An artificial intelligence-based waterway scheduling optimization method

Abstract

The application discloses a kind of channel scheduling optimization methods based on artificial intelligence, it is related to the field of shipping management, this method includes: target channel is dynamically collected according to channel scheduling period, obtains channel multi-source dynamic data set and carries out spatiotemporal fusion processing, generates channel spatiotemporal state matrix;Channel conflict detection is carried out, determines channel traffic conflict parameter, carries out two-way spatiotemporal planning to target channel, and formulates initial scheduling scheme;Real-time ship dynamic position data is sent to ship intelligent terminal in combination and is rolled to optimize, generates channel dynamic optimization instruction;Execute channel dynamic optimization instruction and carry out scheduling feedback, according to the effect of scheduling, initial scheduling scheme is updated, and the channel dynamic scheduling scheme is determined.The technical problem that the existing channel scheduling optimization cannot adapt to dynamic change and lead to unreasonable scheduling is solved, to dynamically adjust according to real-time situation, effectively reduce channel traffic conflict, improve the technical effect of ship traffic efficiency.

Description

Technical Field

[0001] This application relates to the field of shipping management, and in particular to an artificial intelligence-based method for optimizing waterway scheduling. Background Technology

[0002] Optimizing waterway scheduling is crucial for ensuring efficient and safe passage of vessels and improving overall waterway transport efficiency. Currently, the main method for solving waterway scheduling optimization problems is to formulate scheduling plans based on fixed waterway information and preset rules. However, existing methods lack comprehensive collection and in-depth analysis of real-time dynamic waterway data, and fail to fully consider the impact of real-time dynamic changes in vessel positions on scheduling. This results in scheduling schemes that are difficult to adapt to complex and ever-changing waterway environments, easily leading to problems such as waterway traffic conflicts and excessively long vessel waiting times.

[0003] At present, the relevant technologies for waterway scheduling optimization have technical problems such as being unable to adapt to dynamic changes, resulting in unreasonable scheduling. Summary of the Invention

[0004] This application provides an artificial intelligence-based waterway scheduling optimization method. It employs techniques such as dynamically collecting and fusing target waterway data according to a scheduling cycle to generate a waterway spatiotemporal state matrix, conducting waterway conflict detection, formulating an initial scheduling scheme based on bidirectional spatiotemporal planning using detection parameters, continuously optimizing the initial scheme using real-time vessel dynamic position data, generating dynamic optimization instructions, executing these instructions and providing feedback on the scheduling effect, updating the initial scheme accordingly, and determining the dynamic scheduling scheme. This method solves the technical problem of existing waterway scheduling optimization methods being unable to adapt to dynamic changes, leading to unreasonable scheduling. It achieves the technical effect of dynamically adjusting according to real-time conditions, effectively reducing waterway traffic conflicts, improving vessel traffic efficiency, and optimizing the overall transportation efficiency of the waterway.

[0005] This application provides an artificial intelligence-based waterway scheduling optimization method, comprising: dynamically collecting data on a target waterway according to a waterway scheduling cycle, obtaining a multi-source dynamic dataset of the waterway, performing spatiotemporal fusion processing to generate a waterway spatiotemporal state matrix; performing waterway conflict detection on the waterway spatiotemporal state matrix to determine waterway traffic conflict parameters, performing bidirectional spatiotemporal planning on the target waterway according to the waterway traffic conflict parameters, and formulating an initial scheduling scheme; according to the initial scheduling scheme, combining real-time vessel dynamic position data and sending it to a vessel intelligent terminal for rolling optimization to generate a waterway dynamic optimization command; executing the waterway dynamic optimization command to provide scheduling feedback, generating a scheduling effect, updating the initial scheduling scheme according to the scheduling effect, and determining a waterway dynamic scheduling scheme.

[0006] In a possible implementation, the target waterway is dynamically collected according to the waterway scheduling cycle to obtain a multi-source dynamic dataset of the waterway. The following processing is then performed: the tidal cycle and waterway operation cycle of the target waterway are retrieved, and the tidal cycle and the waterway operation cycle are used as dual driving factors to construct the waterway scheduling cycle; a dynamic collection command is issued based on the waterway scheduling cycle, and the ship data acquisition unit is activated to collect data from the target waterway through the dynamic collection command to determine the real-time ship trajectory data; the waterway data acquisition unit is activated to collect data from the target waterway through the dynamic collection command to determine the waterway facility status data; the meteorological data acquisition unit is activated to collect data from the target waterway through the dynamic collection command to determine the waterway meteorological and hydrological data; the real-time ship trajectory data, the waterway facility status data, and the waterway meteorological and hydrological data are integrated from multiple sources to obtain the multi-source dynamic dataset of the waterway.

[0007] In a possible implementation, the multi-source dynamic dataset of the waterway undergoes spatiotemporal fusion processing to generate a waterway spatiotemporal state matrix. The following processing steps are performed: The real-time vessel trajectory data, waterway facility status data, and waterway meteorological and hydrological data are timestamped according to the waterway scheduling cycle to obtain multiple data time tags; the real-time vessel trajectory data, waterway facility status data, and waterway meteorological and hydrological data are time-aligned according to the multiple data time tags to generate a time alignment result; the target waterway is divided into grids to obtain multiple grid cells, and a grid coordinate system is constructed based on the multiple grid cells; the real-time vessel trajectory data, waterway facility status data, and waterway meteorological and hydrological data are mapped to the grid coordinate system for spatial alignment to generate a spatial alignment result; the real-time vessel trajectory data, waterway facility status data, and waterway meteorological and hydrological data are spatiotemporally fused based on the time alignment result and the spatial alignment result to construct a three-dimensional tensor structure; the fusion quality is verified according to the three-dimensional tensor structure, and when the verification passes, the waterway spatiotemporal state matrix is ​​generated.

[0008] In a possible implementation, the real-time ship trajectory data, the waterway facility status data, and the waterway meteorological and hydrological data are spatiotemporally fused based on the time alignment result and the spatial alignment result to construct a three-dimensional tensor structure. The following processing is then performed: Temporal feature encoding of the real-time ship trajectory data is performed based on the time alignment result and the spatial alignment result to construct a ship position probability distribution layer; spatial relationship modeling of the waterway facility status data is performed based on the time alignment result and the spatial alignment result, and the waterway saturation coefficient is calculated based on the modeling results to obtain a waterway congestion index layer; waterway meteorological risk analysis is performed on the waterway meteorological and hydrological data based on the time alignment result and the spatial alignment result to set a meteorological risk level layer; the ship position probability distribution layer, the waterway congestion index layer, and the meteorological risk level layer are then data-linked and fused to construct the three-dimensional tensor structure.

[0009] In a possible implementation, the spatiotemporal state matrix of the waterway is used for waterway conflict detection to determine waterway traffic conflict parameters, and the following processing is performed: target vessels within the target waterway are retrieved for kinematic analysis, and a dynamic safety domain radius for the vessels is set; a four-dimensional analysis is performed based on the dynamic safety domain radius of the vessels, the probability distribution layer of the vessel positions, the waterway congestion index layer, and the meteorological risk level layer to construct a four-dimensional vessel trajectory waterway model; based on the four-dimensional vessel trajectory waterway model, the minimum safe distance is calculated using Euclidean distance detection; based on the minimum safe distance, multi-vehicle spatiotemporal conflict detection is performed on the target waterway, and when multiple vessel trajectories have a waterway path smaller than the minimum safe distance, the waterway traffic conflict parameters are generated.

[0010] In a possible implementation, the target waterway is planned in both directions according to the waterway traffic conflict parameters, an initial scheduling scheme is formulated, and the following processes are performed: multiple conflict events are identified based on the waterway traffic conflict parameters, a conflict resolution unit is triggered, and the multiple conflict events are resolved in three dimensions based on the conflict resolution unit to generate multiple solution spaces; a two-way traffic constraint is constructed based on the multiple solution spaces, the target waterway is divided into N logical resource blocks, and a spatiotemporal conflict analysis is performed on the N logical resource blocks according to the two-way traffic constraint to construct a spatiotemporal resource conflict graph, where N is an integer greater than 1; a two-way spatiotemporal resource allocation strategy is constructed based on the spatiotemporal resource conflict graph to perform collaborative planning on the target waterway, and the initial scheduling scheme is formulated.

[0011] In a possible implementation, bidirectional traffic constraints are constructed based on the multiple solution spaces. The target channel is divided into N logical resource blocks. Spatiotemporal conflict analysis is performed on the N logical resource blocks according to the bidirectional traffic constraints to construct a spatiotemporal resource conflict graph. The following processes are performed: the multiple solution spaces are parsed to obtain time solutions, spatial solutions, and velocity solutions; constraint analysis is performed based on the time solutions, spatial solutions, and velocity solutions to construct a three-dimensional solution constraint set; bidirectional traffic matching is performed on the target channel according to the three-dimensional solution constraint set to set bidirectional traffic constraints; the target channel is divided into N logical resource blocks, and the N logical resource blocks are traversed and defined to determine N four-dimensional tuples; conflict detection is performed on the N four-dimensional tuples according to the bidirectional traffic constraints to obtain inter-block conflict relationships; conflict intensity is defined according to the inter-block conflict relationships to generate conflict weight coefficients; the N logical resource blocks are spatiotemporally associated according to the inter-block conflict relationships and the conflict weight coefficients to construct the spatiotemporal resource conflict graph.

[0012] In a possible implementation, according to the initial scheduling scheme, real-time vessel dynamic position data is sent to the vessel's intelligent terminal for rolling optimization, generating a channel dynamic optimization command, and performing the following processing: based on the vessel's intelligent terminal, the real-time vessel dynamic position is mapped to the initial scheduling scheme for channel scheduling analysis to obtain the spatiotemporal deviation; a rolling time-domain control framework is constructed, and the initial scheduling scheme is optimized and identified according to the rolling time-domain control framework and the spatiotemporal deviation to determine a local replanning label; the local replanning label is located to the target channel, and the channel dynamic optimization command is generated.

[0013] In a possible implementation, the local replanning label is located to the target waterway, the waterway dynamic optimization instruction is generated, and the following processes are performed: impact identification is performed according to the local replanning label to generate a set of affected vessels; the unaffected vessel set is retrieved to freeze vessel scheduling and construct a local optimization subspace; the affected vessel set is calculated based on the local optimization subspace to generate a course turning point sequence; the course turning point sequence is mapped to the target waterway to generate the waterway dynamic optimization instruction.

[0014] In a possible implementation, the dynamic channel optimization command is executed to provide scheduling feedback, generating a scheduling effect. Based on the scheduling effect, the initial scheduling plan is updated to determine the dynamic channel scheduling plan. The following processes are then performed: the dynamic channel optimization command is executed to synchronously collect data from the target channel, obtaining a channel operation log; a multi-dimensional scheduling evaluation is performed based on the channel operation log, generating a scheduling effect; reinforcement learning is used to incrementally update the initial scheduling plan according to the scheduling effect, generating a dynamic channel scheduling plan; digital twin verification is performed based on the dynamic channel scheduling plan; and when the digital twin verification confirms the plan's feasibility, the dynamic channel scheduling plan is deployed for execution.

[0015] This application proposes an AI-based waterway scheduling optimization method. First, it dynamically collects data on the target waterway according to the waterway scheduling cycle, obtaining a multi-source dynamic dataset. This dataset is then spatiotemporally fused to generate a waterway spatiotemporal state matrix. Next, the matrix is ​​used for waterway conflict detection to determine waterway traffic conflict parameters. Based on these parameters, bidirectional spatiotemporal planning is performed on the target waterway to formulate an initial scheduling plan. Following this initial plan, real-time vessel dynamic position data is combined and sent to the vessel's intelligent terminal for rolling optimization, generating a dynamic waterway optimization command. Finally, the dynamic optimization command is executed to provide scheduling feedback and generate a scheduling effect. Based on this effect, the initial scheduling plan is updated to determine the final dynamic waterway scheduling plan. This method achieves the technical effect of dynamically adjusting according to real-time conditions, effectively reducing waterway traffic conflicts, improving vessel traffic efficiency, and optimizing the overall transportation efficiency of the waterway. Attached Figure Description

[0016] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings of the embodiments of the present invention will be briefly described below. Flowcharts are used in this application to illustrate the operations performed by the method according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed precisely in sequence. Instead, various steps can be processed in reverse order or simultaneously as needed. Furthermore, other operations can be added to these processes, or one or more steps can be removed from these processes.

[0017] Figure 1 This is a flowchart illustrating an artificial intelligence-based waterway scheduling optimization method provided in an embodiment of this application.

[0018] Figure 2 This is a flowchart illustrating the process of determining waterway traffic conflict parameters in an artificial intelligence-based waterway scheduling optimization method provided in this application embodiment. Detailed Implementation

[0019] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, specific embodiments of this application are given below.

[0020] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description of this application will be provided in conjunction with the accompanying drawings. The described embodiments should not be considered as limitations on this application. All other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0021] In the following description, references to "some embodiments" describe a subset of all possible embodiments; however, it is understood that "some embodiments" may be the same or different subsets of all possible embodiments and may be combined with each other without conflict. The terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or modules not explicitly listed or inherent to such processes, methods, products, or devices. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only.

[0022] This application provides an artificial intelligence-based waterway scheduling optimization method, such as... Figure 1 As shown, the method includes:

[0023] Step S100: Dynamically collect data on the target waterway according to the waterway scheduling cycle, obtain a multi-source dynamic dataset of the waterway, perform spatiotemporal fusion processing, and generate a waterway spatiotemporal state matrix.

[0024] Specifically, various sensors are deployed along the target waterway (such as at waterway entrances, narrow sections, and junctions), including Automatic Identification System (AIS) receivers, hydro-meteorological sensors (such as anemometers, water level gauges, and current meters), and video surveillance equipment. These sensors are used to collect data on vessel position, speed, and waterway hydro-meteorological conditions in real time. For example, an AIS receiver is installed at the entrance of the target waterway to obtain attribute data of vessels entering the waterway (such as vessel type and tonnage).

[0025] Data is collected from the sensor network using dedicated data acquisition software or hardware according to a set waterway scheduling cycle (e.g., every 10 minutes). The data acquisition module can transmit sensor data to the data processing center based on Internet of Things (IoT) technology via wireless communication protocols (such as LoRa, NB-IoT, etc.).

[0026] Computer algorithms are used to perform spatiotemporal fusion of the collected multi-source data. For example, the Kalman filter algorithm is used to smooth the ship position data and eliminate noise; Geographic Information System (GIS) technology is used to spatially correlate waterway topology data with ship position data, hydrological and meteorological data, etc., to generate a waterway spatiotemporal state matrix.

[0027] For example, three AIS receivers are installed on a narrow section of an inland waterway, located on the left, middle, and right sides of the channel, respectively, to accurately acquire vessel positions. Simultaneously, a water level gauge is installed every 500 meters on both sides of the channel to monitor water level changes. The data acquisition module activates the sensors every 5 minutes to collect data. The collected vessel position data is transmitted to the data processing center via the AIS protocol, while the water level data is transmitted via a LoRa network, with a transmission distance of up to 2 kilometers, ensuring timely and accurate arrival at the data processing center. Upon receiving the data, the data processing center uses GIS software to match the vessel position data with the channel topology data. For example, after matching the position data of vessel A with the channel topology data, it is determined that vessel A is located in the middle of the channel, 2 kilometers away from the next junction. Simultaneously, combined with the water level data (2.5 meters) and wind speed data (3 meters per second) at this location, a spatiotemporal state matrix of the channel is generated, containing information such as vessel position, channel topology, water level, and wind speed.

[0028] In one possible implementation, the target waterway is dynamically collected according to the waterway scheduling cycle to obtain a multi-source dynamic dataset of the waterway. Step S100 further includes step S110, retrieving the tidal cycle and waterway operation cycle of the target waterway, and using the tidal cycle and the waterway operation cycle as dual driving factors to construct the waterway scheduling cycle. Specifically, tidal data of the target waterway is obtained from marine observation stations, including the times of high tide and low tide. For example, the tidal cycle of a certain waterway is 12.5 hours, and the high tide occurs at 06:00 and 18:30 every day. Waterway operation data is obtained from the port management system, including ship loading and unloading operation time, waterway maintenance operation time, etc. For example, the loading and unloading operation cycle of a certain waterway is 8 hours, from 08:00 to 16:00 and from 20:00 to 04:00 every day. A mathematical modeling method is used to construct the waterway scheduling cycle using the tidal cycle and the waterway operation cycle as dual driving factors. For example, the least common multiple method can be used to determine a comprehensive scheduling cycle to ensure that the scheduling cycle can simultaneously meet the requirements of tides and operations.

[0029] Step S120: Based on the waterway scheduling cycle, a dynamic acquisition command is issued. This command activates the ship data acquisition unit to collect data from the target waterway, determining the ship's real-time trajectory data. Specifically, an AIS receiver and a ship positioning system (such as GPS) are used as the ship data acquisition unit. These devices can acquire real-time data such as the ship's position, speed, and heading. The data acquisition control system sends a dynamic acquisition command to activate the ship data acquisition unit. For example, a time-triggered command can be used to collect ship data every 5 minutes. After receiving the command, the ship data acquisition unit collects the ship's real-time trajectory data, including position coordinates (longitude, latitude), speed, and heading, and transmits the data to the data processing center via a wireless communication network (such as 4G / 5G). Examples of real-time ship trajectory data are shown in Table 1.

[0030] Table 1: Examples of Real-Time Ship Trajectory Data

[0031]

[0032] Step S130: The channel data acquisition unit is activated by the dynamic acquisition command to collect data from the target channel and determine the channel facility status data. Specifically, channel monitoring equipment (such as water level gauges, current meters, and channel marker sensors) is used as the channel data acquisition unit. These devices can acquire data such as water level, current velocity, and channel marker status in real time. The data acquisition control system sends a dynamic acquisition command to activate the channel data acquisition unit. For example, channel facility status data is collected every 10 minutes. After receiving the command, the channel data acquisition unit collects the channel facility status data and transmits the data to the data processing center via a wireless communication network (such as LoRa). Examples of channel facility status data are shown in Table 2.

[0033] Table 2: Example of Waterway Facility Status Data

[0034]

[0035] Step S140: The meteorological data acquisition unit is activated by the dynamic acquisition command to collect data from the target waterway and determine the waterway's meteorological and hydrological data. Specifically, meteorological stations (including anemometers, wind direction sensors, temperature and humidity sensors, etc.) are used as meteorological data acquisition units. These devices can acquire real-time meteorological and hydrological data of the waterway, such as wind speed, wind direction, temperature, and humidity. The data acquisition control system sends a dynamic acquisition command to activate the meteorological data acquisition unit. For example, meteorological and hydrological data is collected every 15 minutes. After receiving the command, the meteorological data acquisition unit collects the waterway's meteorological and hydrological data and transmits the data to the data processing center via a wireless communication network (such as NB-IoT). Examples of waterway meteorological and hydrological data are shown in Table 3.

[0036] Table 3: Examples of Meteorological and Hydrological Data for Waterways

[0037]

[0038] Step S150 involves integrating the real-time vessel trajectory data, the waterway facility status data, and the waterway meteorological and hydrological data from multiple sources to obtain the waterway multi-source dynamic dataset. Specifically, after receiving various types of data, the data processing center uses a data fusion algorithm (such as the Kalman filter algorithm) to integrate the real-time vessel trajectory data, waterway facility status data, and waterway meteorological and hydrological data from multiple sources. The algorithm can eliminate noise and errors in the data, generating an accurate waterway multi-source dynamic dataset. The fused data is stored in a database, such as a relational database (MySQL) or a non-relational database (MongoDB), and indexed by fields such as timestamp, vessel ID, and waterway location for easy querying and analysis. This implementation method, through multi-source data collection and fusion, can comprehensively obtain real-time dynamic information about the waterway. The combination of real-time vessel trajectory data with waterway facility status data and meteorological and hydrological data can more accurately reflect the operating environment of vessels in the waterway. This richness of data provides reliable data support for subsequent waterway scheduling optimization.

[0039] In one possible implementation, the multi-source dynamic dataset of the waterway undergoes spatiotemporal fusion processing to generate a waterway spatiotemporal state matrix. Step S100 further includes step S160, which timestamps the real-time ship trajectory data, the waterway facility status data, and the waterway meteorological and hydrological data according to the waterway scheduling cycle, obtaining multiple data time tags. Specifically, a precise timestamp is generated for each data record using a computer algorithm. For example, a timestamp generator synchronized with a high-precision clock (such as an atomic clock) or Network Time Protocol (NTP) is used to ensure the accuracy and consistency of the timestamps. Data processing software (such as Python's Pandas library) is used to timestamp the collected real-time ship trajectory data, waterway facility status data, and waterway meteorological and hydrological data, and the timestamps are appended to each data record.

[0040] Step S170: The real-time ship trajectory data, waterway facility status data, and waterway meteorological and hydrological data are time-aligned according to the multiple data time tags to generate a time alignment result. Specifically, an interpolation algorithm (such as linear interpolation) is used to time-align data with different timestamps. For example, ship data is collected every 5 minutes, and meteorological data is collected every 15 minutes. The meteorological data value at each 5-minute time point is calculated using an interpolation algorithm. The interpolated data is then synchronously processed using data processing software (such as MATLAB) to generate the time alignment result, ensuring consistency of all data in the time dimension.

[0041] Step S180: The target waterway is divided into grids to obtain multiple grid cells, and a grid coordinate system is constructed based on these grid cells. Specifically, Geographic Information System (GIS) technology is used to divide the target waterway into grids, and the boundaries and identifiers of each grid cell are determined. For example, the waterway is divided into multiple uniform grid cells, each with a unique identifier. Based on the grid division results, a grid coordinate system is constructed. For example, a two-dimensional Cartesian coordinate system is established with the starting point of the waterway as the origin, and each grid cell corresponds to a coordinate point. Examples of grid division and coordinate system construction are shown in Table 4.

[0042] Table 4: Examples of Mesh Generation and Coordinate System Construction

[0043]

[0044] Step S190 involves mapping the real-time vessel trajectory data, the waterway facility status data, and the waterway meteorological and hydrological data to the grid coordinate system for spatial alignment, generating a spatial alignment result. Specifically, GIS technology is used to map the real-time vessel trajectory data, waterway facility status data, and waterway meteorological and hydrological data to the grid coordinate system. For example, the position coordinates of vessel V1 are (121.51, 31.21), mapped to grid cell G1. Data processing software (such as ArcGIS) is then used to perform spatial alignment processing on the mapped data, generating a spatial alignment result to ensure the consistency of all data in spatial dimensions.

[0045] Step S1100: Based on the time alignment result and the spatial alignment result, the real-time ship trajectory data, the waterway facility status data, and the waterway meteorological and hydrological data are spatiotemporally fused to construct a three-dimensional tensor structure. Specifically, the time-aligned and spatially aligned data are constructed into a three-dimensional tensor structure. For example, the dimensions of the three-dimensional tensor are (time, space, data type), and each element contains ship trajectory, waterway facility status, and meteorological and hydrological data. A tensor decomposition algorithm (such as Tucker decomposition) is used to fuse the three-dimensional tensor to generate the fused data structure.

[0046] Step S1110: Perform fusion quality verification according to the described three-dimensional tensor structure. When the verification passes, generate the channel spatiotemporal state matrix. Specifically, statistical analysis methods (such as mean squared error, correlation coefficient, etc.) are used to verify the quality of the fused data. For example, the mean squared error between the fused data and the original data is calculated to ensure the accuracy of the fused data. When the verification passes, the channel spatiotemporal state matrix is ​​generated. This implementation method ensures the consistency of multi-source dynamic data of the channel in both time and space dimensions through timestamp identification, time alignment, spatial alignment, and spatiotemporal fusion processing. The target channel is divided into grids and a grid coordinate system is constructed, enabling refined management of the channel. The fusion quality verification algorithm is used to verify the quality of the fused data, ensuring the reliability and accuracy of the data fusion.

[0047] In one possible implementation, the real-time ship trajectory data, the waterway facility status data, and the waterway meteorological and hydrological data are spatiotemporally fused based on the time alignment result and the spatial alignment result to construct a three-dimensional tensor structure. Step S1100 further includes step S1101, which encodes the real-time ship trajectory data based on the time alignment result and the spatial alignment result to construct a ship position probability distribution layer. Specifically, the real-time ship trajectory data is preprocessed to extract time-series features. For example, ship speed, heading, and other data are used as feature inputs. Machine learning algorithms (such as Long Short-Term Memory networks LSTM) are used to encode the real-time ship trajectory data using time-series features to generate a time-series feature vector. The time-series feature vector is input into a Gaussian Mixture Model (GMM), and the GMM model is used to model the probability distribution of the ship's position, calculating the probability distribution of the ship's position in different grid cells. An example of the ship position probability distribution layer is shown in Table 5.

[0048] Table 5: Examples of Ship Position Probability Distribution Layers

[0049]

[0050] Step S1102: Based on the time alignment result and the spatial alignment result, spatial relationship modeling is performed on the waterway facility status data. The waterway saturation coefficient is calculated based on the modeling result to obtain the waterway congestion index layer. Specifically, the waterway facility status data is preprocessed to extract spatial features. For example, water level and flow velocity data are used as feature inputs. A graph neural network (GNN) is used to model the spatial relationships of the waterway facility status data. For example, waterway grid cells are used as nodes, and waterway facility status is used as node features to construct a waterway spatial relationship graph. By analyzing the waterway spatial relationship graph, the waterway saturation coefficient of each grid cell is calculated. For example, the saturation coefficient is calculated based on the number of ships and waterway capacity within the grid cell. Based on the waterway saturation coefficient, the waterway congestion index is calculated. For example, a linear regression model is used to map the saturation coefficient to the congestion index. An example of the waterway congestion index layer is shown in Table 6.

[0051] Table 6: Examples of Channel Congestion Index Layers

[0052]

[0053] Step S1103: Based on the time alignment result and the spatial alignment result, perform a waterway meteorological risk analysis on the waterway meteorological and hydrological data, and set a meteorological risk level layer. Specifically, preprocess the waterway meteorological and hydrological data to extract meteorological features. For example, use data such as wind speed, wind direction, temperature, and humidity as feature inputs. Use statistical analysis methods (such as Bayesian networks) to perform meteorological risk analysis on the waterway meteorological and hydrological data. For example, assess meteorological risk based on data such as wind speed, wind direction, temperature, and humidity. Set meteorological risk levels based on the meteorological risk analysis results. For example, divide the risk levels into low risk (1), medium risk (2), and high risk (3). Examples of meteorological risk level layers are shown in Table 7.

[0054] Table 7: Examples of Meteorological Risk Levels

[0055]

[0056] Step S1104 involves fusing the data from the ship position probability distribution layer, the waterway congestion index layer, and the meteorological risk level layer to construct the three-dimensional tensor structure. Specifically, the data from these layers are correlated. For example, the ship position probability distribution, waterway congestion index, and meteorological risk level of each grid cell at different time points are integrated. Tensor decomposition algorithms (such as Tucker decomposition) are used to construct a three-dimensional tensor structure from the correlated data. For instance, the three levels of data are constructed into a single three-dimensional tensor structure, where the dimensions include time, space, and data type. This implementation, through temporal feature encoding and probability distribution modeling, enables refined prediction of the ship position probability distribution at different time points. Spatial relationship modeling and waterway saturation coefficient calculation allow for real-time assessment of waterway congestion. Meteorological risk analysis and risk level setting enable dynamic assessment of waterway meteorological risk. This comprehensive spatiotemporal data fusion provides comprehensive data support for waterway scheduling optimization and improves the intelligence level of waterway management.

[0057] Step S200: Perform channel conflict detection on the channel spatiotemporal state matrix, determine channel traffic conflict parameters, perform bidirectional spatiotemporal planning on the target channel according to the channel traffic conflict parameters, and formulate an initial scheduling scheme.

[0058] Specifically, channel traffic conflict parameters refer to factors used to determine whether a traffic conflict exists between vessels. These parameters include vessel speed, course, position, channel width, and safe distance. Channel traffic conflict parameters are key constraints in channel conflict detection and two-way spatiotemporal planning. For example, a conflict will occur when two vessels are traveling at high speeds and on intersecting courses, and the distance between them is less than the safe distance. By properly setting and adjusting these parameters, collisions and other accidents between vessels can be effectively avoided, and channel scheduling schemes can be optimized. Two-way spatiotemporal planning refers to considering the passage arrangements of vessels in two-way channels during channel scheduling optimization, planning the passage paths of vessels in two-way channels to achieve orderly passage of vessels in the channel. Two-way spatiotemporal planning can rationally arrange meeting and passing maneuvers of vessels, avoid conflicts between vessels, and improve channel traffic efficiency.

[0059] A rule-based conflict detection algorithm is employed. The generated channel spatiotemporal state matrix is ​​input into the algorithm, which analyzes data such as ship positions and speeds based on preset rules and learned patterns to determine the existence of conflicts. For example, a rule is set that a conflict exists when two ships are less than a safe distance threshold (e.g., 100 meters) in the channel and their courses intersect. Simultaneously, machine learning algorithms (e.g., support vector machines) are used to learn from historical conflict data and identify potential conflict patterns. Upon detecting a conflict, a heuristic algorithm (e.g., genetic algorithm) is used for bidirectional spatiotemporal planning. This algorithm considers channel traffic conflict parameters (e.g., ship speed, channel width, safe distance), and through an iterative optimization process, adjusts the ship passage paths, times, and sequences to formulate an initial scheduling plan. For example, for two ships in conflict, the algorithm might adjust the speed of one ship to delay its entry into the conflict area, or change the course of another ship to bypass the conflict point, thus formulating an initial scheduling plan.

[0060] For example, the channel spatiotemporal state matrix shows that ship A and ship B are 80 meters apart on the channel, and their course intersects. According to the conflict detection rules, the distance between the two ships is less than the safe distance threshold (100 meters), indicating a conflict. Simultaneously, the machine learning algorithm, based on historical data, identifies this situation of intersecting course and close distance as a common conflict pattern, further confirming the existence of the conflict. After receiving the conflict information, the two-way spatiotemporal planning algorithm begins to optimize the scheduling scheme. The algorithm considers parameters such as a channel width of 100 meters, ship A's speed of 10 knots, and ship B's speed of 8 knots. After multiple iterations, the algorithm decides to have ship A slow down to 8 knots, delaying its entry into the conflict area, while simultaneously having ship B slightly change its course, shifting 10 degrees to the left, thus avoiding the conflict. The generated initial scheduling scheme includes the adjusted passage paths, times, and order of ships A and B.

[0061] like Figure 2 As shown, in one possible implementation, the channel spatiotemporal state matrix is ​​used for channel conflict detection to determine channel traffic conflict parameters. Step S200 further includes step S210, which involves retrieving the target vessel within the target channel for kinematic analysis and setting the vessel's dynamic safety zone radius. Specifically, the kinematic parameters of the target vessel, such as speed, acceleration, and heading, are extracted. Kinematic analysis of the target vessel is performed using kinematic equations. For example, by analyzing parameters such as speed, acceleration, and heading, the vessel's trajectory is predicted. Based on the vessel's kinematic characteristics, the vessel's dynamic safety zone radius is set. For example, the safety zone radius is set based on the vessel's minimum turning radius and braking distance. Examples of vessel dynamic safety zone radii are shown in Table 8.

[0062] Table 8: Examples of Dynamic Safety Domain Radius for Ships

[0063]

[0064] Step S220: Perform four-dimensional analysis based on the ship's dynamic safety domain radius, the ship's position probability distribution layer, the waterway congestion index layer, and the meteorological risk level layer to construct a four-dimensional ship trajectory waterway model. Specifically, integrate the data from the ship's dynamic safety domain radius, the ship's position probability distribution layer, the waterway congestion index layer, and the meteorological risk level layer. Use multi-dimensional data analysis methods to construct the four-dimensional ship trajectory waterway model. For example, construct the four-dimensional data into a four-dimensional tensor structure, where the dimensions include time, space, ship's dynamic safety domain radius, and data type.

[0065] Step S230: Based on the four-dimensional ship trajectory channel model, the minimum safe distance is calculated using Euclidean distance detection. Specifically, the safe distance between ships is calculated using the Euclidean distance formula. For example, the safe distance between ship V1 and ship V2 is calculated. The minimum value of the calculated safe distance is then detected. For example, the minimum safe distance between ship V1 and ship V2 is 120 meters.

[0066] Step S240 involves performing multi-ship spatiotemporal conflict detection on the target waterway based on the minimum safe distance value. When the trajectories of multiple ships fall below the minimum safe distance value, waterway traffic conflict parameters are generated. Specifically, data such as ship positions, safe zone radii, waterway congestion index, and meteorological risk levels are extracted from the four-dimensional ship trajectory waterway model. A conflict detection algorithm is used to detect the safe distance between ships. When the safe distance is less than the minimum safe distance value, waterway traffic conflict parameters are generated. This implementation calculates the minimum safe distance value through Euclidean distance detection, enabling real-time detection of conflicts between ships and allowing for proactive avoidance measures to ensure navigation safety.

[0067] In one possible implementation, bidirectional spatiotemporal planning of the target waterway is performed according to the waterway traffic conflict parameters to formulate an initial scheduling scheme. Step S200 further includes step S250, which identifies multiple conflict events based on the waterway traffic conflict parameters, triggers a conflict resolution unit, and performs three-dimensional resolution on the multiple conflict events based on the conflict resolution unit to generate multiple solution spaces. Specifically, conflict events are extracted from the waterway traffic conflict parameters. For example, ships V1 and V2 conflict at timestamp 2024-06-24T08:00:00Z. The conflict resolution unit is triggered to perform three-dimensional resolution on the conflict events, generating multiple solution spaces by adjusting parameters such as ship speed, heading, and time. For example, adjusting the speed of ship V1 from 10 knots to 8 knots, adjusting the course from 90° to 85°, and delaying the adjustment time by 5 minutes generates solution space 1; adjusting the speed of ship V2 from 12 knots to 10 knots, adjusting the course from 270° to 275°, and advancing the adjustment time by 5 minutes generates solution space 2.

[0068] Step S260: Based on the multiple solution spaces, construct bidirectional passage constraints, divide the target waterway into N logical resource blocks, and perform spatiotemporal conflict analysis on the N logical resource blocks according to the bidirectional passage constraints to construct a spatiotemporal resource conflict graph, where N is an integer greater than 1. Specifically, the target waterway is divided into N logical resource blocks. For example, the waterway is divided into 10 logical resource blocks, each corresponding to a grid cell. Graph theory algorithms (such as shortest path algorithms) are used to construct bidirectional passage constraints based on the logical resource block division. For example, passage rules for ships between logical resource blocks are set, such as the passage constraint for ship V1 from logical resource block 1 to logical resource block 2. Multidimensional data analysis methods are used to perform spatiotemporal conflict analysis on the logical resource blocks. For example, by analyzing the position and time of ships in different logical resource blocks, a spatiotemporal resource conflict graph is constructed.

[0069] Step S270: Based on the spatiotemporal resource conflict map, a two-way spatiotemporal resource allocation strategy is constructed to collaboratively plan the target waterway and formulate the initial scheduling scheme. Specifically, based on the spatiotemporal resource conflict map, an optimization algorithm (such as a genetic algorithm) is used to construct the two-way spatiotemporal resource allocation strategy. For example, the passage paths and times of ships in logical resource blocks are optimized to avoid conflicts. The two-way spatiotemporal resource allocation strategy is collaboratively planned to formulate the initial scheduling scheme. For example, the passage paths and times of ships V1 and V2 are adjusted to ensure that they do not conflict in logical resource block 1. This implementation method, by constructing two-way passage constraints and a spatiotemporal resource conflict map, can finely manage the spatiotemporal resources of the waterway, perform spatiotemporal conflict analysis on each resource block, and ensure that the passage arrangements of ships in different resource blocks are reasonable and avoid conflicts.

[0070] In one possible implementation, bidirectional traffic constraints are constructed based on the multiple solution spaces. The target waterway is divided into N logical resource blocks. Spatiotemporal conflict analysis is performed on the N logical resource blocks according to the bidirectional traffic constraints to construct a spatiotemporal resource conflict graph. Step S260 further includes step S261, parsing the multiple solution spaces to obtain time solutions, spatial solutions, and velocity solutions. Specifically, a data parsing algorithm is used to parse the multiple solution spaces, extracting time solutions, spatial solutions, and velocity solutions from the solution spaces. For example, from solution space 1, the speed of ship V1 is adjusted to 8 knots, the heading is adjusted to 85°, and the time delay is 5 minutes.

[0071] Step S262: Based on the time solution, spatial solution, and velocity solution, constraint analysis is performed to construct a three-dimensional constraint set. Bidirectional traffic matching is then performed on the target channel according to this three-dimensional constraint set, setting bidirectional traffic constraints. Specifically, mathematical modeling methods are used to perform constraint analysis on the time solution, spatial solution, and velocity solution to construct the three-dimensional constraint set. For example, by setting a speed limit, heading range, and time window, a three-dimensional constraint set is generated, such as setting the speed limit of ship V1 to 10 knots, the heading range to 80°-90°, and the time window to ±5 minutes. Graph theory algorithms (such as shortest path algorithms) are used to perform bidirectional traffic matching on the target channel, setting bidirectional traffic constraints. For example, based on the three-dimensional constraint set, traffic rules for ships between different logical resource blocks are set.

[0072] Step S263: Divide the target waterway into N logical resource blocks, traverse the N logical resource blocks, define them, and determine N four-dimensional tuples. Specifically, use a grid partitioning algorithm to divide the target waterway into N logical resource blocks. For example, divide the waterway into 10 logical resource blocks, each resource block corresponding to a grid cell. Define the four-dimensional tuple for each logical resource block, including time, space, ship ID, and resource block ID. For example, define the four-dimensional tuple for logical resource block 1 as (time, space, ship ID, resource block ID).

[0073] Step S264: Perform conflict detection on the N four-dimensional tuples according to the bidirectional passage constraints to obtain inter-block conflict relationships. Specifically, multi-dimensional data analysis methods are used to perform conflict detection on logical resource blocks. For example, by analyzing the position and time of ships in different logical resource blocks, inter-block conflict relationships are detected, recorded, and a conflict detection report is generated. Examples of inter-block conflict detection are shown in Table 9.

[0074] Table 9: Examples of Inter-block Conflict Detection

[0075]

[0076] Step S265: Define the conflict intensity based on the inter-block conflict relationship, generate a conflict weight coefficient, and spatiotemporally associate the N logical resource blocks according to the inter-block conflict relationship and the conflict weight coefficient to construct the spatiotemporal resource conflict graph. Specifically, the conflict intensity is defined based on the inter-block conflict relationship. For example, the conflict intensity is defined based on the conflict duration and conflict distance. A conflict weight coefficient is generated based on the conflict intensity. For example, the greater the conflict intensity, the higher the conflict weight coefficient. Graph theory algorithms are used to spatiotemporally associate the logical resource blocks to construct the spatiotemporal resource conflict graph. For example, a weighted graph is constructed to represent the conflict relationship between logical resource blocks. This implementation method obtains temporal, spatial, and velocity solutions by analyzing multiple solution spaces, and constructs a three-dimensional solution constraint set based on these solutions, enabling fine-grained setting of bidirectional passage constraints. By dividing the target channel into multiple logical resource blocks and performing conflict detection on each resource block, the spatiotemporal resources of the channel can be managed efficiently. By defining conflict intensity and generating conflict weight coefficients, logical resource blocks are spatiotemporally correlated to construct a spatiotemporal resource conflict map, which can intuitively represent the spatiotemporal resource conflict situation of the waterway and provide a basis for optimizing scheduling.

[0077] Step S300: According to the initial scheduling scheme, combined with real-time ship dynamic position data, the data is sent to the ship's intelligent terminal for rolling optimization, and a waterway dynamic optimization instruction is generated.

[0078] Specifically, rolling optimization refers to an optimization method that continuously adjusts the scheduling plan based on real-time dynamic data of ships during the channel scheduling optimization process. It is a dynamic optimization process that can respond in real time to changes in ship operations. Rolling optimization can improve the adaptability and accuracy of channel scheduling plans. For example, when ships deviate from their predetermined paths due to factors such as wind and waves, the rolling optimization algorithm can adjust the ship's passage route and time arrangement in a timely manner based on new position data, ensuring that ships can navigate safely and efficiently.

[0079] The initial scheduling plan and real-time vessel dynamic position data are transmitted to the vessel's intelligent terminal via a communication network. The intelligent terminal can be an onboard Electronic Chart Display and Information System (ECDIS) or a dedicated intelligent shipping device. Upon receiving the data, the intelligent terminal employs a rolling optimization algorithm based on Model Predictive Control (MPC). This algorithm predicts the vessel's operational status over a future period based on real-time dynamic data such as the vessel's position and speed, combined with the spatiotemporal state matrix of the waterway, and adjusts the scheduling plan in real time. For example, if vessel A encounters sudden storms and deviates from its planned path, the rolling optimization algorithm recalculates the optimal path for vessel A based on the new position data, adjusts its speed and course, and generates dynamic waterway optimization instructions.

[0080] For example, the initial scheduling plan and the real-time position data of vessel A are transmitted to vessel A's ECDIS equipment via the BeiDou satellite navigation system communication link. The data transmission rate is 100kbps, enabling complete data transmission to the vessel's intelligent terminal within one second. During navigation, vessel A deviates 100 meters from its predetermined path due to sudden wind and waves. Upon receiving the new position data, the rolling optimization algorithm on the vessel's intelligent terminal recalculates the optimal path for vessel A. Based on hydrological and meteorological data (e.g., wind and wave intensity of level 4) in the channel spatiotemporal state matrix and vessel A's performance parameters (e.g., maximum speed of 12 knots), the algorithm adjusts vessel A's speed to 10 knots and its course to shift 5 degrees to the right to avoid the windy and wave-prone area. The generated dynamic channel optimization command, including the new speed and course information, is sent to vessel A's autopilot system for execution.

[0081] In one possible implementation, according to the initial scheduling scheme, real-time vessel dynamic position data is sent to the vessel's intelligent terminal for rolling optimization, generating a channel dynamic optimization command. Step S300 further includes step S310, which maps the real-time vessel dynamic position to the initial scheduling scheme based on the vessel's intelligent terminal to perform channel scheduling analysis and obtain the spatiotemporal deviation. Specifically, GIS technology is used to map the real-time vessel dynamic position data to the initial scheduling scheme. For example, based on the vessel's real-time position coordinates, it is mapped to the corresponding logical resource block. The spatiotemporal deviation is calculated using mathematical modeling methods. For example, by comparing the vessel's actual position and time with the predetermined position and time in the initial scheduling scheme, the deviation is calculated. For instance, if the actual arrival time of vessel V1 is 3 minutes later than the predetermined time, the position deviation is 50 meters.

[0082] Step S320: Construct a rolling time-domain control framework. Based on this framework and the spatiotemporal deviation, optimize and label the initial scheduling scheme, determining local replanning labels. Specifically, the rolling time-domain control framework is constructed using the rolling time-domain control method in control theory. For example, a rolling window is set, with a window size of 10 minutes, scrolling every 5 minutes. Based on the spatiotemporal deviation, an optimization algorithm (such as linear programming) is used to perform local replanning on the initial scheduling scheme. For example, based on the spatiotemporal deviation, the ship's travel path and time are adjusted, generating local replanning labels. For instance, the travel path of ship V1 is adjusted so that it travels directly from logical resource block 1 to logical resource block 3, skipping logical resource block 2.

[0083] Step S330: The local replanning tag is located to the target channel, and the channel dynamic optimization command is generated. Specifically, a data processing algorithm is used to locate the local replanning tag to the target channel and generate the channel dynamic optimization command. For example, the command "V1:G1→G3, speed adjusted to 9 knots, heading adjusted to 88°" is generated. The channel dynamic optimization command is sent to the ship's smart terminal using wireless communication technology (such as 4G / 5G). For example, the optimization command is sent to the smart terminal of ship V1 via a 4G network. This implementation method maps real-time ship dynamic position data to the initial scheduling scheme, enabling real-time analysis of the channel scheduling situation and obtaining the spatiotemporal deviation. By constructing a rolling time-domain control framework, the initial scheduling scheme can be dynamically replanned locally based on the spatiotemporal deviation. By generating the channel dynamic optimization command and sending it to the ship's smart terminal, the ship's navigation can be efficiently guided, ensuring navigation safety and efficiency.

[0084] In one possible implementation, the local replanning label is located to the target channel, and the channel dynamic optimization instruction is generated. Step S330 further includes step S331, which identifies the impact of the local replanning label to generate a set of affected ships. Specifically, graph theory algorithms (such as shortest path algorithms) are used to identify the impact of the local replanning label on other ships. For example, the affected ships are identified by analyzing the position and time of the ships in logical resource blocks. For example, if ship V1 travels directly from logical resource block 1 to logical resource block 3, it may affect ships V2 and V3 in the same area, then the set of affected ships is {V2, V3}.

[0085] Step S332: Retrieve the set of unaffected vessels to freeze their schedules and construct a local optimization subspace. Specifically, data processing algorithms are used to freeze the schedules of unaffected vessels. For example, the travel paths and times of unaffected vessels are fixed to construct a local optimization subspace. For instance, vessels V4 and V5 are not in the set of affected vessels, so their schedules remain unchanged. The local optimization subspace includes logical resource blocks 1, 2, and 3, as well as the travel paths and times of vessels V2 and V3.

[0086] Step S333: Perform steering calculations on the affected ship set based on the local optimization subspace to generate a heading and turning point sequence. Specifically, perform steering calculations on the affected ships using kinematic equations. For example, generate a heading and turning point sequence by calculating the optimal turning path for each ship. For instance, if ship V2 needs to turn from logical resource block 1 to logical resource block 4, calculate its optimal turning path.

[0087] Step S334: Map the heading and turning point sequence to the target channel to generate the channel dynamic optimization command. Specifically, a data processing algorithm is used to map the heading and turning point sequence to the target channel to generate the channel dynamic optimization command. For example, the command "V2: G1→G2→G4, speed adjusted to 10 knots, heading adjusted to 270°" is generated. The generated channel dynamic optimization command is sent to the ship's smart terminal via a wireless communication network. For example, the command is sent to the smart terminal of ship V2 via a 4G network. This implementation method, by identifying the impact of local replanning tags on other ships, ensures the relevance of the optimization command and avoids unnecessary adjustments.

[0088] Step S400: Execute the channel dynamic optimization command to provide scheduling feedback, generate scheduling results, update the initial scheduling scheme based on the scheduling results, and determine the channel dynamic scheduling scheme.

[0089] Specifically, feedback monitoring equipment, such as radar and video surveillance equipment, is installed along the waterway to monitor vessel traffic in real time. This equipment acquires data on vessels' actual travel paths, times, and sequences, and transmits the data to a data processing center. Upon receiving the feedback data, the data processing center employs an adaptive update algorithm based on this data. This algorithm compares and analyzes the scheduling results obtained from the feedback monitoring system with the initial scheduling plan. For example, it calculates the deviation between the actual vessel travel time and the predetermined time in the scheduling plan, and adjusts the initial scheduling plan accordingly. If the deviation is significant, the algorithm re-optimizes the scheduling plan to determine a dynamic scheduling scheme for the waterway.

[0090] For example, a radar device is installed at the junction of waterways to monitor vessel traffic. The radar device can obtain the actual passage path and time of vessel A in real time. For instance, vessel A's actual passage time at the junction is 14:30, 10 minutes later than the scheduled time of 14:20 in the scheduling plan. The radar device transmits this data to the data processing center via a fiber optic communication link. After receiving the feedback data, the data processing center initiates an adaptive update algorithm. The algorithm calculates the passage time deviation of vessel A, considers the passage of other vessels in the waterway (such as vessel B waiting to pass at the junction) and current hydrological and meteorological conditions (such as wind speed of 2 m / s and water level of 3 meters), and readjusts vessel A's passage time. For example, the passage time of vessel A is adjusted to 14:35, and the initial scheduling plan is updated to determine a dynamic scheduling plan for the waterway, enabling vessels A and B to pass in an orderly manner.

[0091] In one possible implementation, the dynamic channel optimization command is executed to provide scheduling feedback, generating a scheduling effect. The initial scheduling plan is then updated based on the scheduling effect to determine the dynamic channel scheduling plan. Step S400 further includes step S410, executing the dynamic channel optimization command to synchronously collect data on the target channel and obtain a channel operation log. Specifically, the dynamic channel optimization command is executed, and a data acquisition system is used to synchronously collect data on the target channel. For example, real-time position data of vessels V2 and V3 are collected using an AIS system. The collected data is recorded using data processing software (such as Python) to generate a channel operation log.

[0092] Step S420: Perform a multi-dimensional scheduling evaluation based on the waterway operation log to generate a scheduling effect. Specifically, use data analysis methods to perform a multi-dimensional evaluation of the waterway operation log. For example, evaluate dimensions such as vessel transit time, transit route, and hydro-meteorological conditions to generate a scheduling effect. For instance, evaluate whether the transit time of vessels V2 and V3 meets expectations, whether the transit route is smooth, and whether hydro-meteorological conditions affect navigation. An example of the scheduling effect is shown in Table 10.

[0093] Table 10: Examples of Scheduling Results

[0094]

[0095] Step S430: Using reinforcement learning, the initial scheduling scheme is incrementally updated according to the scheduling effect to generate a dynamic scheduling scheme for the waterway. Specifically, based on the scheduling effect, a reinforcement learning algorithm (such as Q-learning) is used to incrementally update the initial scheduling scheme to generate a dynamic scheduling scheme for the waterway. For example, based on the scheduling effect, the transit routes and times of ships are adjusted to optimize the scheduling scheme.

[0096] Step S440: Digital twin verification is performed based on the dynamic scheduling scheme for the waterway. When the digital twin verification confirms the scheme's feasibility, the dynamic scheduling scheme is deployed for execution. Specifically, digital twin technology is used to verify the dynamic scheduling scheme. For example, by constructing a digital twin model of the waterway, including its topology, hydrological and meteorological conditions, the passage of ships is simulated to verify the feasibility of the dynamic scheduling scheme and generate a verification report. When the digital twin verification confirms the scheme's feasibility, the dynamic scheduling scheme is deployed for execution. This implementation utilizes reinforcement learning algorithms to incrementally update the initial scheduling scheme, enabling intelligent optimization and improving its adaptability and efficiency.

[0097] This application's embodiments employ techniques such as dynamically collecting and fusing target channel data according to a scheduling cycle to generate a channel spatiotemporal state matrix, performing channel conflict detection, formulating an initial scheduling scheme based on bidirectional spatiotemporal planning using detection parameters, continuously optimizing the initial scheme by combining real-time vessel dynamic position data, generating dynamic optimization instructions, executing the instructions and providing feedback on the scheduling effect, updating the initial scheme accordingly, and determining the dynamic scheduling scheme. These techniques solve the technical problem of existing channel scheduling optimization methods being unable to adapt to dynamic changes, leading to unreasonable scheduling. This achieves the technical effect of dynamically adjusting according to real-time conditions, effectively reducing channel traffic conflicts, improving vessel traffic efficiency, and optimizing the overall transportation efficiency of the channel.

[0098] The specific embodiments described above do not constitute a limitation on the scope of protection of this application. Those skilled in the art should understand that various modifications, combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application should be included within the scope of protection of this application. In some cases, the actions or steps described in this application can be performed in a different order than that shown in the embodiments and still achieve the desired results. Furthermore, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

Claims

1. A waterway scheduling optimization method based on artificial intelligence, characterized in that, The method includes: Dynamic data is collected on the target waterway according to the waterway scheduling cycle. The multi-source dynamic dataset of the waterway is then subjected to spatiotemporal fusion processing to generate a spatiotemporal state matrix of the waterway. The channel spatiotemporal state matrix is ​​used for channel conflict detection to determine channel passage conflict parameters. Based on the channel passage conflict parameters, bidirectional spatiotemporal planning is performed on the target channel to formulate an initial scheduling scheme. According to the initial scheduling plan, real-time vessel dynamic position data is sent to the vessel's intelligent terminal for rolling optimization, generating a waterway dynamic optimization command. The waterway dynamic optimization command is executed to provide scheduling feedback, generating scheduling results. The initial scheduling scheme is updated based on the scheduling results to determine the waterway dynamic scheduling scheme. Based on the aforementioned waterway traffic conflict parameters, a two-way spatiotemporal planning process is performed on the target waterway to formulate an initial scheduling scheme. The methods include: Based on the waterway traffic conflict parameters, multiple conflict events are determined, triggering a conflict resolution unit. The conflict resolution unit then performs three-dimensional resolution on the multiple conflict events, generating multiple solution spaces. Based on the multiple solution spaces, a two-way passage constraint is constructed, and the target channel is divided into N logical resource blocks. Spatiotemporal conflict analysis is performed on the N logical resource blocks according to the two-way passage constraint to construct a spatiotemporal resource conflict graph, where N is an integer greater than 1. Based on the spatiotemporal resource conflict map, a two-way spatiotemporal resource allocation strategy is constructed to collaboratively plan the target waterway and formulate the initial scheduling scheme. Based on the multiple solution spaces, bidirectional passage constraints are constructed, the target channel is divided into N logical resource blocks, and spatiotemporal conflict analysis is performed on the N logical resource blocks according to the bidirectional passage constraints to construct a spatiotemporal resource conflict graph. The method includes: The multiple solution spaces are analyzed to obtain time solutions, space solutions, and velocity solutions; Constraint analysis is performed based on the time solution, the spatial solution, and the velocity solution to construct a three-dimensional solution constraint set. Bidirectional traffic matching of the target channel is performed based on the three-dimensional solution constraint set, and bidirectional traffic constraints are set. The target channel is divided into N logical resource blocks. The N logical resource blocks are traversed and defined to determine N four-dimensional tuples. Conflict detection is performed on the N four-dimensional tuples according to the bidirectional passage constraints to obtain the inter-block conflict relationship; The conflict intensity is defined based on the inter-block conflict relationship, a conflict weight coefficient is generated, and the N logical resource blocks are spatiotemporally associated according to the inter-block conflict relationship and the conflict weight coefficient to construct the spatiotemporal resource conflict graph.

2. The waterway scheduling optimization method based on artificial intelligence as described in claim 1, characterized in that, Dynamic data collection is performed on the target waterway according to the waterway scheduling cycle to obtain a multi-source dynamic dataset of the waterway. The methods include: The tidal cycle and waterway operation cycle of the target waterway are retrieved, and the tidal cycle and waterway operation cycle are used as dual driving factors to construct the waterway scheduling cycle; Based on the waterway scheduling cycle, a dynamic data acquisition command is issued, which activates the ship data acquisition unit to acquire data from the target waterway and determine the ship's real-time trajectory data. The dynamic acquisition command activates the waterway data acquisition unit to acquire data from the target waterway and determine the status data of the waterway facilities. The dynamic acquisition command activates the meteorological data acquisition unit to acquire data from the target waterway, thereby determining the waterway's meteorological and hydrological data. The real-time trajectory data of the vessels, the status data of the waterway facilities, and the meteorological and hydrological data of the waterway are integrated from multiple sources to obtain the multi-source dynamic dataset of the waterway.

3. The waterway scheduling optimization method based on artificial intelligence as described in claim 2, characterized in that, Spatiotemporal fusion processing is performed on multi-source dynamic datasets of waterways to generate a spatiotemporal state matrix of the waterway. Methods include: According to the waterway scheduling cycle, the real-time trajectory data of the ships, the status data of the waterway facilities, and the meteorological and hydrological data of the waterway are timestamped to obtain multiple data time tags; The real-time trajectory data of the ship, the status data of the waterway facilities, and the meteorological and hydrological data of the waterway are time-aligned according to the multiple data time tags to generate a time alignment result; The target waterway is divided into grids to obtain multiple grid cells, and a grid coordinate system is constructed based on the multiple grid cells. The real-time ship trajectory data, the waterway facility status data, and the waterway meteorological and hydrological data are mapped to the grid coordinate system for spatial alignment, generating a spatial alignment result. Based on the time alignment result and the spatial alignment result, the real-time trajectory data of the ship, the status data of the waterway facilities, and the meteorological and hydrological data of the waterway are spatiotemporally fused to construct a three-dimensional tensor structure; The fusion quality is verified according to the three-dimensional tensor structure. When the verification is successful, the spatiotemporal state matrix of the waterway is generated.

4. The waterway scheduling optimization method based on artificial intelligence as described in claim 3, characterized in that, Based on the time alignment result and the spatial alignment result, the real-time trajectory data of the ship, the status data of the waterway facilities, and the meteorological and hydrological data of the waterway are spatiotemporally fused to construct a three-dimensional tensor structure. The method includes: Based on the time alignment result and the spatial alignment result, the real-time trajectory data of the ship is encoded with temporal features to construct a ship position probability distribution layer. Based on the time alignment result and the spatial alignment result, a spatial relationship model is performed on the waterway facility status data. The waterway saturation coefficient is calculated based on the modeling result to obtain the waterway congestion index layer. Based on the time alignment results and the spatial alignment results, a waterway meteorological risk analysis is performed on the waterway meteorological and hydrological data, and a meteorological risk level layer is set. The three-dimensional tensor structure is constructed by associating and fusing the data of the ship position probability distribution layer, the waterway congestion index layer, and the meteorological risk level layer.

5. The waterway scheduling optimization method based on artificial intelligence as described in claim 4, characterized in that, The method involves performing channel conflict detection on the spatiotemporal state matrix of the waterway to determine channel passage conflict parameters, including: Retrieve target vessels within the target channel for kinematic analysis and set the dynamic safety zone radius of the vessels; A four-dimensional ship trajectory channel model is constructed by performing a four-dimensional analysis based on the ship dynamic safety domain radius, the ship position probability distribution layer, the channel congestion index layer, and the meteorological risk level layer. Based on the aforementioned four-dimensional ship trajectory channel model, the minimum safe distance is calculated using Euclidean distance detection. Based on the minimum safe distance, multi-ship spatiotemporal conflict detection is performed on the target waterway. When the trajectories of multiple ships are less than the minimum safe distance, the waterway passage conflict parameters are generated.

6. The waterway scheduling optimization method based on artificial intelligence as described in claim 1, characterized in that, According to the initial scheduling plan, real-time vessel dynamic position data is sent to the vessel's intelligent terminal for rolling optimization, generating a waterway dynamic optimization command. The method includes: Based on the ship's intelligent terminal, the real-time dynamic position of the ship is mapped to the initial scheduling scheme to perform waterway scheduling analysis and obtain the spatiotemporal deviation. Construct a rolling time-domain control framework, and optimize and identify the initial scheduling scheme according to the rolling time-domain control framework and the spatiotemporal deviation, and determine the local replanning label; The local replanning label is located to the target channel, and the channel dynamic optimization instruction is generated.

7. The waterway scheduling optimization method based on artificial intelligence as described in claim 6, characterized in that, The method involves locating the local replanning label to the target waterway and generating the waterway dynamic optimization instruction, including: Impact identification is performed based on the local replanning labels to generate a set of affected vessels; The unaffected vessel set is retrieved and the vessel scheduling is frozen to construct a local optimization subspace; Based on the local optimization subspace, the affected set of ships is subjected to turning calculations to generate a sequence of heading turning points; The heading and turning point sequence is mapped to the target channel to generate the channel dynamic optimization command.

8. The waterway scheduling optimization method based on artificial intelligence as described in claim 1, characterized in that, The method includes executing the dynamic channel optimization command to generate scheduling feedback, generating scheduling results, updating the initial scheduling plan based on the scheduling results, and determining the dynamic channel scheduling plan. The waterway dynamic optimization command is executed to synchronously collect data on the target waterway and obtain the waterway operation log. A multi-dimensional scheduling evaluation is performed based on the aforementioned waterway operation log to generate scheduling results; Reinforcement learning is used to incrementally update the initial scheduling scheme according to the scheduling effect, thereby generating a dynamic scheduling scheme for the waterway. Digital twin verification is performed based on the aforementioned dynamic scheduling scheme for waterways. When the digital twin verification confirms the scheme's feasibility, the dynamic scheduling scheme for waterways is then issued and executed.

Images