A method and device for coordinating the operation of a terminal vehicle and a handling device

Abstract

The application provides a wharf vehicle and unloading equipment cooperative operation method and device, which is applied to the field of information technology, regards the whole wharf vehicle and unloading equipment as a unified and dynamically schedulable resource pool, virtualizes the wharf resource by constructing an Actor distributed model, dynamically predicts the operation time length based on historical data by the operation point Actor, and generates an operation plan ensuring continuity in combination with bidirectional space-time deduction; the task is dispatched to the optimal vehicle through a global task matching mechanism, and the subsequent scheduling instruction is dynamically adjusted by using a negative feedback control strategy based on the probability distribution formed by the actual execution deviation, so that the cooperative optimization of unloading operation continuity, vehicle utilization rate and system adaptability is realized, the unloading operation interruption is avoided, the ship time in port is shortened, and the overall operation efficiency and service quality of the wharf are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of information technology, in particular to a method and device for collaborative operation of vehicles and loading and unloading equipment at a wharf. BACKGROUND

[0002] In the process of loading and unloading at a wharf, improving the operation efficiency of a single machine and shortening the time of a ship in port are core demands of a wharf. At present, the operation road scheduling mode is generally used in shore operation, that is, a fixed number of vehicles are allocated to each loading and unloading equipment of a ship to be operated as a resource pool, and there are N resource pools, that is, N operation roads, for N loading and unloading equipment, and vehicle scheduling is based on the operation road. The continuity problem of shore operation is solved by reserving a sufficient number of vehicles in the resource pool, which will inevitably cause uneven busy and idle among the operation roads and cause waste of resources.

[0003] Although the existing improved method integrates multiple resource pools to balance the uneven busy and idle among the operation roads and can conditionally realize the re-entry and re-exit operation. For example, under the condition of no conflict, the ship can be loaded and unloaded simultaneously. However, the underlying logic of the improved scheme is still limited to the basic logic of the operation road: scheduling depends on the experience of personnel, a sufficient number of vehicles need to be reserved, and a group of tasks need to be arranged in advance to make the shore operation as continuous as possible. Vehicles and tasks can only be matched in a local range, vehicles cannot be fully utilized, and tasks may not be matched to more suitable vehicles.

[0004] Therefore, a new collaborative operation scheme of vehicles and loading and unloading equipment at a wharf is needed. SUMMARY

[0005] Therefore, the embodiments of the present application provide a method and device for collaborative operation of vehicles and loading and unloading equipment at a wharf, which abandon the operation road mode, and under the premise of considering the operation continuity and vehicle utilization rate, adaptively control the collaborative operation of vehicles and loading and unloading equipment at a wharf to achieve a dynamic and optimal balance effect.

[0006] The embodiments of the present application provide the following technical solutions:

[0007] The embodiments of the present application provide a method for collaborative operation of vehicles and loading and unloading equipment at a wharf, comprising:

[0008] An Actor distributed parallel computing model is constructed and deployed to an Actor service container, and a job point and a vehicle in the wharf are respectively mapped to a job point Actor and a vehicle Actor;

[0009] The job point Actor receives a job task message, parses the job conditions of the current task, and screens the corresponding historical job records according to the job conditions and performs mode statistics to obtain the predicted job duration of the current task;

[0010] The job point Actor predicts the planned job time period of the current task according to the planned completion time of the previous job point or the planned start time of the next job point, in combination with the predicted job duration, so as to continuously connect the planned completion time of the previous job point and the planned start time of the next job point;

[0011] A specified time window for the arrival of a vehicle at the job point is generated according to the planned job time period, a task message is issued by the job point Actor to a vehicle job near-junction service, the task is matched to the corresponding vehicle Actor through the vehicle job near-junction service, and the task message is pushed; wherein the task message at least includes the specified time window;

[0012] The job point Actor continuously receives a task feedback message of the vehicle Actor, wherein the task feedback message at least includes the actual arrival time of the vehicle;

[0013] According to the task feedback message, a deviation value of the actual arrival time relative to the specified time window is obtained, and a deviation probability distribution is constructed based on the deviation value;

[0014] Based on the deviation probability distribution, a negative feedback control strategy is applied to dynamically adjust the specified time window of the subsequent task.

[0015] The embodiments of the present specification also provide a wharf vehicle and unloading equipment cooperative operation device, comprising:

[0016] A model construction and deployment module is configured to construct an Actor distributed parallel computing model and deploy it to an Actor service container, and map the job points and vehicles in the wharf to job point Actors and vehicle Actors, respectively;

[0017] A job duration estimation module is configured to receive a job task message, parse the job conditions of the current task, and screen the corresponding historical job records according to the job conditions and perform mode statistics to obtain the predicted job duration of the current task;

[0018] A job time period prediction module is configured to predict the planned job time period of the current task according to the planned completion time of the previous job point or the planned start time of the next job point, in combination with the predicted job duration, so as to continuously connect the planned completion time of the previous job point and the planned start time of the next job point;

[0019] a task matching module configured to generate a specified time window for a vehicle to arrive at a work point according to the planned work period, publish a task message to the vehicle work near-matching service by the work point Actor, match a task to a corresponding vehicle Actor through the vehicle work near-matching service, and push the task message, wherein the task message at least includes the specified time window;

[0020] a dynamic adjustment module configured to continuously receive a task feedback message of the vehicle Actor, wherein the task feedback message at least includes an actual arrival time of the vehicle;

[0021] According to the task feedback message, a deviation value of the actual arrival time relative to the specified time window is obtained, and a deviation probability distribution is constructed based on the deviation value.

[0022] Based on the deviation probability distribution, a negative feedback control strategy is applied to dynamically adjust the specified time window of a subsequent task.

[0023] Compared with the prior art, the above at least one technical solution adopted by the embodiments of the present application can achieve at least the following beneficial effects:

[0024] Based on the Actor distributed parallel computing model, the port global vehicle and the handling equipment are abstracted as a virtual Actor resource pool that can be elastically expanded and contracted. Through dynamic estimation of the work duration, planning of the work period, and combination of the fastest driving path search and vehicle near-matching, adaptive collaborative scheduling of the vehicle and the handling equipment is realized. At the same time, a negative feedback control strategy is used to dynamically adjust the work time window, so that the vehicle utilization rate is optimized while ensuring continuous work at the loading and unloading point, the ship time in port is significantly shortened, and the overall work efficiency and resource utilization rate of the port are improved. BRIEF DESCRIPTION OF DRAWINGS

[0025] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed in the embodiments will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0026] Figure 1 is a flow chart of a port vehicle and handling equipment collaborative work method in the present application;

[0027] Figure 2 is a schematic diagram of predicting a planned work period in the present application;

[0028] Figure 3 is a schematic diagram of a negative feedback control strategy when the range deviation is small in the present application;

[0029] Figure 4 is a schematic diagram of the negative feedback control strategy when the range is too large in the present application;

[0030] Figure 5 is a class diagram of the method for vehicle and loading and unloading device cooperative operation in the present application. DETAILED DESCRIPTION

[0031] The embodiments of the present application will be described in detail below with reference to the drawings.

[0032] The above and other aspects of the present application will become more apparent by describing in detail the embodiments thereof with reference to the attached drawings in which:

[0033] It is to be understood that the foregoing description is that of certain examples of the application and that numerous changes in the details of construction and the combination and arrangement of parts can be made by those skilled in the art without departing from the scope of the application.

[0034] It is also to be understood that the following description is only illustrative of the aspects of the present application and that no limitation of the scope of the application is intended by either the inclusion or non-inclusion of specific aspects. It is intended, however, that the application cover any and all modifications and variations of the various aspects included within the scope of the overall concept.

[0035] In addition, in the following description, numerous specific details are provided, such as examples of components and / or methods, to provide a thorough understanding of examples incorporating the present application. One skilled in the relevant art will recognize, however, that the application can be practiced without one or more of the specific details, or with other methods, components, etc. In other instances, well-known structures and components are shown in block diagram form in order to avoid obscuring the concepts of the application.

[0036] In the process of wharf loading and unloading operation, the traditional "operation road" mode is simple and intuitive in management, but its underlying logic has the following problems:

[0037] Firstly, the "operation road" mode statically and fixedly allocates vehicles to specific quays, but ship operation is dynamic, and the operation amount and difficulty of different ship holds are greatly different, resulting in uneven operation of each operation line: the busy operation road lacks vehicles, and the quay waits, while the idle operation road has a large number of idle vehicles, causing structural waste of resources.

[0038] Secondly, in the "operation road" mode, the dispatching decision is limited within a single operation road, vehicles cannot flow across paths, and it is difficult to achieve optimal matching of vehicles and tasks from the overall efficiency of the wharf, which restricts the improvement of the overall efficiency of the system.

[0039] Thirdly, the operation continuity highly depends on the redundant vehicles reserved by the dispatching personnel based on experience, and once the actual situation does not match the experience (such as vehicle failure and traffic congestion), the entire plan is easy to be disrupted, lacking flexibility and self-adaptive ability.

[0040] Therefore, the inventors have found that, although the existing scheme alleviates the uneven busy and idle through resource pool integration to a certain extent and supports conditional re-entry and re-discharge operations (such as simultaneous loading and unloading), the underlying logic is still based on the static resource division centering on the equipment (quay), and the dispatching still relies on manual experience to pre-configure the vehicles and tasks, the matching range is limited to the local, and the global self-adaptive optimization cannot be realized in a true sense.

[0041] The following professional terms are explained:

[0042] Horizontal transportation task (referred to as "task / operation task"), refers to the tasks of ship to yard (unloading cargo from ship to yard), yard to ship (loading cargo from yard to ship), ship to ship (transshipment), and yard to yard (transshipment) for transporting cargo between ship loading and unloading equipment and yard loading and unloading equipment by horizontal transportation equipment (i.e. wharf vehicles, referred to as "vehicles"). The task includes a set of instructions, which requires the vehicle to be engaged in the task to arrive at the loading point within the specified time window and complete the loading within the planned operation period and then be transported to the designated unloading point.

[0043] Loading point and unloading point, refer to the operation points for vehicles to load and unload cargo distributed on the road structure of the wharf network. On the shore, it is the loading and unloading position under the ship loading and unloading equipment, and the operation is shore operation; in the yard, it is the loading and unloading position under the yard loading and unloading equipment, and the operation is yard operation.

[0044] The planned operation period refers to a period of time from the planned start time to the planned completion time of the vehicle at the operation point, and the time span is the expected operation time. The expected operation time is at least divided into the expected loading time of the loading point and the expected unloading time of the unloading point, and other analogies.

[0045] The specified time window refers to the best time period for the vehicle to drive to the operation point, from the earliest arrival time to the latest arrival time. If the best time period is exceeded, the order of loading and unloading operations will be disrupted if it arrives early, and the loading and unloading operations will be stopped if it arrives late. Generally, the earliest arrival time of the specified time window is the planned start time of the planned operation period of the previous vehicle, and the latest arrival time is the planned completion time of the planned operation period of the previous vehicle.

[0046] The Actor distributed parallel computing model originated from Carl Hewitt's concept model for parallel computing in 1973. Today's mainstream cloud-native tools and runtime environments can implement the construction, registration, addressing, activation, running and release of virtual Actor computing units under the distributed system architecture, automatically exclude faults, recover normal operation, and realize event-driven and parallel computing of virtual Actor computing units under the distributed system environment based on the publish-subscribe pattern (Pub-Sub), with technical features such as real-time, reliability, robustness and scalability, which can support large-scale real-time data processing and high-performance computing. For example, the open-source Dapr is the preferred Actor distributed parallel computing platform of the present application. Dapr is a language-independent, event-driven, and sidecar mode running microservice runtime that helps developers build portable, scalable, and maintainable stateless and stateful applications running in cloud platforms or edge computing.

[0047] Virtual Actor computing unit (Actor) is also called distributed virtual state machine, which is a computing and state independent distributed computing unit deployed in a distributed computing hosting service cluster and registered on a hosting service instance (Actor service container) of the cluster. The Actor instance has its own behavior and data and can subscribe to external messages. When receiving a message or being remotely called, the Actor instance can execute appropriate behaviors according to its own state. The Actor instance can also be activated by its own timer and reminder mechanism (Timers and Reminders) at a predetermined time to execute a preset process and affect the outside world. The Actor is transparent to the outside world, and the outside world can access a specific Actor instance through an agreed Actor keyword (ID) without knowing where it runs. Generally, an Actor instance is not released immediately after being activated, but retains its own data until it is no longer accessed and has no new messages, and then is released by the hosting service instance, thereby achieving high-performance response and computation and greatly reducing the operation pressure of the system service on the database.

[0048] Based on this, the embodiment of the present specification provides a method for collaborative operation of terminal vehicles and handling equipment, which breaks down the barriers between operation roads, and the overall idea is as follows: regarding the vehicles and handling equipment in the whole terminal as a unified and dynamically schedulable resource pool, virtualizing the terminal resources by constructing an Actor distributed model, dynamically predicting the operation time based on historical data by the job point Actor, and generating a job plan that guarantees continuity in combination with bidirectional space-time deduction; dispatching tasks to the optimal vehicle through a global task matching mechanism, and dynamically adjusting the subsequent scheduling instructions based on the probability distribution formed by the actual execution deviation, so as to realize the collaborative optimization of handling operation continuity, vehicle utilization rate and system adaptability, avoid handling operation interruption, shorten the ship's time in port, and improve the overall operation efficiency and service quality of the terminal.

[0049] The technical solutions provided by the embodiments of the present application are described below with reference to the accompanying drawings.

[0050] As shown in Figure 1 The embodiment of the present specification provides a method for collaborative operation of terminal vehicles and handling equipment, which breaks down the barriers between operation roads, and the overall idea is as follows: regarding the vehicles and handling equipment in the whole terminal as a unified and dynamically schedulable resource pool, virtualizing the terminal resources by constructing an Actor distributed model, dynamically predicting the operation time based on historical data by the job point Actor, and generating a job plan that guarantees continuity in combination with bidirectional space-time deduction; dispatching tasks to the optimal vehicle through a global task matching mechanism, and dynamically adjusting the subsequent scheduling instructions based on the probability distribution formed by the actual execution deviation, so as to realize the collaborative optimization of handling operation continuity, vehicle utilization rate and system adaptability, avoid handling operation interruption, shorten the ship's time in port, and improve the overall operation efficiency and service quality of the terminal.

[0051] Step S1, an Actor distributed parallel computing model is constructed and deployed to an Actor service container, and the job points and vehicles in the terminal are respectively mapped to job point Actors and vehicle Actors.

[0052] In implementation, a distributed runtime environment is created in the Actor service container, and each job point (including shore crane loading and unloading sites, yard operation areas, etc.) in the physical world is abstracted as a job point Actor instance, and each transport vehicle is mapped as a vehicle Actor instance.

[0053] These Actor instances serve as parallel computing units, each maintaining independent state and communicating through asynchronous message passing: the job point Actor is responsible for receiving task instructions, estimating job duration, and scheduling job plans, while the vehicle Actor executes transportation instructions and provides real-time feedback on operating status, enabling large-scale real-time data processing and high-performance computing, sufficient to support large-scale port production scenarios.

[0054] Step S2, the job point Actor receives a job task message, parses the job conditions of the current task, and filters the corresponding historical job records according to the job conditions and performs mode statistics to obtain the estimated job duration of the current task.

[0055] Specifically, the single job duration under different job conditions (such as only the last ten thousand historical records under different job conditions are retained by default) is collected and recorded, and the historical records are filtered out using the job conditions of the current task on the job point and the mode is counted as the estimated job duration of the task.

[0056] In implementation, after receiving the job task message, the job point Actor parses the current task job conditions contained therein, such as: job mode, device performance, loading and unloading depth, loading and unloading span, weather influence, driver skill, and job time, etc. multi-dimensional feature parameters.

[0057] Subsequently, based on these conditions, the stored historical job records are filtered from coarse to fine, such as: first, the primary filter is performed according to the job mode and device performance, and then the effective historical records are filtered layer by layer according to the loading and unloading position, environmental factors, personnel skills, etc. single job duration in the effective historical records is statistically analyzed, and the longest duration value is taken as the estimated job duration of the current task. This intelligent estimation method based on multi-dimensional condition hierarchical filtering and mode statistics effectively overcomes the limitations of traditional fixed experience value, and realizes the precision and adaptive optimization of job duration prediction through data-driven.

[0058] Step S3, the job point Actor predicts the planned job time period of the current task according to the planned completion time of the previous job point or the planned start time of the next job point, combined with the estimated job duration, so that the planned completion time of the previous job point and the planned start time of the next job point are continuous.

[0059] In practice, generally speaking, as long as the continuity of operations at the loading point is ensured, the planned operation period at the loading point directly connects to the previous operation task. In this case, the planned operation period at the unloading point can be calculated through forward spatiotemporal extrapolation. In special cases, however, it may be necessary to ensure the continuity of operations at the unloading point, where the planned operation period directly connects to the previous operation task. In such cases, the planned operation period at the loading point needs to be calculated through backward spatiotemporal extrapolation. For example, in a yard-to-ship operation, the unloading point is at the shore. To shorten the ship's time in port, cargo vehicles need to arrive at the shore unloading point on time. The planned operation period at the shore unloading point directly connects to the previous operation task, while the planned operation period at the yard loading point needs to be calculated through backward spatiotemporal extrapolation.

[0060] Specifically, when the system aims to ensure continuous operation at loading points, the planned start time of the current task at the loading point is directly set as the planned completion time of the previous work point, and the planned completion time is calculated by combining the estimated operation time of that point. When it is necessary to prioritize ensuring continuous operation at unloading points (such as key quay cranes), the planned start time of the unloading point is used as the benchmark, and the planned completion time is determined by combining the estimated operation time of that point with the reverse calculation mode. Based on this, the time requirements are passed to the preceding work points, thereby ensuring seamless connection of the planned time periods of each work point and effectively guaranteeing the continuous and efficient operation of key loading and unloading equipment.

[0061] Step S4: Generate a designated time window for vehicle arrival at the work point according to the planned work period. The work point Actor publishes a task message to the vehicle work proximity matching service, and matches the task to the corresponding vehicle Actor through the vehicle work proximity matching service, and pushes the task message; wherein, the task message includes: the designated time window.

[0062] Specifically, after calculating the planned work period, the work point Actor generates a designated time window for vehicles to arrive at the work point. This window uses the planned start time as the latest arrival deadline and extrapolates the estimated work duration to obtain the earliest arrival time, forming a precise arrival time period that vehicles must adhere to. Subsequently, the work point Actor publishes the task message containing this designated time window to the vehicle task proximity matching service. Based on the global vehicle status and the principle of optimal empty driving time, this service constructs a weighted bipartite graph and uses a matching algorithm to select the most suitable vehicle Actor to perform the task from all vehicle resources. Finally, the task message is dispatched to the successfully matched vehicle, thus completing the closed-loop scheduling process from work planning to vehicle assignment.

[0063] In practice, more than one vehicle can often reach the loading point within the specified time window, and a single vehicle may also have the opportunity to reach multiple loading points. Therefore, the vehicle operation proximity connection service in this application can use the dock vehicle operation proximity connection method to achieve the best matching of tasks and vehicles.

[0064] The method for connecting terminal vehicles to the nearest location is a dynamic task matching method based on the Actor distributed parallel computing model. It finds the matching relationship with the shortest empty running time globally to maximize the utilization rate of terminal vehicles and improve the overall operational efficiency of the terminal.

[0065] Specifically, the fastest travel path search method is used to find vehicles that meet the conditions for obtaining tasks and can reach the loading point within a specified time window, including heavy-load vehicles that can reach the loading point after unloading. These vehicles form many-to-many, one-to-many, and one-to-one relationships with the tasks. The weighted bipartite graph and its corresponding bipartite Actor instances are constructed and updated using the negative of the expected empty-run time as the weight of the relationship. Based on distributed graph partitioning and disjoint-set data structures, graph partitioning methods, and graph merging methods, the number and size of the graphs are dynamically controlled and adjusted. To address the problem of multiple graphs competing for matching vehicles, the vehicle Actor makes decisions using methods to resolve matching conflicts. This allows the Actor distributed parallel computing model to apply the Kuhn-Munkres algorithm, processing large-scale real-time data pushed from the outside in parallel, selecting the matching relationship with the shortest global empty-run time, allowing vehicles to connect to the next most suitable task earlier, and finding even more suitable tasks before execution.

[0066] Once the optimal match between vehicles and tasks is achieved, the collaborative operation of all vehicles and loading / unloading equipment can be further optimized, so that after the previous vehicle completes its task at the work point, the next vehicle can immediately follow and enter the work point, allowing the loading / unloading equipment at the work point to operate continuously.

[0067] Step S5: The work point Actor continuously receives task feedback messages from the vehicle Actor, wherein the task feedback messages include at least: the actual arrival time of the vehicle;

[0068] Based on the task feedback message, the deviation value of the actual arrival time relative to the specified time window is obtained, and a deviation probability distribution is constructed based on the deviation value;

[0069] Based on the aforementioned deviation probability distribution, a negative feedback control strategy is applied to dynamically adjust the specified time window for subsequent tasks.

[0070] In practice, although a time window for vehicle arrival at the work site is specified when the task is issued, the actual arrival time is probabilistically distributed and may be early or late, exceeding the specified time window range. To make the coordinated operation of vehicles and loading and unloading equipment more resilient, it is necessary to dynamically specify the time window based on feedback, reducing the occurrence of vehicles arriving early or late at the source.

[0071] The statistical analysis of probability distribution data is handled by the job point entities mapped to job point Actor instances in the distributed parallel computing model.

[0072] Specifically, the system first subscribes to and receives task messages from external sources for the current work site; then it calculates the specified time window for the task and publishes the task to the nearest vehicle work connection service; based on the received task feedback messages, it records the specified time window and actual arrival time of each task, calculates the deviation value, and statistically analyzes the probability distribution.

[0073] By comparing the difference between the probability distribution range (the absolute value of the difference between the maximum and minimum deviation values) and the specified time window span (the absolute value of the difference between the earliest and latest arrival times), a corresponding negative feedback control strategy is adopted. The offset of the specified time window of the next task relative to the ideal target is calculated and used to gradually stabilize the vehicle arrival time within a reasonable range.

[0074] This application optimizes the coordinated operation of vehicles and loading / unloading equipment throughout the terminal, maintains the continuity of loading and unloading operations at work points, especially at key work points, avoids interruptions in loading and unloading operations, thereby shortening the time vessels spend in port and improving the overall operational efficiency and service quality of the terminal.

[0075] Explanation of technical terms in this embodiment:

[0076] The duration of a single operation refers to the time consumed by the loading and unloading equipment to complete one loading and unloading operation, and it records the time interval between consecutive operations. For example, for container terminals, the time is defined by the two unlocking indication signals (signals indicating the separation of the spreader from the container) from the loading and unloading equipment's PLC (Programmable Logic Controller).

[0077] Operating conditions refer to the essential characteristics of the task, resources, and operating environment that significantly affect the time required for a single loading and unloading operation.

[0078] The probability distribution refers to the statistical data on the number of times the deviation value (in seconds) of the actual arrival time occurs. The statistical scope is limited to the tasks currently being carried out continuously at the work point, and the content involved includes but is not limited to the specified time window and the actual arrival time of each task.

[0079] The deviation value (in seconds) refers to the difference between the median time of a specified time window and the actual arrival time. A positive deviation value indicates arrival earlier than the median time, while a negative deviation value indicates arrival later than the median time. In other words, the deviation value equals the median time of the specified time window minus the actual arrival time, accurate to the second.

[0080] In some embodiments, the estimated duration of the current task is obtained by filtering corresponding historical task records based on the task conditions and performing mode statistics, including:

[0081] Based on the current task's operating conditions and the type of loading and unloading equipment, determine the corresponding filtration level, wherein the filtration level includes multiple filtration conditions from coarse to fine.

[0082] The filtering conditions are used to filter the duration of each single task in the historical task records layer by layer.

[0083] The mode of the final selected single-job duration is statistically analyzed;

[0084] If a unique mode is obtained from the statistics, then the mode is taken as the estimated operation time.

[0085] If multiple modes are obtained from the statistics, the average of the modes shall be taken as the estimated operation time.

[0086] If no historical records are found, the preset default value is used as the estimated job duration.

[0087] Based on the above embodiments and through research and analysis, the operating conditions for ship loading and unloading equipment include, but are not limited to: operating mode (continuous loading / continuous unloading / loading while unloading), equipment performance (equipment ID), loading and unloading depth (inside the hold / on the deck), loading and unloading span (sea side / land side), weather impact (minor impact / moderate impact / high impact), driver skill (driver ID), and operating time (Zi / Chou / Yin / Mao / Chen / Si / Wu / Wei / Shen / You / Xu / Hai).

[0088] For yard loading and unloading equipment, operating conditions include, but are not limited to: operating mode (continuous loading / continuous unloading / loading and unloading simultaneously), equipment performance (equipment ID), loading and unloading span (inner side / roadside), weather impact (minor impact / moderate impact / high impact), driver skills (driver ID), and operating time (Zi / Chou / Yin / Mao / Chen / Si / Wu / Wei / Shen / You / Xu / Hai).

[0089] The content of these operational conditions can be added or removed according to the actual situation of the terminal. These operational conditions are collected from operational task messages and task feedback messages pushed from the outside. The outside needs to convert the raw data into categorized data (tags or categories) that can distinguish the differences in impact according to the interface specifications. For example, for container terminals, the row position (Row, which represents the position of the container in the width direction of the ship, even-numbered rows are port side, and odd-numbered rows are starboard side) needs to be converted into loading and unloading span (sea side / land side). Assuming that a ship is berthed on the starboard side, even-numbered rows are sea side and odd-numbered rows are land side.

[0090] After accumulating historical records under different operating conditions, the operating conditions of the current task at the work point can be used to filter the data layer by layer, from coarse to fine, until the finest layer is reached, or until no further filtering is possible. Then, the last layer of historical records is taken for the mode calculation. The specific content of each layer of filtering conditions can be set and adjusted according to the actual situation of the loading and unloading equipment and the dock.

[0091] For example, for ship loading and unloading equipment, the first layer of filtering conditions is: operation mode and equipment performance; the second layer of filtering conditions is: loading and unloading depth and loading and unloading span; the third layer of filtering conditions is: weather influence; and the finest fourth layer of filtering conditions is: driver skills and operation time.

[0092] For example, for yard loading and unloading equipment, the first layer of filtering conditions is: operation mode and equipment performance; the second layer of filtering conditions is: loading and unloading span; the third layer of filtering conditions is: weather influence; and the finest fourth layer of filtering conditions is: driver skills and operating time.

[0093] When the mode cannot be calculated, the default value is used instead. When multiple modes appear, the average value is taken and the working conditions are further optimized and adjusted.

[0094] In some embodiments, predicting the planned work period for the current task by combining the estimated work duration includes:

[0095] If the current task is a loading operation, the operation point Actor calls the fastest travel path search service to perform forward spatiotemporal extrapolation to obtain the planned operation time period of the unloading point;

[0096] If the current task is an unloading operation, the operation point Actor calls the fastest travel path search service to perform reverse spatiotemporal simulation to obtain the planned operation time period of the loading point.

[0097] Based on the above embodiments, the fastest travel path search method mentioned in patent CN120489165A, "A Method and Device for Searching the Fastest Travel Path for Terminal Vehicles," is used to estimate the estimated travel time. To ensure the continuity of operations at the loading point, the fastest travel path is searched forward from the planned completion time of the loading point to calculate the estimated travel time and thus estimate the planned start time of the unloading point. The planned start time plus the estimated unloading time equals the planned completion time of the unloading point, and the calculated result represents the planned operating period of the unloading point.

[0098] To ensure operational continuity at the unloading point, the fastest route is determined by reverse searching starting from the earliest arrival time within the designated time window. The estimated travel time is then calculated, leading to the projected completion time at the loading point. Subtracting the estimated loading time from the projected completion time gives the projected start time at the loading point, thus calculating the planned operating period for that point.

[0099] like Figure 2 As shown, for container terminals, yard-to-ship operations can occur in two scenarios: one truck with one container or one truck with two containers. The latter often requires passing through two loading points. In addition, before loading onto the ship, it may be necessary to pass through a locking station (a centralized lock installation and removal point) to install twist locks. Taking this complex scenario as an example, an empty vehicle loads the first small container at the first loading point in the yard, loads the second small container at the second loading point, installs twist locks at the locking station, and finally arrives at the unloading point on the shore to complete the loading operation. Following this route and reverse sequence, first, based on the planned operating hours of the unloading point at the shore, determine the designated time window. Starting from the earliest arrival time within the designated time window, perform a reverse time-space calculation to determine the estimated third travel time from the lock station to the unloading point at the shore. Subtracting the earliest arrival time within the designated time window gives the planned completion time of the lock station. Subtracting the estimated locking time (the default fixed time) gives the planned start time of the lock station, thus obtaining the planned operating hours and designated time window of the lock station. Continuing with the reverse time-space calculation from the earliest arrival time within the designated time window of the lock station, determine the estimated second travel time from the second loading point to the lock station. Subtracting the designated time window... The earliest arrival time of the time window is the planned completion time of the second loading point. Subtracting the estimated loading time of the second loading point gives the planned start time of the second loading point, thus deriving the planned operating period and designated time window of the second loading point. Finally, by working backward from the earliest arrival time of the designated time window of the second loading point, the estimated first travel time from the first loading point to the second loading point is calculated. Subtracting the earliest arrival time of the designated time window gives the planned completion time of the first loading point. Subtracting the estimated loading time of the first loading point gives the planned start time of the first loading point, thus deriving the planned operating period and designated time window of the first loading point. In this way, the time cycle for completing the task includes: the designated time window and estimated loading time of the first loading point, the estimated first travel time, the designated time window and estimated loading time of the second loading point, the estimated second travel time, the designated time window and estimated locking time of the locking station, the estimated third travel time, and the designated time window and estimated unloading time of the quay unloading point.

[0100] In some embodiments, during the spatiotemporal simulation, if the calculated planned operation period conflicts with existing tasks, the priorities of the current task and the conflicting task are compared:

[0101] If the current task has a higher priority, the planned operation period of the conflicting task and its subsequent tasks will be postponed, and a new time window will be dynamically assigned to the conflicting task and its subsequent tasks, and the task will be re-matched.

[0102] If the conflicting task has a higher priority, the current task is placed after the conflicting task, the planned operation period is recalculated, the planned operation period of the current task's subsequent tasks is postponed, and a new time window is dynamically assigned to the current task and its subsequent tasks, and the task is re-matched.

[0103] In conjunction with the above embodiments, when a calculated planned work period conflicts with an existing task, a priority comparison process is initiated. If the current task has a higher priority, the overall planned time sequence of the conflicting task and all its subsequent tasks is shifted backward, and the planned work periods of these tasks are recalculated. At the same time, the time constraints are updated through a dynamic time window mechanism, and a task rematching process is triggered to ensure that vehicle resources are re-adapted. If the conflicting task has a higher priority, the current task is placed after the conflicting task, the planned work period of the current task is recalculated, and the time sequence of all its subsequent tasks is correspondingly postponed. At the same time, a new time window is generated for the current task and its subsequent tasks, and a rematching process is triggered. Through this priority-based recursive time sequence adjustment strategy, the system achieves dynamic optimization of task scheduling while maintaining the continuity of the work chain.

[0104] In some embodiments, the application of a negative feedback control strategy to dynamically adjust the specified time window for subsequent tasks includes:

[0105] Based on the deviation probability distribution, the range of the deviation probability distribution is obtained;

[0106] When the range is less than the span of the specified time window, the specified time window is slid along the opposite direction of the deviation by an offset corresponding to the deviation amount, based on the deviation between the median of the deviation probability distribution and the target median value.

[0107] When the range is greater than or equal to the span, the specified time window is slid along the opposite direction of the deviation by an offset corresponding to the deviation amount, based on the deviation between the minimum value of the probability distribution and the target minimum value.

[0108] In practice, precise negative feedback control is implemented by comparing the range of the probability distribution of the actual arrival time deviation with the span of a specified time window.

[0109] When the range is less than the span of the specified time window, such as Figure 3 As shown, the arrival times of vehicles are relatively concentrated but there is a systematic deviation. In this case, the deviation between the median of the probability distribution (the value in the middle after arranging the deviation values ​​in order of magnitude) and the target median value can be calculated, and the specified time window can be slid along the opposite direction of the deviation by the corresponding offset (i.e., using the negative value of the deviation) to make the arrival time tend to the median value.

[0110] Specifically, the deviation (in seconds) is calculated by subtracting the median from the median. The opposite of this deviation is the offset by which the specified time window slides in the opposite direction to the target. In other words, if there is a high probability of arriving earlier, the specified time window is slideped later, and vice versa. By sliding the time window—that is, dynamically specifying the time window—the vehicle's arrival time can be controlled within the target range.

[0111] When the range is greater than or equal to the span, such as Figure 4 As shown, the dispersion of vehicle arrival time has exceeded the preset tolerance, which will inevitably lead to early arrivals and late arrivals. For early arrivals, vehicles can wait nearby as appropriate. For late arrivals, it is necessary to solve the problem by making the minimum value consistent with the target minimum value.

[0112] Specifically, the deviation (in seconds) is obtained by subtracting the minimum target value from the minimum value. The opposite of the deviation is the offset (in seconds) of the specified time window sliding in the opposite direction to the target. By sliding the time window, that is, dynamically specifying the time window, the phenomenon of vehicle lateness can be controlled to be limited to extremely rare events.

[0113] In some embodiments, the method further includes, before adjusting based on the deviation between the minimum and the target minimum of the probability distribution:

[0114] Based on the median absolute deviation, extreme values ​​in the deviation probability distribution are identified and eliminated to obtain the non-extreme minimum value of the deviation.

[0115] In conjunction with the above embodiments, to reduce the sensitivity of the negative feedback control strategy to extreme, random probability events, it is necessary to identify and remove extreme values ​​in the statistical data based on MAD (Median Absolute Deviation). The judgment criterion is: |Data Point - Median| > Multiple * MAD. The multiple is defaulted to 3 * 1.4826 (multiplying by a constant factor of 1.4826 assumes a normal probability distribution), and can be set and adjusted according to the actual situation of the terminal. The MAD is calculated as follows: first, calculate the absolute deviation of each data point from the median, then calculate the median of these absolute deviations, which is the MAD. Multiple * MAD is the threshold; values ​​exceeding the threshold are judged as extreme values. During processing, the minimum value is taken from the statistical data, and it is determined whether the absolute value of its difference from the median exceeds the threshold. If it does, it is removed, and a new minimum value is taken for judgment until a minimum value within the threshold is found.

[0116] For example, assuming the probability follows a normal distribution, a default multiple can be used. If the calculated MAD value is 6.5, the default threshold is 3 * 1.4826 * 6.5, which is 28.9107. Then, the judgment starts from the minimum value. Assuming the minimum value is -30, its absolute value 30 is greater than 28.9107, so it is identified as an extreme value and is eliminated. After that, a new minimum value is selected for judgment until a minimum value within -28.9107 is found.

[0117] As the judgment criteria indicate, the multiples in the formula are empirical values, and multiplying by a constant factor (1.4826) is to ensure that MAD is comparable to the standard deviation scale under a normal distribution. These experiences and assumptions can easily lead to misjudgments in practical applications. Therefore, it is necessary to improve the judgment criteria and supplement additional threshold judgment logic to better suit actual business scenarios. For example, if the minimum value is generated by the Nth most recent task (defaulting to the 1st), it can be used as the minimum value within the threshold. For flexibility, an additional threshold judgment method is implemented by an external plugin. This plugin is embedded into the job point Actor through a standard interface, allowing the job point Actor to call it and make the judgment based on the returned result.

[0118] Regardless of whether the work site is an unloading point or a loading point, this application can arrange tasks according to the requirements of continuous operation by estimating the expected operation time and calculating the planned operation period. It also allows for dynamic and continuous adjustment to match more suitable vehicles. In response to the chaotic phenomenon caused by its own limitations and external factors, it uses the method of vehicles waiting nearby in a timely manner and the method of dynamically specifying time windows by negative feedback control strategy to ensure that the loading and unloading equipment at the work site can operate continuously, and realize the coordinated operation of vehicles and loading and unloading equipment throughout the site.

[0119] In some embodiments, the following is further included before applying the negative feedback control strategy:

[0120] The statistical data used to construct the deviation probability distribution are weighted using a time decay weighting algorithm to improve the sensitivity to recent deviation trends.

[0121] Specifically, before applying the negative feedback control strategy, the statistical data is weighted using a time decay weighting algorithm to improve the sensitivity to recent trends.

[0122] The core of the algorithm is to define a weight function w(t) that decreases over time, which can be either an exponential decay function or a linear decay function. This invention uses the former, causing the weight to decrease exponentially over time, as shown in the formula: ;in: It is the attenuation factor The decay rate is controlled by the weight; the larger the weight, the faster the decay. t represents the time interval between the data generation time and the current time, accurate to the second. The decay factor λ can be indirectly adjusted by setting the half-life T_half according to the actual situation of the dock. The relationship between the half-life T_half and the decay factor λ is: T_half = ln(2) / λ.

[0123] For example, the half-life can be set to 30 minutes, and then converted to seconds to calculate the corresponding λ. Suppose a data point was generated 30 minutes ago, and its calculated deviation value is within the statistical range and was calculated only by it. Then the number of occurrences of this deviation value is 1, and after weighting, it is adjusted to 0.5.

[0124] This application dynamically statistically analyzes the probability distribution of vehicle arrival deviation, and uses a time decay weighted algorithm and a self-developed negative feedback control strategy to adaptively slide the operation time window.

[0125] In some embodiments, the method for coordinated operation of terminal vehicles and loading / unloading equipment further includes:

[0126] Waiting spaces should be set up on the side of the lane closest to the work site;

[0127] During task execution, the work point Actor determines whether the vehicle can reach the work point within the specified time window based on the task feedback message.

[0128] If the arrival time is earlier than the specified time window, the estimated waiting time is calculated, and a waiting message is sent to the corresponding vehicle Actor. The waiting message includes at least: waiting position information and waiting time.

[0129] If the vehicle arrives later than the specified time window and there are subsequent tasks, it is determined whether the order can be changed. If so, the Actor at the work point recalculates the planned work period and regenerates the specified time window for the vehicle to arrive at the work point. The task rematch message is then sent to the vehicle work nearest connection service to trigger the vehicle rematch process.

[0130] In practice, the estimated operation time based on mode statistics and the planned operation period calculated based on spatiotemporal extrapolation are subject to fluctuations due to their own limitations and external factors, inevitably leading to deviations between the plan and reality, and exhibiting randomness. Although their calculation accuracy can increase over time, and deviations can be detected and recalculated and adjusted in real time to achieve the effect of "rolling optimization and global decision-making," the room for maneuver decreases as the vehicle gets closer to the destination. Therefore, it is necessary to set up several waiting spaces on the side of the lane closest to the operation point that do not obstruct the passage of other vehicles, allowing early-arriving vehicles to temporarily stop and wait, in order to mitigate the chaos phenomenon.

[0131] Additionally, vehicles that have completed their tasks, or those that have finished other tasks (such as refueling / charging) or other situations (such as driver rest / meal breaks) and are now in standby mode but have not yet been assigned a task, need to find the nearest waiting area to park. As time progresses, they will become vehicles capable of reaching a loading point within a specified time window and will be assigned a task. This is especially important in situations where there is a nearby loading point but arriving too early allows them to wait in the nearest waiting area in exchange for a chance to be assigned a task nearby.

[0132] It should be noted that the waiting area of ​​this invention is different from the waiting area conventionally set up at the dock in terms of underlying logic. Although their functions are similar, both being responsible for "water storage" and "peak shaving", this application is an auxiliary plan based on refined management under a specified time window, while the waiting area conventionally set up at the dock is a routine operation based on extensive management under the operation route mode.

[0133] This application abandons the traditional work route model. In the ever-changing and complex production environment, it achieves coordinated operation of all vehicles and loading and unloading equipment by estimating the expected operation time, calculating the planned operation period, allowing vehicles to wait nearby in a timely manner, and dynamically specifying time windows. This avoids interruption of loading and unloading operations, shortens the time ships spend in port, improves the overall operational efficiency and service quality of the terminal, maximizes vehicle utilization, and completely solves the problem of empty runs.

[0134] The following is another example of a collaborative operation method between dock vehicles and loading / unloading equipment, which is a schematic illustration combining the previous examples.

[0135] Step S1: Build the Actor distributed parallel computing model and deploy it to the Actor service container:

[0136] S1.1, Build a fastest driving route search service according to the fastest driving route search method, and encapsulate the relevant functional modules and their Actor instances:

[0137] S1.1.1, Subscribe to and receive actual / expected vehicle entry and / or exit messages, and dynamically refresh the spatiotemporal simulation basic data (dense volume fluctuation time series record list and feature combination count record list).

[0138] S1.1.2 provides an API service for searching the fastest driving route. Based on the vehicle's planned driving information, it performs spatiotemporal extrapolation to search for the fastest driving route (the IDs of each lane and intersection along the route and their estimated entry and exit times), the planned departure time / estimated arrival time, and returns the results.

[0139] S1.2, Build a vehicle operation proximity connection service according to the vehicle operation proximity connection method, and encapsulate the relevant functional modules and their Actor instances:

[0140] S1.2.1, Subscribe to and receive vehicle messages and task messages, sort out the relationship and weight of vehicles and tasks, build and refresh weighted bipartite graphs, and dynamically control and adjust the number and size of bipartite graphs;

[0141] S1.2.2 Select the matching relationship with the shortest global empty driving time from the bipartite graph to resolve the problem of competition for matching vehicles between multiple graphs, and publish the matching result message using the task ID as the message identifier;

[0142] S1.3, Define the work point Actor, such as Figure 5 As shown, all work point entities on the dock roadway are mapped to work point Actor instances, marked with a work point ID, and defined with, but not limited to, the following attributes. These attributes are initialized by retrieving data from the database when the work point Actor is activated by the managed service, and written to the database when there are updates:

[0143] S1.3.1, Define the lane ID;

[0144] S1.3.2, Define a list of nearby waiting bit IDs, sorted by distance from nearest to farthest;

[0145] S1.3.3 defines a task feedback message list to store task feedback messages pushed from the outside and the dynamic tracking of its own tasks. The task feedback message includes, but is not limited to: task ID, task status (pending assignment / assigned / in execution / completed / cancelled), planned operation period, specified time window, vehicle ID, actual arrival time, single operation duration, and operation task message. The operation task message includes, but is not limited to: task ID, priority, planned start time (if empty, it is a seamless connection to the previous task), task type (ship to site / site to ship / ship to ship / site to site), operation type (loading operation / unloading operation), operation conditions, and an array of assisting operation tasks (sorted by operation points along the way).

[0146] S2, subscribes to, receives, and processes job task messages at this job point:

[0147] S2.1, the task message is only pushed to the task point that needs to be continuously operated. In general, it is pushed to the loading point, and only in special cases is it pushed to the unloading point. For example, the task from the yard to the ship is pushed to the unloading point on the shore when it is necessary to ensure the continuity of shore operations. The specific task point to which it is pushed is determined by the external pusher.

[0148] S2.2 When a job point Actor is activated by the managed service, it begins to subscribe to job task messages with the job point ID as the message identifier.

[0149] S2.3, continuously listen for job task messages pushed from the outside, and trigger and execute the event of the Actor at this job point when the job task message is received upon receipt;

[0150] S2.4 defines the event that the job point Actor receives when it receives a job task message:

[0151] S2.4.1, using the job task message as a parameter, asynchronously call the method of scheduling job tasks of the current job point Actor;

[0152] S2.4.2, End this event.

[0153] S3: Subscribes to, receives, and processes task feedback messages at this work point.

[0154] S3.1 When a job point Actor is activated by the managed service, it begins to subscribe to task feedback messages identified by the message ID of this job point;

[0155] S3.2, continuously listen for task feedback messages pushed from the outside, and trigger and execute the event of the Actor at this job point when the task feedback message is received upon receipt;

[0156] S3.3 defines the event that occurs when a job point Actor receives a task feedback message:

[0157] S3.3.1, Locate the current task record in the task feedback message list;

[0158] S3.3.2 Determine whether the vehicle will arrive early. If so, calculate the estimated waiting time and use the vehicle ID as a message identifier to publish a waiting message (work point ID, estimated waiting time, and list of waiting position IDs) to the outside world to remind the vehicle driver / vehicle equipment.

[0159] S3.3.3, determine whether there will be a delay. If so, determine whether there are subsequent tasks. If so, determine whether the order can be changed. If so, recalculate the planned operation period and dynamically specify the time window. Use the task ID as the message identifier to publish the task message (including the specified time window) to the vehicle operation for the nearest connection service.

[0160] S3.3.4, refresh the task feedback message list;

[0161] S3.3.5, End this event.

[0162] S4, Method for executing the scheduled work tasks at this work point:

[0163] S4.1, retrieve the task ID, priority, planned start time, task type, job type, job conditions, and assist task array from the job task message with input parameters;

[0164] S4.2 If there is a task with the same task ID in the task feedback message list, it is considered that a data conflict has occurred. The exception message is sent to the outside world with the task ID as the message identifier to resolve the data conflict problem, and the following steps are not continued.

[0165] S4.3 Determine the type of loading / unloading equipment at the current work point based on the task type and operation type:

[0166] S4.3.1, if the starting point in the task type is a ship and the operation type is loading operation, then it refers to ship loading and unloading equipment;

[0167] S4.3.2, if the starting point in the task type is yard and the operation type is loading operation, then it is yard loading and unloading equipment;

[0168] S4.3.3, if the destination in the task type is a ship and the operation type is unloading operation, then it refers to ship loading and unloading equipment;

[0169] S4.3.4, if the destination in the task type is yard and the operation type is unloading operation, then it refers to yard loading and unloading equipment;

[0170] S4.4, Execute the method for estimating the estimated operation time at this operation point, input the operation conditions and loading / unloading equipment type, and obtain the estimated operation time of the current task;

[0171] S4.5, retrieve the tasks from the task feedback message list whose task status is "pending assignment / assigned / in execution" and whose planned job time period is the latest as follow-up tasks;

[0172] S4.6, Calculate the planned operation period for the current task:

[0173] S4.6.1, if the input planned start time is empty, then:

[0174] S4.6.1.1, If ​​there is a follow-up task, the planned start time of the planned work period shall be the planned completion time of the follow-up task;

[0175] S4.6.1.2 If no task is followed, it is considered that a data conflict has occurred. An exception message is sent to the outside world using the task ID as the message identifier, requesting the planned start time. The following steps are not continued.

[0176] S4.6.2, if the input planned start time is not empty, then:

[0177] S4.6.2.1, If ​​the planned start time is later than or equal to the planned completion time of the following task, then the planned start time of the planned work period is the planned start time passed in.

[0178] S4.6.2.2 If the planned start time is earlier than the planned completion time of the following task, a data conflict is considered to have occurred. An exception message is sent to the outside world using the task ID as the message identifier to request a reasonable planned start time, and the following steps are not continued.

[0179] S4.6.3, the planned completion time for a planned work period is the planned start time plus the estimated work duration;

[0180] S4.7 Calculate the target within the specified time window. The earliest arrival time is the planned start time of the planned work period minus the estimated work duration, and the latest arrival time is the planned start time of the planned work period.

[0181] S4.8, Executes the method of dynamically specifying the time window at this job point, passing in the specified time window target, and obtains the specified time window of the current task;

[0182] S4.9 If the job type is loading job, then iterate forward through the array of assisting job tasks, remotely call the API service to search for the fastest driving route to find the fastest driving route and the estimated arrival time, remotely call the method to add assisting job tasks at the assisting job points, and forward calculate the estimated driving time and the fastest driving route between each job point, as well as the planned working period and specified time window at the job point, until the last unloading point.

[0183] S4.10 If the job type is unloading, then iterate backward through the array of assisting job tasks, remotely call the API service to search for the fastest driving route to find the fastest driving route and the planned departure time, remotely call the method to add assisting job tasks at the assisting job points, and reversely calculate the estimated driving time and the fastest driving route between each job point, as well as the planned working period and specified time window at the job point, until the earliest loading point.

[0184] S4.11, refresh the task feedback message list;

[0185] S4.12, End this method.

[0186] S5, Method for estimating the expected operation time at this work point:

[0187] S5.1, retrieve the operating conditions and loading / unloading equipment type from the input parameters;

[0188] In S5.2, an external plugin is responsible for filtering statistical data. The plugin is embedded into the job point actor through a standard interface, allowing the job point actor to call it and implement the following functions:

[0189] S5.2.1 Select the corresponding filter using the input loading and unloading equipment type, that is, determine which filtration conditions are used for each layer;

[0190] S5.2.2, decompose the incoming job conditions into the content of each layer of filter conditions;

[0191] S5.2.3, use filtering conditions to filter the historical records in the task feedback message list whose task status is "executed" layer by layer from coarse to fine until the finest layer, or until no more can be filtered, and return the filtering results;

[0192] S5.3 performs mode statistics on the statistical data filtered by the plug-in. When multiple modes appear, the average value is taken. If no mode can be filtered, the configured default value is used instead.

[0193] S5.4, take the mode statistical result as the estimated operation time and return the result;

[0194] S5.5, End of this method.

[0195] S6, Method for dynamically specifying time windows at this work point:

[0196] S6.1, retrieve the target for the specified time window from the input parameters and calculate the span of the specified time window;

[0197] S6.2 Extract the most recent consecutive jobs with a task status of "completed" from the task feedback message list as statistical data;

[0198] S6.3 uses a time decay weighting algorithm to weight statistical data to improve sensitivity to recent trends;

[0199] S6.4, Perform probability distribution statistics on the number of times the actual arrival time deviation value (seconds) occurs on the weighted statistical data;

[0200] S6.5, if the range of the probability distribution is less than the specified time window span, then:

[0201] S6.5.1, Subtract the target median value for the specified time window from the median of the probability distribution to obtain the deviation (in seconds).

[0202] S6.5.2, slide the specified time window in the opposite direction of the deviation (seconds) to the specified time window and return the result;

[0203] S6.6, If the range of the probability distribution is greater than or equal to the specified time window span, then:

[0204] S6.6.1, based on MAD, identifies and removes extreme values ​​in statistical data. For flexibility, an additional threshold judgment method is implemented by an external plugin. The threshold judgment plugin is embedded into the job point Actor through a standard interface, which is called by the job point Actor and makes a judgment based on the returned result.

[0205] S6.6.2, Subtract the target minimum value within the specified time window from the minimum value to obtain the deviation (in seconds).

[0206] S6.6.3, slide the specified time window in the opposite direction of the deviation (seconds) to the specified time window and return the result.

[0207] S6.7, End of this method.

[0208] S7, call the method to add an assistance task at the assistance task point:

[0209] S7.1, retrieve the job task message and estimated arrival time / planned departure time from the input parameters;

[0210] S7.2, if the input is the estimated arrival time, it indicates that a forward spacetime deduction is being performed, then:

[0211] S7.2.1, Execute the method for estimating the estimated operation time at this work point, input the operation conditions and loading / unloading equipment type, and obtain the estimated operation time of the current task;

[0212] S7.2.2 Calculate the planned work period for the current task. The planned start time is the estimated arrival time, and the planned completion time is the planned start time plus the estimated work duration.

[0213] S7.2.3, If the planned work period conflicts with other existing tasks, the priorities of the conflicting tasks are compared:

[0214] S7.2.3.1 If the current task has a higher priority, the planned operation period of the conflicting task and its subsequent tasks will be postponed, the time window will be dynamically reassigned, and the task IDs will be used as message identifiers to publish task messages (including the specified time window) to the vehicle operation nearest connection service.

[0215] S7.2.3.2 If the conflicting task has a higher priority, the current task will be placed after the conflicting task to recalculate the planned operation period. The planned operation period of subsequent tasks will also be postponed, and a new time window will be dynamically specified. The task IDs of these tasks will be used as message identifiers to publish task messages (including the specified time window) to the vehicle operation nearest connection service.

[0216] S7.2.4 Calculate the target within a specified time window. The earliest arrival time is the planned start time of the planned work period minus the estimated work duration, and the latest arrival time is the planned start time of the planned work period.

[0217] S7.2.5, Execute the method of dynamically specifying the time window at this job point, pass in the specified time window target, and obtain the specified time window of the current task;

[0218] S7.3, if the input is the planned departure time, it indicates that a reverse spacetime simulation is being performed, then:

[0219] S7.3.1, Execute the method for estimating the estimated operation time at this operation point, input the operation conditions and loading / unloading equipment type, and obtain the estimated operation time of the current task;

[0220] S7.3.2, calculate the planned work period for the current task. The planned start time is the planned departure time minus the estimated work duration, and the planned completion time is the planned departure time.

[0221] S7.3.3, If the planned work period conflicts with other existing tasks, the priorities of the conflicting tasks are compared:

[0222] S7.3.3.1 If the current task has a higher priority, the planned operation period of the conflicting task and its subsequent tasks will be postponed, the time window will be dynamically reassigned, and the task IDs will be used as message identifiers to publish task messages (including the specified time window) to the vehicle operation nearest connection service.

[0223] S7.3.3.2 If the conflicting task has a higher priority, the current task will be placed after the conflicting task to recalculate the planned operation period. The planned operation period of subsequent tasks will also be postponed, and a new time window will be dynamically specified. The task IDs of these tasks will be used as message identifiers to publish task messages (including the specified time window) to the vehicle operation nearest connection service.

[0224] S7.3.4 Calculate the target within a specified time window. The earliest arrival time is the planned start time of the planned work period minus the estimated work duration, and the latest arrival time is the planned start time of the planned work period.

[0225] S7.3.5, Executes the method of dynamically specifying the time window at this job point, passing in the specified time window target, and obtains the specified time window of the current task;

[0226] S7.4 returns the planned work period and specified time window for the current task at this assisted work point;

[0227] S7.5, End of this method.

[0228] S8, After completing the above, publish the task message:

[0229] S8.1, using the current task ID as the message identifier, publish a task message (including a specified time window) to the vehicle operation nearest connection service;

[0230] S8.2, based on the fastest driving route found, use the IDs of each lane and intersection along the route as message identifiers to publish the vehicle's expected entry and exit messages to the fastest driving route search service;

[0231] S8.3, the Actor at this work point enters standby mode;

[0232] The external system drives the Actor entities deployed in the Actor service container through push messages. These Actor entities then collaboratively process the messages and feed back the final confirmed task orchestration results to the external system.

[0233] Based on the same inventive concept, this application also provides a device for coordinated operation of dock vehicles and loading / unloading equipment, comprising:

[0234] The model building and deployment module is used to build the Actor distributed parallel computing model and deploy it to the Actor service container, mapping the work points and vehicles in the dock to work point Actor and vehicle Actor, respectively.

[0235] The task duration estimation module is used to receive task messages, parse the task conditions of the current task, filter the corresponding historical task records according to the task conditions and perform mode statistics to obtain the estimated task duration of the current task.

[0236] The work period prediction module is used to predict the planned work period of the current task based on the planned completion time of the previous work point or the planned start time of the next work point, combined with the estimated work duration, so that the planned completion time of the previous work point and the planned start time of the next work point are continuously connected.

[0237] The task matching module is used to generate a specified time window for vehicles to arrive at the work point according to the planned work period. The work point Actor publishes a task message to the vehicle work proximity connection service, and the vehicle work proximity connection service matches the task to the corresponding vehicle Actor and pushes the task message. The task message includes at least the specified time window.

[0238] The dynamic adjustment module is used to continuously receive task feedback messages from the vehicle Actor, wherein the task feedback messages include at least the actual arrival time of the vehicle;

[0239] Based on the task feedback message, the deviation value of the actual arrival time relative to the specified time window is obtained, and a deviation probability distribution is constructed based on the deviation value;

[0240] Based on the aforementioned deviation probability distribution, a negative feedback control strategy is applied to dynamically adjust the specified time window for subsequent tasks.

[0241] In some embodiments, the dock vehicle and loading / unloading equipment collaborative operation device further includes: a scheduling module;

[0242] The scheduling module is used to determine whether the vehicle can reach the work point within the specified time window based on the task feedback message during task execution.

[0243] If the arrival time is earlier than the specified time window, the estimated waiting time is calculated, and a waiting message is sent to the corresponding vehicle Actor. The waiting message includes at least: waiting position information and waiting time.

[0244] If the vehicle arrives later than the specified time window and there are subsequent tasks, it is determined whether the order can be changed. If so, the Actor at the work point recalculates the planned work period and regenerates the specified time window for the vehicle to arrive at the work point. The task rematch message is then sent to the vehicle work nearest connection service to trigger the vehicle rematch process.

[0245] This application dynamically schedules planned operation tasks for single-machine loading and unloading equipment according to priority by subscribing to, receiving and processing operation task messages and task feedback messages. It supports not only continuous operation at loading points, but also continuous operation at unloading points, especially continuous operation of shore-side loading and unloading equipment.

[0246] This application organically combines the fastest travel path search method and the vehicle operation proximity connection method. While taking into account both operation continuity and vehicle utilization, it adaptively controls the coordinated operation of dock vehicles and loading and unloading equipment to achieve dynamic controllability and optimal balance.

[0247] This application is based on the Actor distributed parallel computing model, which builds job point Actors under a distributed architecture to realize distributed scheduling of job tasks.

[0248] This application abstracts and summarizes the work points and their tasks, decomposes the tasks into the work tasks of the work points and the auxiliary work tasks of the assisting work points, and provides a plug-in interface to scalably introduce unknown statistical logic for different working conditions and loading and unloading equipment types. It is applicable not only to traditional or automated container terminals, but also to bulk cargo and general cargo terminals where goods are transported by vehicles within the terminal.

[0249] This application ensures the absolute continuity of loading and unloading operations at a single work point by dynamically scheduling planned tasks, adaptively sliding the operation time window, and using physical waiting positions for buffering. This can significantly shorten the time ships spend in port and improve the overall operational efficiency and service quality of the terminal.

[0250] In this specification, the same or similar parts between the various embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the descriptions of the embodiments described later are relatively simple, and relevant parts can be referred to the descriptions of the foregoing embodiments.

[0251] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for coordinated operation of dock vehicles and loading / unloading equipment, characterized in that, include: Construct an Actor distributed parallel computing model and deploy it to an Actor service container, mapping the work points and vehicles in the dock to work point Actor and vehicle Actor, respectively; The Actor receives the task message, parses it to obtain the task conditions, filters the corresponding historical task records according to the task conditions and performs mode statistics to obtain the estimated task duration. The work point Actor predicts the planned work period of the current task based on the planned completion time of the previous work point or the planned start time of the next work point, combined with the estimated work duration, so that the planned completion time of the previous work point and the planned start time of the next work point are continuously connected. A designated time window for vehicle arrival at the work site is generated based on the planned work period. The work site Actor publishes a task message to the vehicle work proximity matching service, and the vehicle work proximity matching service matches the task to the corresponding vehicle Actor and pushes the task message. The task message includes at least the designated time window. The work point Actor continuously receives task feedback messages from the vehicle Actor, wherein the task feedback messages include at least: the actual arrival time of the vehicle; Based on the task feedback message, the deviation value of the actual arrival time relative to the specified time window is obtained, and a deviation probability distribution is constructed based on the deviation value; Based on the aforementioned deviation probability distribution, a negative feedback control strategy is applied to dynamically adjust the specified time window for subsequent tasks.

2. The method for coordinated operation of dock vehicles and loading / unloading equipment according to claim 1, characterized in that, Based on the stated task conditions, corresponding historical task records are filtered and the mode is calculated to obtain the estimated task duration for the current task, including: Based on the current task's operating conditions and the type of loading and unloading equipment, determine the corresponding filtration level, wherein the filtration level includes multiple filtration conditions from coarse to fine. The filtering conditions are used to filter the duration of each single task in the historical task records layer by layer. The mode of the final selected single-job duration is statistically analyzed; If a unique mode is obtained from the statistics, then the mode is taken as the estimated operation time; If multiple modes are obtained from the statistics, the average of the modes shall be taken as the estimated operation time. If no historical records are found, the preset default value is used as the estimated job duration.

3. The method for coordinated operation of dock vehicles and loading / unloading equipment according to claim 1, characterized in that, The step of predicting the planned work period for the current task by combining the estimated work duration includes: If the current task is a loading operation, the operation point Actor calls the fastest travel path search service to perform forward spatiotemporal extrapolation to obtain the planned operation time period of the unloading point; If the current task is an unloading operation, the operation point Actor calls the fastest travel path search service to perform reverse spatiotemporal simulation to obtain the planned operation time period of the loading point.

4. The method for coordinated operation of dock vehicles and loading / unloading equipment according to claim 3, characterized in that, During the spatiotemporal simulation, if the calculated planned operation period conflicts with existing tasks, the priorities of the current task and the conflicting task are compared: If the current task has a higher priority, the planned operation period of the conflicting task and its subsequent tasks will be postponed, and a new time window will be dynamically assigned to the conflicting task and its subsequent tasks, and the task will be re-matched. If the conflicting task has a higher priority, the current task is placed after the conflicting task, the planned operation period is recalculated, the planned operation period of the current task's subsequent tasks is postponed, and a new time window is dynamically assigned to the current task and its subsequent tasks, and the task is re-matched.

5. The method for coordinated operation of dock vehicles and loading / unloading equipment according to claim 1, characterized in that, The application of negative feedback control strategy to dynamically adjust the specified time window for subsequent tasks includes: Based on the deviation probability distribution, the range of the deviation probability distribution is obtained; When the range is less than the span of the specified time window, the specified time window is slid along the opposite direction of the deviation by an offset corresponding to the deviation amount, based on the deviation between the median of the deviation probability distribution and the target median value. When the range is greater than or equal to the span, the specified time window is slid along the opposite direction of the deviation by an offset corresponding to the deviation amount, based on the deviation between the minimum value of the probability distribution and the target minimum value.

6. The method for coordinated operation of dock vehicles and loading / unloading equipment according to claim 5, characterized in that, Before adjusting based on the deviation between the minimum and the target minimum of the probability distribution, the following is also included: Based on the median absolute deviation, extreme values ​​in the deviation probability distribution are identified and eliminated to obtain the non-extreme minimum value of the deviation.

7. The method for coordinated operation of dock vehicles and loading / unloading equipment according to claim 5, characterized in that, Before applying the negative feedback control strategy, the following is also included: The statistical data used to construct the deviation probability distribution are weighted using a time decay weighting algorithm to improve the sensitivity to recent deviation trends.

8. The method for coordinated operation of terminal vehicles and loading / unloading equipment according to any one of claims 1-7, characterized in that, The method for coordinated operation of terminal vehicles and loading / unloading equipment also includes: Waiting spaces should be set up on the side of the lane closest to the work site; During task execution, the work point Actor determines whether the vehicle can reach the work point within the specified time window based on the task feedback message. If the arrival time is earlier than the specified time window, the estimated waiting time is calculated, and a waiting message is sent to the corresponding vehicle Actor. The waiting message includes at least: waiting position information and waiting time. If the vehicle arrives later than the specified time window and there are subsequent tasks, it is determined whether the order can be changed. If so, the Actor at the work point recalculates the planned work period and regenerates the specified time window for the vehicle to arrive at the work point. The task rematch message is then sent to the vehicle work nearest connection service to trigger the vehicle rematch process.

9. A device for coordinated operation of dock vehicles and loading / unloading equipment, characterized in that, include: The model building and deployment module is used to build the Actor distributed parallel computing model and deploy it to the Actor service container, mapping the work points and vehicles in the dock to work point Actor and vehicle Actor, respectively. The task duration estimation module is used to receive task messages, parse the task conditions of the current task, filter the corresponding historical task records according to the task conditions and perform mode statistics to obtain the estimated task duration of the current task. The work period prediction module is used to predict the planned work period of the current task based on the planned completion time of the previous work point or the planned start time of the next work point, combined with the estimated work duration, so that the planned completion time of the previous work point and the planned start time of the next work point are continuously connected. The task matching module is used to generate a specified time window for vehicles to arrive at the work point according to the planned work period. The work point Actor publishes a task message to the vehicle work proximity connection service, and the vehicle work proximity connection service matches the task to the corresponding vehicle Actor and pushes the task message. The task message includes at least the specified time window. The dynamic adjustment module is used to continuously receive task feedback messages from the vehicle Actor, wherein the task feedback messages include at least the actual arrival time of the vehicle; Based on the task feedback message, the deviation value of the actual arrival time relative to the specified time window is obtained, and a deviation probability distribution is constructed based on the deviation value; Based on the aforementioned deviation probability distribution, a negative feedback control strategy is applied to dynamically adjust the specified time window for subsequent tasks.

10. The dock vehicle and loading / unloading equipment collaborative operation device according to claim 9, characterized in that, The dock vehicle and loading / unloading equipment collaborative operation device also includes: a scheduling module; The scheduling module is used to determine whether the vehicle can arrive at the work point within the specified time window based on the task feedback message during task execution. If the arrival time is earlier than the specified time window, the estimated waiting time is calculated, and a waiting message is sent to the corresponding vehicle Actor. The waiting message includes at least: waiting position information and waiting time. If the vehicle arrives later than the specified time window and there are subsequent tasks, it is determined whether the order can be changed. If so, the Actor at the work point recalculates the planned work period and regenerates the specified time window for the vehicle to arrive at the work point. The task rematch message is then sent to the vehicle work nearest connection service to trigger the vehicle rematch process.

Images