i , choose the weight w ij The smallest side (v i ,v j );

[0041] Prioritize each transport resource r to the task node v that best matches its current status i ;

[0042] Introduce an adjustment mechanism to recalculate w through a sliding window ij and x ij , dynamically update paths and task allocations.

[0043] Furthermore, the sorting strategy of the access sequence is:

[0044]

[0045] Among them, S is the access sequence, argsort is the sorting function, V is the total number of nodes, T end,i is the end window, T start,i is the start window;

[0046] The dynamic path selection rule for path planning by the dynamic time-effectiveness graph model is:

[0047]

[0048] Among them, f ij is an additional path constraint, λ is a regularization coefficient used to adjust the importance of the additional constraint, x i t j is the initial path;

[0049] The method allocates each transport resource r to the task node v that best matches its current state. i , specifically:

[0050] Based on the distance D between the current location of the resource and the location of the task r , no-load rate η r and task requirements i , computing resource adaptation, expressed as:

[0051]

[0052] Select φ r Minimal resources to perform tasks;

[0053] The sliding window recalculates w ij and x ij , dynamically update the path and task allocation, specifically:

[0054] In each sliding window period, the updated path selection rule is:

[0055]

[0056] in, For the updated path, is the initial path, Δx ij Path changes caused by real-time data flow.

[0057] Furthermore, the S3 specifically includes:

[0058] Adjust the path weights based on real-time traffic information and calculate the actual weights of each path in real time The adjusted path planning formula is:

[0059]

[0060] Among them, w ij is the initial path weight, traffic ij For the path (v i ,v j ) real-time traffic information, weather ij For the path (v i ,v j Weather influence, time of day ij is the real-time period factor on the path, λ1,λ2,λ3 are weight coefficients, which adjust the impact of real-time information on path selection;

[0061] If a delay or idle resources occurs during task execution, the scheduling service of the microservice architecture needs to dynamically adjust the task allocation strategy according to the deviation time of the task, including:

[0062] First, define the task deviation time ΔT i :

[0063] ΔT i =T actual,i -T planned,i

[0064] Among them, T actual,i represents the actual execution time of task i, T planned,i represents the planned execution time of task i, ΔTi represents the execution deviation time of task i.

[0065] When the task deviation is too large, the scheduling algorithm will automatically adjust the resource allocation of the task and select the resource r with a closer distance and more suitable status to reschedule the task. The adjustment strategy is as follows:

[0066]

[0067] in, For the updated strategy, dist r is the distance between resource r and task i, avail r is the current availability of resource r, priority r is the priority of resource r, reflecting its applicable priority requirements, and β1 and β2 are adjustment weights that determine the impact of distance and resource priority on scheduling decisions.

[0068] Furthermore, the S3 also includes that the real-time monitoring service detects the state changes during the task execution process in real time. If a deviation or anomaly is detected, the real-time monitoring service will pass the real-time monitoring data to the task scheduling service, triggering the task scheduling service to make dynamic adjustments, which is specifically expressed as follows:

[0069] ΔT i =T actual,i -T planned,i

[0070] Where, ΔT i is the deviation time of each task, T actual,i is the actual time, T planned,i is the planned time, if ΔT i >δ i , then the task rescheduling is triggered through the scheduling system;

[0071] Monitor the usage status of each resource, expressed as:

[0072]

[0073] Among them, T assigned,r is the allocated working time of resource r, T total,r is the total working time (maximum available time) of resource r. When the resource load exceeds the set threshold, L r >τ r , it triggers resource reallocation to avoid overuse of a certain resource;

[0074] At the same time, in the microservice architecture, dynamic adjustment is performed through the load balancing strategy to ensure that the response time of each service does not exceed the predetermined threshold. When a service overload is detected, it automatically switches to the backup service to ensure that the entire scheduling process is not interrupted; wherein, the load balancing strategy is expressed as:

[0075]

[0076] Among them, response time s is the response time of service s, τ s is the threshold of response time, Status s The current status of the service.

[0077] Furthermore, the S4 specifically includes:

[0078] According to the deviation time of each task and the change in resource load, the deviation of the scheduling model is calculated, and the historical feedback data is combined for learning. The original path planning and task scheduling strategy are adjusted according to the deviation of the scheduling model, so that the scheduling result tends to be optimized, which is expressed as:

[0079]

[0080] Where, ΔT i is the execution time deviation of task i, L r is the current load of resource r, α, β, γ are the impact coefficients of scheduling error on path selection and task allocation strategy, T actual,i and T planned,i are the actual execution time and planned execution time of task i, respectively, and w ij is the original path weight, η1 is the sensitivity factor of path adjustment, ΔM i is the scheduling error output by the error model, and α is a hyperparameter that controls the impact of the error;

[0081] Design dynamic task adjustment strategies, combine the optimization results of path planning, and reallocate resources, including:

[0082] If the actual execution time of the task exceeds the expected deviation range ΔT i , then through the rescheduling mechanism, the most suitable resources are selected for task reallocation. When reallocating, the task priority i Adjustments will be made based on the urgency and importance of the task. The formula for task rescheduling is:

[0083]

[0084] Among them, dist r is the distance between resource r and task i, avail ris the availability of resource r, priority r is the priority of resources, γ is the impact coefficient of feedback error on resource scheduling;

[0085] Collect user execution feedback data, update the scheduling error output by the error model through incremental learning, and dynamically update path planning and task scheduling.

[0086] In a second aspect of the present invention, a dedicated transport line handling optimization system based on intelligent scheduling is provided, the system comprising:

[0087] A dynamic timeliness graph model unit is used to construct a dynamic timeliness graph model and perform real-time data collection; wherein the structure of the dynamic timeliness graph model is as follows: each node represents a key location, including a starting point, an end point or a transfer point, and contains location information and key attributes of the task; each edge represents a transportation path, and the weight of the edge reflects the timeliness of the path; the update mechanism of the dynamic timeliness graph model adopts a sliding window mechanism to update the weight of the edge;

[0088] The path planning and task allocation unit is used to perform path planning and task allocation based on a dynamic time-effectiveness graph model; wherein the objective function of the path planning and task allocation is to minimize the total transportation cost and satisfy the task time window and resource constraints. The total transportation cost includes time cost and resource allocation cost. The objective function of the total transportation cost is:

[0089]

[0090] Where C is the total transportation cost, w ij For the edge (v i ,v j ) represents the path cost, x ij ∈{0,1} is a decision variable, indicating that the edge (v i ,v j ) is selected; φ r Assign costs to resources, combining the empty load rate and allocation efficiency of each transportation resource;

[0091] A microservice architecture construction unit is used to design a microservice architecture based on a distributed architecture, split functions into different modules, and different microservices are responsible for independent functions. Each microservice will dynamically adjust the execution strategy based on the output of the dynamic timeliness graph model and real-time monitoring data, and in combination with the real-time monitoring data. If a delay or idle resources occur during the task execution process, the scheduling service of the microservice architecture needs to dynamically adjust the task allocation strategy according to the deviation time of the task; wherein, the functions are split into different modules, including: path planning service, task scheduling service, real-time monitoring service and adaptive scheduling adjustment service;

[0092] The path planning service is used to calculate and dynamically adjust the path selection of the task;

[0093] The task scheduling service is used to schedule tasks according to the path planning results and real-time data;

[0094] The real-time monitoring service is used to monitor the status changes during task execution in real time and provide feedback;

[0095] The adaptive scheduling adjustment service is used to automatically adjust the resource allocation strategy according to real-time feedback and task status;

[0096] The self-learning optimization unit is used to design a self-learning mechanism to automatically adjust according to historical tasks and real-time feedback to optimize the transportation line handling tasks.

[0097] The beneficial technical effects of the present invention are at least as follows:

[0098] First, the present invention adopts a graph optimization algorithm. By modeling the transport line as a weighted directed graph, it can comprehensively consider various time and space constraints, resource allocation, path selection and other multi-dimensional factors in the transport network, and use a dynamic time-effective graph algorithm to adjust the transport path and task allocation in real time. This method not only solves the problem of static path selection in the existing scheduling system, but also can respond to emergencies such as traffic congestion and weather changes in real time, realize dynamic optimization of transport tasks, and significantly improve transport efficiency and reliability.

[0099] In order to solve the scalability and flexibility problems of the existing centralized system architecture, the present invention adopts a microservice architecture design. Each transportation scheduling task is handled by an independent microservice module, including multiple core modules such as path planning, resource management, scheduling execution, and real-time monitoring. These modules work together through message queues, so that the system can quickly adjust and expand as needed, and respond to changes in transportation tasks in real time. For example, when the demand for transportation surges, the system can quickly add new scheduling modules to share tasks, thereby maintaining scheduling efficiency. In addition, the microservice architecture also effectively solves the problem of information islands, and uses API interfaces to efficiently flow and coordinate information between different modules, ensuring the real-time and comprehensiveness of decision-making.

[0100] Finally, the present invention introduces self-learning algorithms such as reinforcement learning, so that the scheduling system can continuously learn from historical data and real-time execution results, and gradually optimize scheduling decisions and resource allocation strategies. The feedback data after each transportation task is completed will be used as training data to continuously adjust the scheduling strategy to adapt to the diverse needs of transportation tasks. Through this self-learning mechanism, the system can achieve dynamic self-optimization, gradually improve scheduling efficiency, and avoid inefficient scheduling problems caused by artificial rules or lack of experience.

[0101] In summary, the present invention effectively overcomes the problems of low efficiency, low flexibility, poor scalability and insufficient real-time scheduling capabilities in traditional dedicated transport line scheduling methods by introducing graph theory optimization, microservice architecture and intelligent self-learning algorithms. The system can not only provide efficient, dynamic and intelligent transportation paths and task scheduling solutions, but also ensure the efficient operation and flexible expansion of the system in a large-scale transportation network, greatly improving the overall efficiency and responsiveness of dedicated transport line handling. Therefore, the dedicated transport handling optimization system of the present invention is not only innovative in technology, but also can significantly improve the actual application effect of the existing system. BRIEF DESCRIPTION OF THE DRAWINGS

[0102] The present invention is further described using the accompanying drawings, but the embodiments in the accompanying drawings do not constitute any limitation to the present invention. A person skilled in the art can obtain other drawings based on the following drawings without creative work.

[0103] Figure 1 The present invention is a flow chart of a method for optimizing transportation line handling based on intelligent scheduling.

[0104] Figure 2 This is a framework diagram of a dedicated transport line handling optimization system based on intelligent scheduling according to the present invention. DETAILED DESCRIPTION

[0105] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be understood as limiting the present invention.

[0106] like Figure 1 As shown, an embodiment of the present invention provides a transportation line handling optimization method based on intelligent scheduling, and the method includes the following:

[0107] S1. Build a dynamic time-effectiveness graph model and collect real-time data.

[0108] Specifically, we collect graph data to represent the network structure of the transportation task. The nodes in the graph represent the key locations of the transportation task (such as the starting point, the end point, and the transfer point), and the edges represent the transportation paths connecting these locations. Each edge contains weight information, reflecting timeliness factors such as estimated transportation time, traffic conditions, etc.

[0109] Among them, real-time data stream (D t ) contains real-time data from IoT sensors, GPS devices, traffic monitoring systems, etc. These data mainly include traffic flow, weather conditions, equipment status, etc. The input of real-time data streams is the key to continuously updating the graph model.

[0110] Furthermore, the dynamic time-effectiveness graph model is constructed as follows:

[0111] Each node represents a key location, such as a starting point, an end point, or a transfer point, and contains location information and other key attributes of the task, such as priority, time window, etc.

[0112] Node attributes can be expressed as: i =(x i ,y i ,P i ,T window ,U i )

[0113] Among them, v i Represents the location information of node i (x i ,y i ), P i is the task priority, T window is the time window of the task (earliest start and latest end time), U i The availability of the node indicates whether the device is available or the path is unobstructed.

[0114] Each node represents a key location, such as a starting point, an end point, or a transfer point, and contains location information and other key attributes of the task, such as priority, time window, etc.

[0115] Node attributes can be expressed as: i =(x i ,y i ,P i ,T window ,U i ), where v i Represents the location information of node i (x i ,y i ), P i is the task priority, T window is the time window of the task (earliest start and latest end time), U i The availability of the node indicates whether the device is available or the path is unobstructed.

[0116] Furthermore, each edge represents a transportation route, and the weight of the edge reflects the timeliness of the route. In addition to distance and time, the weight also takes into account real-time data such as traffic flow and weather.

[0117] The weight of the edge w ij The calculation formula is as follows:

[0118] w ij =α·D ij +β·T ij +γ·Cij +δ·W ij +λ·f ij

[0119] Among them, D ij is the distance between nodes i and j, T ij is the estimated time of the transportation route, C ij is the impact of traffic flow on the path (such as traffic congestion), W ij is the impact of weather on the route (such as delays caused by bad weather), f ij are additional path constraints, such as equipment availability, traffic restrictions on special road sections, etc. α, β, γ, δ, and λ are adjustment parameters that adjust the impact of different factors on the weights.

[0120] As you can understand, an additional constraint f is added to the edge weight calculation ij , allowing the model to flexibly respond to special needs in transportation tasks, such as equipment availability or safety factors.

[0121] Furthermore, the weight of the timeliness graph is dynamically updated as real-time data streams are input. In order to ensure the real-time and accuracy of the graph, a sliding window mechanism is used to update the weight of the edge.

[0122] The sliding window update formula is:

[0123] Updated ij (t) = w ij (t-Δt)+Δw ij (t)

[0124] Among them, w ij (t) represents the edge weight at time t, Δt is the time interval, Δw ij 9t) is the edge weight change caused by real-time data flow. The sliding window mechanism can ensure that the system responds to environmental changes in real time within each time window and ensure the timeliness of the data.

[0125] Furthermore, in order to ensure the accuracy of real-time data, Kalman filtering is used to optimize sensor data. Kalman filtering helps reduce noise and improve data accuracy, especially in complex transportation scheduling scenarios.

[0126] The core formula of Kalman filtering is:

[0127]

[0128] in, represents the estimated value of the state (such as position, speed, etc.) at time k, z k is the measured value at time k (data from the sensor), K kis the Kalman gain, H k is the measurement matrix, which represents the relationship between measurement and state. The Kalman filter algorithm not only processes basic sensor data, but also integrates a variety of external environmental data (such as traffic and weather) to provide more accurate real-time data for the timeliness graph.

[0129] Finally, the output dynamic time-effectiveness graph (G t ) includes the latest status of all nodes and edges. The edge weight reflects multiple factors such as real-time traffic, weather, and equipment status. This graph can be directly used for subsequent path planning and task scheduling. The key point of dynamic timeliness graph update is the adjustment of edge weights to ensure that each transportation task makes the best decision based on the latest real-time data.

[0130] Through these steps, the present invention not only realizes a time-sensitive graph model that can respond to changes in transportation tasks in real time, but also can perform efficient and accurate path planning and task scheduling on this basis, solving the bottleneck problem of traditional methods that cannot be adjusted in real time.

[0131] S2. Path planning and task allocation based on dynamic time-effectiveness graph model.

[0132] Specifically, based on the dynamic timeliness graph G t , perform path planning and task allocation at the same time, minimize the total transportation cost C (including time cost and resource allocation cost), and meet the task time window and resource constraints. Objective function:

[0133]

[0134] Among them, w ij For the edge (v i ,v j ) represents the path cost, x ij ∈{0,1} is a decision variable, indicating that the edge (v i ,v j ) is selected, φ r Assign costs to resources, taking into account the empty load rate and allocation efficiency of each transportation resource.

[0135] Furthermore, the constraints are:

[0136] Each task node must be visited exactly once:

[0137]

[0138] Where V is the set of nodes in the graph.

[0139] All tasks must meet the time window constraints:

[0140]

[0141] Among them, T i is the task start time, T start,i and T end,i The earliest and latest time for the task.

[0142] The load of a transport resource must not exceed its capacity:

[0143]

[0144] Among them, d i is the task demand, C r is the maximum load of resource r.

[0145] Furthermore, the node priority is sorted:

[0146] According to the priority of the task P i and the time window constraint T window,i , Generate the initial access sequence. Give priority to visiting nodes with high priority and tight time windows.

[0147] Sorting strategy:

[0148]

[0149] Furthermore, path selection and allocation optimization:

[0150] Use dynamic time-efficiency graph model for path planning. For each task node v i , choose the weight w ij The smallest side (v i ,v j ).

[0151] Define dynamic path selection rules:

[0152]

[0153] Among them, f ij is an additional path constraint (such as road condition changes and equipment limitations), and λ is a regularization coefficient used to adjust the importance of additional constraints.

[0154] Furthermore, the resource allocation rules are:

[0155] Each transport resource r is preferentially allocated to the task node v that best matches its current status i . Based on the distance D between the current location of the resource and the location of the task r , no-load rate η r and task requirements i , calculation resource adaptability:

[0156]

[0157] Select φ r Minimal resources to perform tasks.

[0158] Furthermore, in order to cope with real-time data changes (such as traffic jams, new tasks, etc.), an adjustment mechanism is introduced after the initial planning. ij and x ij , dynamically update paths and task allocations.

[0159] In each sliding window period, the updated path selection rule is:

[0160]

[0161] Where Δx ij Path changes caused by real-time data flow.

[0162] Furthermore, the optimized path planning scheme and task allocation results are output, including the selection decision x for each path ij and the task allocation strategy φ for each resource r The results are directly used as input for subsequent steps (scheduling execution and monitoring) to ensure efficient resource scheduling and optimal path selection.

[0163] This step realizes the joint optimization of path planning and task allocation, making full use of the weight information of the dynamic timeliness graph and the real-time status of transportation resources. The introduction of priority sorting, dynamic path selection rules and resource adaptation optimization methods solves the problem of separation of task allocation and path planning in traditional methods, and provides the best decision basis for subsequent task execution.

[0164] S3. Design a microservice architecture based on a distributed architecture, split the functions into different modules, and different microservices are responsible for independent functions. Each microservice will dynamically adjust the execution strategy based on the output of the dynamic timeliness graph model and real-time monitoring data, and in combination with real-time monitoring data. If delays or idle resources occur during task execution, the scheduling service of the microservice architecture needs to dynamically adjust the task allocation strategy according to the deviation time of the task.

[0165] Specifically, the core of this step is to manage path planning and task scheduling execution under the microservice architecture. By splitting different modules, the system is ensured to have high flexibility, scalability and dynamic response capabilities. Different microservices are responsible for independent functional modules, including path planning, task allocation, execution monitoring and adjustment. Each microservice will dynamically adjust the execution strategy based on the output of the previous step and real-time monitoring data.

[0166] Path Planning Service (S path ): Responsible for calculating and dynamically adjusting the path selection of the task.

[0167] Task Scheduling Service (S task ): Responsible for scheduling tasks based on path planning results and real-time data.

[0168] Real-time monitoring service (S monitor ): Monitor status changes during task execution in real time and provide feedback.

[0169] Adaptive Scheduling Adjustment Service (S adaptive ): Automatically adjust resource allocation strategies based on real-time feedback and task status.

[0170] Each microservice communicates through a message queue (such as Kafka) or an asynchronous messaging mechanism to ensure data flow and execution coordination between modules.

[0171] Furthermore, the scheduling decision is based on the path planning result x ij , task allocation strategy φ r and real-time monitoring data t , dynamically adjust the execution strategy. The main factors considered in scheduling decisions include real-time traffic conditions, resource availability, and task execution deviation, ensuring that tasks can be completed efficiently and accurately.

[0172] Furthermore, the path weights are adjusted based on real-time traffic information, and the actual weights of each path are calculated in real time. The adjusted path planning formula is:

[0173]

[0174] Among them, w ij is the initial path weight, traffic ij For the path (v i ,v j ) real-time traffic information (such as traffic delays, road congestion, etc.), weather ij For the path (v i ,v j ) weather impact (such as traffic slowdown caused by rain or snow), time of day ij is the real-time time factor on the path (such as traffic changes during peak hours), and λ1,λ2,λ3 are weight coefficients that adjust the impact of real-time information on path selection. This adjustment formula ensures the dynamic adaptability of the path, can timely reflect the impact of changing traffic, weather and time, and avoid delays in path selection.

[0175] Furthermore, if a delay or idle resources occurs during task execution, the scheduling service needs to dynamically adjust the task allocation strategy based on the task deviation time. To achieve this, we first define the task deviation time ΔTi :

[0176] ΔT i =T actual,i -T planned,i

[0177] Among them, T actual,i represents the actual execution time of task i, T planned,i represents the planned execution time of task i, ΔT i represents the execution deviation time of task i. When the task deviation is too large, the scheduling algorithm will automatically adjust the resource allocation of the task and select a resource r with a closer distance and more suitable status to reschedule the task. The adjustment strategy is as follows:

[0178]

[0179] Among them, dist r is the distance between resource r and task i, avail r is the current availability of resource r (such as idle time, load condition), priority r is the priority of resource r, reflecting its applicable priority requirements, β1, β2 are adjustment weights, determining the impact of distance and resource priority on scheduling decisions. This adjustment formula ensures the timely execution of tasks and can automatically adjust resource allocation in the event of task delays or resource shortages to ensure the smooth completion of tasks.

[0180] Furthermore, the core function of the real-time monitoring module is to detect the state changes during the task execution process in real time, especially the task deviation, resource status, system load, etc. If a deviation or anomaly is detected, the monitoring module will pass the monitoring data to the scheduling system, triggering the scheduling service to make dynamic adjustments.

[0181] Furthermore, in order to effectively monitor the real-time performance of task execution, the present invention introduces a real-time feedback mechanism. The real-time monitoring formula for task execution is:

[0182] ΔT i =T actual,i -T planned,i

[0183] If ΔT i >δ i (set tolerance range), the task rescheduling is triggered through the scheduling system.

[0184] Furthermore, for each resource usage status, the present invention defines the resource workload L r :

[0185]

[0186] Among them, T assigned,r is the allocated working time of resource r, T total,r is the total working time (maximum available time) of resource r. When the resource load exceeds the set threshold (L r >τ r ), will trigger resource reallocation to avoid overuse of a certain resource.

[0187] Furthermore, in addition to basic path adjustment and task rescheduling, the scheduling service must also have the ability to dynamically balance loads to ensure that resources can be reasonably allocated when the system load is high and to avoid overloading of a single node.

[0188] In the microservice architecture, the system dynamically adjusts the load balancing strategy (such as load monitoring based on request volume or service response time) to ensure that the response time of each service does not exceed the predetermined threshold τ max .

[0189] Load balancing adjustment formula:

[0190]

[0191] Among them, response time s is the response time of service s, τ s is the threshold of response time, Status s The current status of the service (normal or standby). When a service overload is detected, the system will automatically switch to the standby service to ensure that the entire scheduling process is not interrupted.

[0192] The final output includes the execution status of the task, path selection, resource usage, and real-time monitoring data. By analyzing these results, the system can further optimize the scheduling algorithm and improve the path planning and task allocation strategies.

[0193] This step achieves real-time scheduling and optimization of path planning and task allocation through microservice architecture and dynamic scheduling mechanism. By innovatively combining path planning, task scheduling and real-time monitoring, the complex problems of path selection and resource scheduling in a dynamic environment are solved. The use of microservice architecture ensures the efficiency, scalability and stability of the system. At the same time, through real-time monitoring and feedback mechanism, the system can adaptively adjust the task execution process to ensure the smooth completion of tasks in any complex environment.

[0194] S4. Design a self-learning mechanism to automatically adjust according to historical tasks and real-time feedback to optimize the transportation line handling tasks.

[0195] Specifically, the goal of the self-learning feedback mechanism is to dynamically adjust the path planning and task scheduling model based on the real-time feedback data of the system to cope with complex actual scenarios. This mechanism optimizes path selection, task allocation and resource scheduling strategies by continuously accumulating the actual performance of task execution and scheduling.

[0196] The key to self-learning is to use feedback data to establish a scheduling decision-making mechanism that can dynamically learn and adjust. Its main process includes the following parts:

[0197] Error analysis and adjustment model:

[0198] According to the deviation time ΔT of each task i and resource load variation L r , calculate the deviation ΔM of the scheduling model i , and combined with historical feedback data for learning. The update formula of the scheduling error model is as follows:

[0199]

[0200] Where, ΔT i is the execution time deviation of task i, L r is the current load of resource r, α, β, γ are the impact coefficients of scheduling error on path selection and task allocation strategy, T actual,i and T planned,i are the actual execution time and the planned execution time of task i respectively. The function of this formula is to combine the deviation of each task execution with the resource load to form a feedback error model ΔM i , providing a basis for the next round of scheduling decisions.

[0201] Model optimization and adjustment:

[0202] Using the error model ΔM i Adjust the original path planning and task scheduling strategy. According to the execution effect of the task, the load of resources and the deviation time, adjust the path planning model and resource allocation strategy to optimize the scheduling results. The specific path planning optimization formula is:

[0203]

[0204] Among them, w ij is the original path weight, η1 is the sensitivity factor of path adjustment, ΔM i is the scheduling error output by the error model, and α is a hyperparameter that controls the degree of influence of the error. This formula is combined with the error model ΔM i , dynamically adjust the path planning and achieve path optimization based on feedback data.

[0205] Furthermore, under the self-learning feedback mechanism, the task scheduling and execution process will be dynamically adjusted according to real-time feedback. The core purpose is to improve the accuracy and efficiency of task execution and ensure that the system can adaptively handle various complex situations. To this end, the present invention proposes a dynamic task adjustment strategy and reallocates resources in combination with the optimization results of path planning.

[0206] Furthermore, during the task execution process, if the actual execution time of the task exceeds the expected deviation range ΔT i , then through the rescheduling mechanism, the most suitable resources are selected for task reallocation. When reallocating, the task priority i Adjustments will be made based on the urgency and importance of the task. The formula for task rescheduling is:

[0207]

[0208] Among them, dist r is the distance between resource r and task i, avail r is the availability of resource r (such as idle time, current load, etc.), priority r is the priority of resources, and γ is the impact coefficient of feedback error on resource scheduling. During task execution, the system optimizes the resource allocation of tasks to ensure that each task can get the best resource support, thereby improving the success rate and efficiency of task completion.

[0209] Furthermore, the system will continuously collect execution feedback data and continuously update the scheduling model through incremental learning. In order to achieve the adaptive capability of the system, the present invention introduces an incremental learning update mechanism to dynamically update the path planning and task scheduling model based on historical feedback data. Incremental learning update formula

[0210] After each execution feedback, the system will update the path planning model according to the new data. The update formula is:

[0211] θ new =θ old +η2·ΔM i

[0212] Among them, θ old are the model parameters before updating (such as path planning weights, resource allocation priorities, etc.), η2 is the learning rate, ΔM i is the feedback error generated by the error model. The updated model θ new This will be used in the next path planning and task scheduling decision. Through this incremental learning and updating mechanism, the system can gradually optimize the scheduling strategy in the process of continuously accumulating feedback data, and form a more efficient path planning and task allocation solution.

[0213] Furthermore, in the output of this step, the system adjusts the path planning strategy, task allocation strategy and resource scheduling strategy according to the results of the self-learning feedback mechanism. The final output scheduling results include:

[0214] Optimized path planning decision,

[0215] New task allocation scheme,

[0216] The adjusted resource scheduling strategy,

[0217] Feedback error data during adaptive scheduling.

[0218] The core of this step is to dynamically optimize path planning and task scheduling strategies through self-learning feedback mechanisms and incremental learning methods. Through real-time monitoring feedback and analysis of scheduling errors, the system can continuously update the model and make adaptive adjustments to path planning, task allocation, and resource scheduling. The self-learning feedback mechanism ensures that the system can maintain efficiency and accuracy in the face of a constantly changing task execution environment, thereby improving the overall scheduling execution efficiency, optimizing resource usage, and ensuring that tasks are completed on time and accurately.

[0219] like Figure 2 As shown, the embodiment of the present invention also provides a dedicated transport line handling optimization system based on intelligent scheduling, the system comprising:

[0220] The dynamic timeliness graph model unit 601 is used to construct a dynamic timeliness graph model and perform real-time data collection; wherein the structure of the dynamic timeliness graph model is as follows: each node represents a key location, including a starting point, an end point or a transfer point, and contains location information and key attributes of a task; each edge represents a transportation path, and the weight of the edge reflects the timeliness of the path; the update mechanism of the dynamic timeliness graph model adopts a sliding window mechanism to update the weight of the edge;

[0221] The path planning and task allocation unit 602 is used to perform path planning and task allocation based on the dynamic time-effectiveness graph model; wherein the objective function of the path planning and task allocation is to minimize the total transportation cost and satisfy the task time window and resource constraints. The total transportation cost includes time cost and resource allocation cost. The objective function of the total transportation cost is:

[0222]

[0223] Where C is the total transportation cost, w ij For the edge (v i ,v j ) represents the path cost, x ij ∈{0,1} is a decision variable, indicating that the edge (v i ,vj ) is selected; φ r Assign costs to resources, combining the empty load rate and allocation efficiency of each transportation resource;

[0224] The microservice architecture construction unit 603 is used to design a microservice architecture based on a distributed architecture, split the functions into different modules, and different microservices are responsible for independent functions. Each microservice will dynamically adjust the execution strategy based on the output of the dynamic timeliness graph model and the real-time monitoring data, and in combination with the real-time monitoring data. If a delay or idle resources occur during the task execution process, the scheduling service of the microservice architecture needs to dynamically adjust the task allocation strategy according to the deviation time of the task; wherein, the functions are split into different modules, including: path planning service, task scheduling service, real-time monitoring service and adaptive scheduling adjustment service;

[0225] The path planning service is used to calculate and dynamically adjust the path selection of the task;

[0226] The task scheduling service is used to schedule tasks according to the path planning results and real-time data;

[0227] The real-time monitoring service is used to monitor the status changes during task execution in real time and provide feedback;

[0228] The adaptive scheduling adjustment service is used to automatically adjust the resource allocation strategy according to real-time feedback and task status;

[0229] The self-learning optimization unit 604 is used to design a self-learning mechanism to automatically adjust according to historical tasks and real-time feedback to optimize the transportation line handling tasks.

[0230] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0231] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a division of logical functions. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, and the indirect coupling or communication connection of the device or unit can be electrical, mechanical or other forms.

[0232] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, server or network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, and other media that can store program codes.

[0233] Although embodiments of the present invention have been shown and described, those skilled in the art will appreciate that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the claims and their equivalents.

Claims

1. A transportation line handling optimization method based on intelligent scheduling, characterized in that: The method comprises: S1. Construct a dynamic time-efficiency graph model and collect real-time data; wherein the structure of the dynamic time-efficiency graph model is as follows: each node represents a key location, including a starting point, an end point or a transfer point, and contains location information and key attributes of the task; each edge represents a transportation path, and the weight of the edge reflects the timeliness of the path; the update mechanism of the dynamic time-efficiency graph model uses a sliding window mechanism to update the weight of the edge; S2. Perform path planning and task allocation based on a dynamic time-effectiveness graph model; wherein the objective function of the path planning and task allocation is to minimize the total transportation cost and satisfy the task time window and resource constraints. The total transportation cost includes time cost and resource allocation cost. The objective function of the total transportation cost is: Where C is the total transportation cost, w ij For the edge (v i ,v j ) represents the path cost, x ij ∈{0,1} is a decision variable, indicating that the edge (v i ,v j ) is selected; φ r Assign costs to resources, combining the empty load rate and allocation efficiency of each transportation resource; S3. Design a microservice architecture based on a distributed architecture, split functions into different modules, and different microservices are responsible for independent functions. Each microservice will dynamically adjust the execution strategy based on the output of the dynamic timeliness graph model and real-time monitoring data, and in combination with the real-time monitoring data. If a delay or idle resources occur during task execution, the scheduling service of the microservice architecture needs to dynamically adjust the task allocation strategy according to the deviation time of the task; wherein, the functions are split into different modules, including: path planning service, task scheduling service, real-time monitoring service and adaptive scheduling adjustment service; The path planning service is used to calculate and dynamically adjust the path selection of the task; The task scheduling service is used to schedule tasks according to the path planning results and real-time data; The real-time monitoring service is used to monitor the status changes during task execution in real time and provide feedback; The adaptive scheduling adjustment service is used to automatically adjust the resource allocation strategy according to real-time feedback and task status; S4. Design a self-learning mechanism to automatically adjust according to historical tasks and real-time feedback to optimize the transportation line handling tasks.

2. According to claim 1, a method for optimizing transportation line handling based on intelligent scheduling is characterized in that: The edge weights of the dynamic temporal graph model are calculated as follows: w ij =α·D ij +β·T ij +γ·C ij +δ·W ij +λ·f ij Among them, D ij is the distance between nodes i and j, T ij is the estimated time of the transportation route, C ij is the impact of traffic flow on the path, W ij is the impact of weather on the path, f ij is an additional path constraint, α, β, γ, δ, and λ are adjustment parameters that adjust the impact of different factors on the weights; The sliding window mechanism is used to update the edge weights, which is expressed as: Updatedw ij (t)=w ij (t-Δt)+Δw ij (t) Among them, w ij (t0 represents the edge weight at time t, Δt is the time interval, Δw ij (t) is the edge weight change caused by real-time data flow.

3. The method for optimizing transportation line handling based on intelligent scheduling according to claim 2 is characterized in that: The S1 also includes optimizing the sensor data using Kalman filtering, which is expressed as: in, represents the estimated value of the state (such as position, speed, etc.) at time k, z k is the measured value at time k, K k is the Kalman gain, H k is the measurement matrix, which represents the relationship between measurement and state.

4. The method for optimizing transportation line handling based on intelligent scheduling according to claim 1 is characterized in that: The total transportation cost is also subject to constraints, including: Each task node must be visited exactly once: Where V is the set of nodes in the graph; All tasks must meet the time window constraints: Where T i is the task start time, T start,i and T end,i The earliest and latest time for the task; The load of a transport resource must not exceed its capacity: where d i is the task demand, C r is the maximum load of resource r.

5. The method for optimizing transportation line handling based on intelligent scheduling according to claim 4 is characterized in that: The S2 also includes the joint optimization of path planning and task allocation, specifically: According to the task priority and time window constraints, an initial access sequence is generated, where nodes with high priority and tight time windows are accessed first; Use the dynamic time-effective graph model for path planning, and for each task node v i , choose the weight w ij The smallest side (v i ,v j ); Prioritize each transport resource r to the task node v that best matches its current status i ; Introduce an adjustment mechanism to recalculate w through a sliding window ij and x ij , dynamically update paths and task allocations.

6. The method for optimizing transportation line handling based on intelligent scheduling according to claim 5 is characterized in that: The sorting strategy of the access sequence is: Among them, S is the access sequence, argsort is the sorting function, V is the total number of nodes, T end,i is the end window, T start,i is the start window; The dynamic path selection rule for path planning by the dynamic time-effectiveness graph model is: Among them, f ij is an additional path constraint, λ is a regularization coefficient used to adjust the importance of the additional constraint, is the initial path; The method allocates each transport resource r to the task node v that best matches its current state. i , specifically: Based on the distance D between the current location of the resource and the location of the task r , no-load rate η r and task requirements i , computing resource adaptation, expressed as: Select φ r Minimal resources to perform tasks; The sliding window recalculates w ij and x ij , dynamically update the path and task allocation, specifically: In each sliding window period, the updated path selection rule is: in, For the updated path, is the initial path, Δx ij Path changes caused by real-time data flow.

7. The method for optimizing transportation line handling based on intelligent scheduling according to claim 1 is characterized in that: The S3 specifically includes: Adjust the path weights based on real-time traffic information and calculate the actual weights of each path in real time The adjusted path planning formula is: Among them, w ij is the initial path weight, traffic ij For the path (v i ,v j ) real-time traffic information, weather ij For the path (v i ,v j Weather influence, time of day ij is the real-time period factor on the path, λ1,λ2,λ3 are weight coefficients, which adjust the impact of real-time information on path selection; If a delay or idle resources occurs during task execution, the scheduling service of the microservice architecture needs to dynamically adjust the task allocation strategy according to the deviation time of the task, including: First, define the task deviation time ΔT i : ΔT i =T actual,i -T planned,i Among them, T actual,i represents the actual execution time of task i, T planned,i represents the planned execution time of task i, ΔT i represents the execution deviation time of task i; When the task deviation is too large, the scheduling algorithm will automatically adjust the resource allocation of the task and select the resource r with a closer distance and more suitable status to reschedule the task. The adjustment strategy is as follows: in, For the updated strategy, dist r is the distance between resource r and task i, avail r is the current availability of resource r, priority r is the priority of resource r, reflecting its applicable priority requirements, and β1 and β2 are adjustment weights that determine the impact of distance and resource priority on scheduling decisions.

8. The method for optimizing transportation line handling based on intelligent scheduling according to claim 7 is characterized in that: S3 also includes that the real-time monitoring service detects the state changes during the task execution process in real time. If a deviation or anomaly is detected, the real-time monitoring service will pass the real-time monitoring data to the task scheduling service, triggering the task scheduling service to make dynamic adjustments, which is specifically expressed as follows: ΔT i =T actual,i -T planned,i Where, ΔT i is the deviation time of each task, T actual,i is the actual time, T planned,i is the planned time, if ΔT i >δ i , then the task rescheduling is triggered through the scheduling system; Monitor the usage status of each resource, expressed as: Among them, T assigned,r is the allocated working time of resource r, T total,r is the total working time (maximum available time) of resource r. When the resource load exceeds the set threshold, L r >τ r , it triggers resource reallocation to avoid overuse of a certain resource; At the same time, in the microservice architecture, dynamic adjustment is performed through the load balancing strategy to ensure that the response time of each service does not exceed the predetermined threshold. When a service overload is detected, it automatically switches to the backup service to ensure that the entire scheduling process is not interrupted; wherein, the load balancing strategy is expressed as: Among them, response time s is the response time of service s, τ s is the threshold of response time, Status s The current status of the service.

9. The method for optimizing transportation line handling based on intelligent scheduling according to claim 1 is characterized in that: The S4 specifically includes: According to the deviation time of each task and the change in resource load, the deviation of the scheduling model is calculated, and the historical feedback data is combined for learning. The original path planning and task scheduling strategy are adjusted according to the deviation of the scheduling model, so that the scheduling result tends to be optimized, which is expressed as: Where, ΔT i is the execution time deviation of task i, L r is the current load of resource r, α, β, γ are the impact coefficients of scheduling error on path selection and task allocation strategy, T actual,i and T planned,i are the actual execution time and planned execution time of task i, respectively, and w ij is the original path weight, η1 is the sensitivity factor of path adjustment, ΔM i is the scheduling error output by the error model, and α is a hyperparameter that controls the impact of the error; Design dynamic task adjustment strategies, combine the optimization results of path planning, and reallocate resources, including: If the actual execution time of the task exceeds the expected deviation range ΔT i , then through the rescheduling mechanism, the most suitable resources are selected for task reallocation. When reallocating, the task priority i Adjustments will be made based on the urgency and importance of the task. The formula for task rescheduling is: Among them, dist r is the distance between resource r and task i, avail r is the availability of resource r, priority r is the priority of resources, γ is the impact coefficient of feedback error on resource scheduling; Collect user execution feedback data, update the scheduling error output by the error model through incremental learning, and dynamically update path planning and task scheduling.

10. A dedicated transport line handling optimization system based on intelligent scheduling, characterized in that: The system comprises: A dynamic timeliness graph model unit is used to construct a dynamic timeliness graph model and perform real-time data collection; wherein the structure of the dynamic timeliness graph model is as follows: each node represents a key location, including a starting point, an end point or a transfer point, and contains location information and key attributes of the task; each edge represents a transportation path, and the weight of the edge reflects the timeliness of the path; the update mechanism of the dynamic timeliness graph model adopts a sliding window mechanism to update the weight of the edge; The path planning and task allocation unit is used to perform path planning and task allocation based on a dynamic time-effectiveness graph model; wherein the objective function of the path planning and task allocation is to minimize the total transportation cost and satisfy the task time window and resource constraints. The total transportation cost includes time cost and resource allocation cost. The objective function of the total transportation cost is: Where C is the total transportation cost, w ij For the edge (v i ,v j ) represents the path cost, x ij ∈{0,1} is a decision variable, indicating that the edge (v i ,v j ) is selected; φ r Assign costs to resources, combining the empty load rate and allocation efficiency of each transportation resource; A microservice architecture construction unit is used to design a microservice architecture based on a distributed architecture, split functions into different modules, and different microservices are responsible for independent functions. Each microservice will dynamically adjust the execution strategy based on the output of the dynamic timeliness graph model and real-time monitoring data, and in combination with the real-time monitoring data. If a delay or idle resources occur during the task execution process, the scheduling service of the microservice architecture needs to dynamically adjust the task allocation strategy according to the deviation time of the task; wherein, the functions are split into different modules, including: path planning service, task scheduling service, real-time monitoring service and adaptive scheduling adjustment service; The path planning service is used to calculate and dynamically adjust the path selection of the task; The task scheduling service is used to schedule tasks according to the path planning results and real-time data; The real-time monitoring service is used to monitor the status changes during task execution in real time and provide feedback; The adaptive scheduling adjustment service is used to automatically adjust the resource allocation strategy according to real-time feedback and task status; The self-learning optimization unit is used to design a self-learning mechanism to automatically adjust according to historical tasks and real-time feedback to optimize the transportation line handling tasks.

Images