Ship dynamic speed limit and ship lock cooperative scheduling method and system based on environmental feature fusion

Abstract

The present application relates to a kind of environmental feature fusion ship dynamic speed limit and ship lock coordinated scheduling method and system, method includes the following steps: S1, the static intrinsic characteristics of target ship are obtained, and dynamic trajectory data containing real-time latitude and longitude and time stamp are obtained;S2, space-time state alignment is carried out using space-time interpolation algorithm;S3, the hydrological data after alignment, weather data and the static intrinsic characteristics of target ship and dynamic trajectory data are carried out feature engineering fusion, and derivative coupling feature is calculated and generated, and multi-dimensional sailing situation feature vector is constructed;S4, the adaptive safety speed limit interval for target ship in current voyage is output;S5, generate exclusive dynamic speed limit instruction and issue to ship terminal for real-time control;The updated estimated time of arrival is sent to ship lock scheduling system, to trigger the coordinated automatic optimization of scheduling scheme.The present application realizes the fine control of one ship one strategy and the great release of the overall navigation efficiency of regional waterway large traffic.

Description

Technical Field

[0001] This invention relates to intelligent water transportation, and more specifically, to a method and system for dynamic speed limiting of ships and coordinated scheduling of locks based on the integration of environmental characteristics. Background Technology

[0002] Currently, vessel speed limits on inland waterways primarily rely on static regulations established by waterway management departments, typically employing a segmented, "one-size-fits-all" fixed speed limit approach. This existing technical solution suffers from the following significant drawbacks and bottlenecks: it cannot adapt to environmental changes, as hydrological conditions (such as surges in flow velocity during flood season and drops in water levels during dry season) and meteorological conditions (such as sudden fog, strong crosswinds, and severe convective weather) are dynamic and subject to dramatic changes. Fixed speed limits cannot provide sufficient safety margins during severe weather and periods of hydrological anomalies, nor can they maximize waterway throughput when weather is favorable and visibility is good, leading to a significant waste of waterway transportation resources. Furthermore, the lack of a tailored management mechanism means that different vessels, due to differences in construction standards, current draft, tonnage distribution, and maneuverability, have drastically different safety perceptions of the same speed. A uniform speed limit may be too fast for fully loaded large vessels (posing risks of bank effects, increased seabed negative pressure, and grounding due to insufficient water depth), while it may be too slow for empty or lightly loaded small vessels, failing to leverage their maneuverability advantages.

[0003] The dynamic speed limits for waterways are severely disconnected from the scheduling of locks at key transportation hubs. While the construction of smart waterways has led to the deployment of numerous sensing devices along the waterways and the introduction of intelligent scheduling algorithms at the locks, "refined speed control of waterways" and "upper-level hub scheduling" are currently in a state of micro-isolation. On the one hand, speed limit instructions issued by waterway management departments often fail to penetrate the lock scheduling system in real time and quantitatively, resulting in the scheduling end's inability to accurately predict the drift in the estimated arrival time of individual vessels due to changes in speed limits. This leads to misallocation of lock chamber resources, empty lock operation, or severe periodic congestion in the pilot channel. On the other hand, existing lock scheduling relies heavily on static and dynamic data from vessel arrival for one-time scheduling, lacking a closed-loop traffic linkage mechanism that can directly map the "calculus-like change in the estimated arrival time of individual vessels caused by physical environmental constraints" to "automatic shuffling and multi-objective reorganization optimization of the hub lock queue." This technological disconnect severely restricts the overall navigation efficiency of inland waterways and navigation hubs. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a method and system for dynamic speed limiting of ships and coordinated scheduling of locks that integrates environmental characteristics, which realizes refined management and control of each ship and greatly releases the overall navigation efficiency of regional waterway transportation.

[0005] The technical solution adopted by this invention to solve its technical problem is: to construct a method for coordinated scheduling of ship dynamic speed limits and locks by integrating environmental characteristics, including the following steps: S1. Obtain the static intrinsic characteristics of the target vessel and acquire dynamic trajectory data containing real-time latitude, longitude and timestamp through the Automatic Identification System (AIS). S2. Based on latitude, longitude and timestamp, real-time hydrological and meteorological data are extracted through the waterway perception system, and spatiotemporal interpolation algorithm is used to align spatiotemporal states. S3. The aligned hydrological data, meteorological data and the static intrinsic features and dynamic trajectory data of the target ship are fused by feature engineering to calculate and generate derived coupled features, and construct a multi-dimensional navigation situation feature vector. S4. Input the multi-dimensional navigation situation feature vector into the pre-trained dynamic speed limit decision network model, calculate and output the adaptive safe speed limit range for the target ship in the current segment. S5. Generate a dedicated dynamic speed limit command based on the adaptive safe speed limit range and send it to the ship terminal for real-time control; at the same time, recalculate the estimated arrival time of the target ship based on the upper limit of the dynamic speed limit and send the updated estimated arrival time to the lock scheduling system to trigger the collaborative automatic optimization of the scheduling scheme.

[0006] According to the above scheme, in step S1, the static intrinsic characteristics of the target ship are collected by the Internet of Things sensing devices such as meteorological stations, flow velocity and direction instruments, and high-precision water level gauges deployed along the waterway, on cross-river bridges and navigation marks, at different high and low frequency cycles. The static intrinsic characteristics of the target ship are aggregated, converted, and cleaned of outliers by a standardized distributed API gateway cluster, and then uniformly packaged and pushed to the shipping big data base.

[0007] According to the above scheme, the specific method for performing spatiotemporal state alignment in step S2 is as follows: S201. Extract the current precise latitude and longitude of the target vessel, select multiple hydrological and meteorological sensing nodes that are closest to its upstream and downstream, and use the Kriging statistical spatial interpolation algorithm to evaluate and calculate the estimated values ​​of the flow field environment and meteorological parameters under the current absolute position coordinates of the vessel. S202. Using the timestamp of the dynamic trajectory data as the reference axis, establish a sliding window-type virtual timestamp window, map the interpolated environmental and meteorological parameter estimates into the virtual timestamp window, and generate a unified state composite vector containing spatiotemporal characteristic attributes.

[0008] According to the above scheme, in step S3, the derived coupling features include dynamic excess water depth ratio and crosswind drift risk vector; The dynamic margin of water depth is calculated as follows: extract the theoretical reference water depth of the water section, add the real-time water level compensation, subtract the real-time draft of the target vessel, and then divide by the real-time draft to obtain the result, which is used to quantify the bottom distance safety redundancy. The crosswind drift risk vector is calculated by performing sine decomposition on the real-time wind speed vector and the target ship's heading vector to the ground, and extracting the crosswind component perpendicular to the ship's pitch direction, which is used to quantify the yaw drift risk under complex weather conditions.

[0009] According to the above scheme, the dynamic surplus water depth ratio, which integrates real-time water level elevation, basic geographic information of the navigation section, and ship draft, quantifies the risk of bottoming out and towing resistance. The formula is as follows:

[0010] in, The theoretical reference water depth for the navigation segment, This refers to the real-time water level compensation amount transmitted back and interpolated by the waterway sensing system. Real-time draft of the ship; The relative flow velocity and crosswind drift risk vector are used to extract the crosswind component that has the greatest impact on the longitudinal maneuvering of the ship through vector projection. The formula is as follows:

[0011] in, The wind speed after spatiotemporal alignment. and These are the wind direction angle and the ship's heading angle relative to the ground, respectively.

[0012] According to the above scheme, in step S4, the network structure and reasoning process of the dynamic speed limit decision network model are as follows: using gated recurrent units or temporal convolutional networks to extract time series features of the weather and water flow evolution trends within a continuous time step, capturing the dynamic change characteristics of navigation resistance, and outputting a safe speed limit range containing the lower and upper speed limits through the quantile regression algorithm of the model output layer; when the crosswind drift risk vector exceeds the set critical threshold and the dynamic margin water depth ratio is lower than the preset safety threshold, the network model adaptively activates the penalty weight mapping and actively lowers the upper threshold of the safe speed limit range.

[0013] According to the above scheme, the network structure and reasoning process of the dynamic rate limiting decision network model include the following steps: S401: Receive a multidimensional tensor matrix, and use a gated recurrent unit or a temporal convolutional network to backtrack and extract the dynamic evolution trend of meteorological and water flow fields within a continuous time step. S402. Introduce a quantile regression algorithm on the output side to output an adaptive safe speed limit range with confidence intervals. When the model detects that the crosswind drift risk vector exceeds the critical threshold, and the dynamic surplus water depth ratio is extremely low due to the dry season, the neural network adaptively activates the penalty weights and nonlinearly adjusts the upper limit of velocity. And fine-tune the lower limit. ; S403. Extract historical safe navigation records of a large number of ships and use their actual speeds on specific sections as a reference for safe speed limits. During the training phase, use the quantile loss function for backpropagation updates of network weights, enabling the model to fit the mean of historical safe speeds and output a speed limit upper limit that meets the preset safety confidence level. Speed ​​limit lower limit This completes the offline pre-training of the network. According to the above scheme, the method for triggering the collaborative automatic optimization of the scheduling scheme in step S5 specifically includes: after the lock scheduling system receives the updated estimated arrival time, it matches and recalculates it with the current lock master data queuing sequence and lock chamber resource occupancy status; if it is determined that the target vessel cannot safely arrive at the designated lock according to the original scheduling plan, it is automatically moved to the subsequent feasible lock passage batch using a multi-objective optimization algorithm, and the current lock chamber space capacity is released in advance and redistributed to other queuing vessels that meet the lock passage conditions.

[0014] According to the above scheme, the method for triggering collaborative automatic optimization of the scheduling scheme includes the following steps: S501. The calculated safe speed limit range is digitally encrypted and encapsulated into a standard control command for issuance. If the actual speed continues to exceed the limit within the set penalty time window, the system will trigger an audible and visual alarm and record the violation. S502. Divide the remaining channel from the target vessel to the downstream lock into $n$ discrete mileage segments. Combined with the dynamic speed limit upper limit for each mileage segment output by the model. Based on the ship's current speed, the estimated arrival time is recalculated using the discrete integral formula:

[0015] in, This is the current timestamp. To account for the speed reduction factor due to water flow resistance and navigation density; S503, the updated high-precision estimated arrival time is pushed through to the lock scheduling system; after receiving the estimated arrival time drift signal, the lock scheduling algorithm kernel performs multi-target state matching and recalculation with the current lock chamber physical scale matrix and pre-queue sequence; if the evaluation result shows that the target vessel cannot arrive within the original strict time window, the algorithm will automatically start reorganization optimization, smoothly move it to the subsequent feasible lock passage batch, and release the theoretical space capacity of the current lock chamber originally occupied in advance, and redistribute it to other vessels in the queue.

[0016] This invention also provides a ship dynamic speed limiting and lock coordinated scheduling system based on environmental feature fusion, comprising: The multimodal sensing module is used to acquire static and dynamic data of the target vessel and obtain high-frequency real-time hydrological and meteorological data through the standard API interface of the distributed waterway sensing system. The spatiotemporal alignment and feature fusion module is used to perform spatial interpolation and temporal alignment on massive heterogeneous data, and to calculate derived coupled features to construct multidimensional navigation situation feature vectors; The decision-making module is internally deployed with a deep neural network containing a spatiotemporal attention mechanism, which is used to map multidimensional navigation situation feature vectors into safe dynamic speed limit zones. The control and coordination module is used to convert the calculated speed limit range into underlying control commands and send them to the ship traffic service system or shipboard terminal. At the same time, it coordinates with the back-end lock scheduling system to recalculate the estimated arrival time and rearrange the lock sequence.

[0017] The method and system for dynamic speed limiting of ships and coordinated scheduling of locks based on environmental feature fusion of the present invention have the following beneficial effects: 1. This invention breaks through human experience and improves navigation safety under extreme conditions. It uses AI algorithms to keenly capture the impact of subtle changes in hydrology and meteorology on the handling characteristics of specific ships. When encountering severe conditions such as strong winds, high water levels and current speeds, the model adaptively tightens safety redundancy, effectively curbing the occurrence of yaw, insufficient bottom distance and grounding accidents. 2. This invention completely breaks the constraints of static, one-size-fits-all speed limits. Under favorable environmental conditions and with suitable ship handling performance, the system can adaptively relax the speed limit range for specific waterways, significantly improving the overall traffic throughput and operational economic benefits of the waterway. It achieves flexible capacity expansion, maximizing the release of waterway traffic efficiency. 3. The speed limit results output by this invention are deeply interwoven with the real-time draft, beam and full load parameters of each ship, constructing a digital twin management system with one policy for each ship, and truly realizing the refined progress from "managing waterways" to "managing individual ships". 4. This invention is the first to directly map and calculate changes in navigation speed limits triggered by the environment into an early warning of the estimated arrival time drift of the navigation hub, and guides the efficient reorganization and allocation of the physical resource pool of the lock, thus opening up the "last mile" of intelligent joint control of regional transportation. Attached Figure Description

[0018] The present invention will be further described below with reference to the accompanying drawings and embodiments. In the accompanying drawings: Figure 1 This is a flowchart of the method and system for ship dynamic speed limiting and lock coordinated scheduling based on the fusion of environmental features, as described in this invention. Detailed Implementation

[0019] To provide a clearer understanding of the technical features, objectives, and effects of the present invention, specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0020] like Figure 1 As shown, the environmental feature fusion-based ship dynamic speed limiting and lock coordinated scheduling method of the present invention includes the following steps: S1. Obtain the static intrinsic characteristics of the target vessel and acquire dynamic trajectory data containing real-time latitude, longitude and timestamp through the Automatic Identification System (AIS). The system's underlying layer relies on a highly distributed intelligent waterway perception system to acquire comprehensive navigation environment data. Data acquisition and routing: Through IoT sensing devices deployed along the waterway, on bridges across the river, and on navigation marks—including meteorological stations, current velocity and direction meters, and high-precision water level gauges—the static intrinsic characteristics of target vessels are collected at different high and low frequency cycles. A standardized distributed API gateway cluster then aggregates, converts protocols, and cleans outliers from these static intrinsic characteristics before uniformly encapsulating and pushing them to the shipping big data platform. Master data association: The system pre-establishes a waterway master data asset library and a vessel lifecycle archive based on digital twins. When a target vessel enters a key controlled waterway, the system parses its MMSI code via AIS and retrieves its static physical parameter master data in real time, such as design full-load speed, overall length, beam, and draft.

[0021] S2. Based on latitude, longitude, and timestamps, real-time hydrological and meteorological data are extracted using a waterway sensing system, and spatiotemporal state alignment is performed using a spatiotemporal interpolation algorithm. The specific method for spatiotemporal state alignment is as follows: S201. Based on the real-time latitude and longitude of the target vessel, select multiple hydrological and meteorological sensing nodes that are closest to its upstream and downstream, and use the Kriging spatial interpolation algorithm to calculate the estimated environmental parameters at the current absolute position of the vessel. Spatial interpolation aggregation: Extract the target ship's current precise latitude and longitude, select multiple hydrological / meteorological sensing physical nodes closest to its upstream and downstream, and use the Kriging geostatistical spatial interpolation algorithm to evaluate and calculate the estimated values ​​of the flow field environment and meteorological parameters under the ship's current absolute position coordinates.

[0022] S202. Using the timestamp of the dynamic trajectory data as the reference axis, establish a sliding window-type virtual timestamp window, map the interpolated environmental and meteorological parameter estimates into the virtual timestamp window, and generate a unified state composite vector containing spatiotemporal characteristic attributes.

[0023] Virtual timestamp alignment model: Based on the timestamp of the ship's AIS high-frequency trajectory, it is defined as... Establish a sliding window-style virtual timestamp field The spatially interpolated heterogeneous environmental data is mapped and filled into the discrete grid points of the time window using a time series smoothing algorithm, generating a unified comprehensive state composite vector containing spatiotemporal characteristic attributes. .

[0024] S3. The aligned hydrological data, meteorological data and the static intrinsic features and dynamic trajectory data of the target ship are fused by feature engineering to calculate and generate derived coupling features, and construct a multi-dimensional navigation situation feature vector.

[0025] The derived coupling features include the dynamic margin of safety depth ratio and the crosswind drift risk vector. The dynamic margin of safety depth ratio is calculated by extracting the theoretical reference depth for the navigation segment, adding the real-time water level compensation, subtracting the real-time draft of the target vessel, and then dividing by the real-time draft. This is used to quantify the safety redundancy of the bottom distance. The crosswind drift risk vector is calculated by performing sine decomposition of the real-time wind speed vector and the target vessel's heading vector to the ground, and extracting the crosswind component perpendicular to the ship's heave direction. This is used to quantify the yaw drift risk under complex weather conditions.

[0026] After spatiotemporal alignment, this system performs in-depth feature engineering extraction: Dynamic Under-Keel Clearance Ratio By combining real-time water level elevation, basic geographic information of the navigation section, and ship draft, the risk of bottoming out and towing resistance is quantified using the following formula:

[0027] in, The theoretical reference water depth for the navigation segment, This refers to the real-time water level compensation amount transmitted back and interpolated by the waterway sensing system. Real-time draft of the ship; Relative velocity and crosswind risk vector By using vector projection, the crosswind component that has the greatest impact on the longitudinal maneuvering of the ship is extracted. The formula is as follows:

[0028] in, The wind speed after spatiotemporal alignment. and These are the wind direction angle and the ship's heading angle relative to the ground, respectively.

[0029] S4. Input the multi-dimensional navigation situation feature vector into the pre-trained dynamic speed limit decision network model, calculate and output the adaptive safe speed limit range for the target ship in the current segment. The network structure and reasoning process of the dynamic speed limit decision network model are as follows: A gated recurrent unit (GRU) or a temporal convolutional network (TCN) is used to extract time-series features of the weather and water flow evolution trends within a continuous time step, capturing the dynamic changes in navigation resistance. The model output layer outputs a safe speed limit range containing both a lower and upper speed limit through a quantile regression algorithm. When the crosswind drift risk vector exceeds a set critical threshold and the dynamic margin of water depth is lower than a preset safety threshold, the network model adaptively activates the penalty weight mapping, actively lowering the upper threshold of the safe speed limit range.

[0030] After feature engineering is completed, the system feeds the multi-dimensional navigation situation feature vector into the deep neural network constructed in this invention, which integrates environmental rule constraints, for speed limit inference calculation. The network structure and inference process of the dynamic speed limit decision network model include the following steps: S401, Input and Feature Extraction Layer Receive multidimensional tensor matrices, and use gated recurrent units (GRUs) or temporal convolutional networks (TCNs) to trace back and extract the dynamic evolution trend of meteorological and water flow fields within continuous time steps, such as the "abrupt variance" of wind speed. S402, Adaptive Constraints and Output Layer A quantile regression algorithm is introduced on the output side to output an adaptive safe speed limit range with confidence intervals. When the model detects that the crosswind drift risk vector exceeds the critical threshold, and the dynamic surplus water depth ratio is extremely low due to the dry season, the neural network adaptively activates the penalty weights and nonlinearly adjusts the upper limit of velocity. And fine-tune the lower limit. ; S403, Offline Training and Label Acquisition of the Model Extracting historical safe navigation records of numerous vessels, their actual speeds on specific routes are used as the ground truth for safe speed limits. During training, the Pinball Loss quantile loss function is employed for backpropagation updates of network weights, enabling the model to fit the mean of historical safe speeds and output a speed limit that conforms to a preset safety confidence level (e.g., 95% confidence interval). Speed ​​limit lower limit This allows for offline pre-training of the network. S5. Generate a dedicated dynamic speed limit command based on the adaptive safe speed limit range and send it to the ship terminal for real-time control; at the same time, recalculate the estimated arrival time of the target ship based on the upper limit of the dynamic speed limit and send the updated estimated arrival time to the lock scheduling system to trigger the collaborative automatic optimization of the scheduling scheme.

[0031] The specific methods for triggering the collaborative automatic optimization of the scheduling scheme include: after receiving the updated estimated arrival time, the lock scheduling system matches and recalculates it with the current lock master data queuing sequence and lock chamber resource occupancy status; if it is determined that the target vessel cannot safely arrive at the designated lock according to the original scheduling plan, a multi-objective optimization algorithm is used to automatically move it to the subsequent feasible lock passage batch, and the current lock chamber space capacity is released in advance and reallocated to other queuing vessels that meet the lock passage conditions.

[0032] The method for triggering collaborative automatic optimization of the scheduling scheme includes the following steps: S501, Terminal Command Constraints The calculated safe speed limit range is encrypted digitally and then packaged into a standard control command for issuance. If the actual speed continues to exceed the limit within the set penalty time window, the system will trigger an audible and visual alarm and record the violation. S502, Recalculation of Discrete Calculus for Estimated Arrival Time The system divides the remaining channel from the target vessel to the downstream lock into $n$ discrete mileage segments. Combined with the dynamic speed limit upper limit for each mileage segment output by the model. Based on the ship's current speed, the estimated arrival time is recalculated using the discrete integral formula:

[0033] in, This is the current timestamp. To account for the speed reduction factor due to water flow resistance and navigation density; S503, Coordinated Optimization of Lock Shift Scheduling (Closed-Loop of Comprehensive Transportation Control): The updated high-precision estimated arrival time is pushed through to the lock scheduling system. After receiving the estimated arrival time drift signal, the lock scheduling algorithm kernel performs multi-target state matching and recalculation with the current lock chamber physical scale matrix and pre-queue sequence. If the evaluation result shows that the target vessel cannot arrive within the original strict time window, the algorithm will automatically start reorganization optimization, smoothly move it to the subsequent feasible lock-passing batch, and release the theoretical space capacity of the current lock chamber originally occupied in advance, and redistribute it to other vessels in the queue.

[0034] The collaborative automatic optimization of the trigger-based scheduling scheme of this invention specifically includes: after receiving the updated estimated arrival time, the lock scheduling system matches and recalculates it with the current lock master data queuing sequence and lock chamber resource occupancy status; if it is determined that the target vessel cannot safely arrive at the designated lock according to the original scheduling plan, a multi-objective optimization algorithm is used to automatically move it to a subsequent feasible lock passage batch, and the current lock chamber space capacity is released in advance and reallocated to other queuing vessels that meet the lock passage conditions. The system monitors the waterway traffic density and vessel trajectory response characteristics in real time after the issuance of speed limit instructions. When the flow rate of the target channel section exceeds the safe passage threshold or the vessel trajectory shows abnormal fluctuations, the system triggers an online incremental learning mechanism to extract abnormal feature vectors and update model parameters, further reducing the upper limit benchmark of the safe speed limit for that section. At the same time, the system integrates an abnormal degradation strategy mechanism. When a core sensing node of the waterway perception system goes offline, the system uses the time sequence residual of adjacent nodes to fit and replace it, ensuring high system availability.

[0035] This invention also provides a ship dynamic speed limit and lock coordinated scheduling system based on environmental feature fusion, comprising: a multimodal perception module for acquiring static and dynamic data of the target ship and acquiring high-frequency real-time hydrological and meteorological data through the standard API interface of the distributed waterway perception system; a spatiotemporal alignment and feature fusion module for performing spatial interpolation and temporal alignment on massive heterogeneous data and calculating derived coupling features to construct a multidimensional navigation situation feature vector; a decision module, internally deployed with a deep neural network containing a spatiotemporal attention mechanism, for mapping the multidimensional navigation situation feature vector to a safe dynamic speed limit range; and a control coordination module for converting the calculated speed limit range into underlying control commands and issuing them to the ship traffic service system or shipborne terminal, while simultaneously linking with the backend lock scheduling system to perform estimated arrival time recalculation and lock sequence reordering.

[0036] The core operational logic of the ship navigation dynamic speed limiting system based on multi-dimensional environmental feature fusion of the present invention is deployed on a cloud server or edge computing node, including at least one processor and a memory connected in communication with the processor; the memory stores computer instructions that can be executed by the processor, and when the instructions are executed, they can realize all the methods and steps of the above-mentioned spatiotemporal state alignment, derived feature calculation, AI model speed limiting inference and lock collaborative scheduling.

[0037] This invention encompasses a closed-loop linkage from underlying multimodal environment perception and virtual timestamp feature alignment to top-level lock scheduling. The core working process is as follows: Step 1: Data Access and Spatiotemporal Alignment The system extracts the real-time latitude and longitude coordinates and dynamic trajectory timestamps of the target vessel. It automatically retrieves the raw data collected by the nearest hydrological monitoring station and meteorological observation station through the API gateway. With the help of spatial interpolation algorithm and time series completion algorithm, it unifies the heterogeneous data from different sensing devices with different collection frequencies, and finally achieves complete alignment of high and low frequency data in time and space.

[0038] Step 2: Construction of Core Security Features Based on aligned environmental data and ship parameters, a dynamic margin of water depth is generated through feature cross-calculation to quantify the risk of a ship hitting the bottom and running aground, and a crosswind drift risk vector is generated to quantify the lateral yaw resistance of a ship under strong crosswind conditions. This forms a navigation situation feature vector that covers multiple safety constraints, providing a basic input for subsequent speed limit decisions.

[0039] Step 3: AI Model-Driven Speed ​​Limit Decision Reasoning A dynamic speed limit decision network model based on a deep learning framework is introduced. This model comprehensively evaluates the coupling risk between the dynamic changes in the current water environment and the target vessel's own maneuverability, and finally outputs a safe speed range that adapts to the vessel's current navigation state. This range includes both the maximum speed limit and the minimum speed limit that are allowed to navigate.

[0040] Step 4: Dynamic speed limit command issuance and coordinated operation with the lock hub Based on the calculated safe speed range, the system generates a dynamic speed limit control command applicable only to the target vessel and sends it to the vessel terminal to achieve real-time constraints on vessel speed. At the same time, based on the new speed limit upper limit, the system recalculates the estimated arrival time of the target vessel at the downstream lock and actively pushes the updated estimated arrival time to the lock scheduling system to trigger the automatic rearrangement and optimization of the lock scheduling plan, thereby forming a closed-loop linkage between waterway control and hub scheduling.

[0041] The embodiments of the present invention have been described above with reference to the accompanying drawings. However, the present invention is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of the present invention without departing from the spirit and scope of the claims. All of these forms are within the protection scope of the present invention.

Claims

1. A method for coordinated scheduling of ship dynamic speed limits and locks based on environmental feature integration, characterized in that, Includes the following steps: S1. Obtain the static intrinsic characteristics of the target vessel and acquire dynamic trajectory data containing real-time latitude, longitude and timestamp through the Automatic Identification System (AIS). S2. Based on latitude, longitude and timestamp, real-time hydrological and meteorological data are extracted through the waterway perception system, and spatiotemporal interpolation algorithm is used to align spatiotemporal states. S3. The aligned hydrological data, meteorological data and the static intrinsic features and dynamic trajectory data of the target ship are fused by feature engineering to calculate and generate derived coupled features, and construct a multi-dimensional navigation situation feature vector. S4. Input the multi-dimensional navigation situation feature vector into the pre-trained dynamic speed limit decision network model, calculate and output the adaptive safe speed limit range for the target ship in the current segment. S5. Generate a dedicated dynamic speed limit command based on the adaptive safe speed limit range and send it to the ship terminal for real-time control; at the same time, recalculate the estimated arrival time of the target ship based on the upper limit of the dynamic speed limit and send the updated estimated arrival time to the lock scheduling system to trigger the collaborative automatic optimization of the scheduling scheme.

2. The method for ship dynamic speed limiting and lock coordinated scheduling based on environmental feature fusion according to claim 1, characterized in that, In step S1, the static intrinsic characteristics of the target vessel are collected at different high and low frequency cycles by IoT sensing devices such as meteorological stations, flow velocity and direction instruments, and high-precision water level gauges deployed along the waterway, on cross-river bridges and navigation marks. The static intrinsic characteristics of the target vessel are aggregated, converted, and cleaned of outliers by a standardized distributed API gateway cluster, and then uniformly packaged and pushed to the shipping big data platform.

3. The method for ship dynamic speed limiting and lock coordinated scheduling based on environmental feature fusion according to claim 1, characterized in that, In step S2, the specific method for performing spatiotemporal state alignment is as follows: S201. Extract the current precise latitude and longitude of the target vessel, select multiple hydrological and meteorological sensing nodes that are closest to its upstream and downstream, and use the Kriging statistical spatial interpolation algorithm to evaluate and calculate the estimated values ​​of the flow field environment and meteorological parameters under the current absolute position coordinates of the vessel. S202. Using the timestamp of the dynamic trajectory data as the reference axis, establish a sliding window-type virtual timestamp window, map the interpolated environmental and meteorological parameter estimates into the virtual timestamp window, and generate a unified state composite vector containing spatiotemporal characteristic attributes.

4. The method for coordinated scheduling of ship dynamic speed limits and locks based on environmental feature fusion according to claim 1, characterized in that, In step S3, the derived coupling features include dynamic margin water depth ratio and crosswind drift risk vector; The dynamic margin of water depth is calculated as follows: extract the theoretical reference water depth of the water section, add the real-time water level compensation, subtract the real-time draft of the target vessel, and then divide by the real-time draft to obtain the result, which is used to quantify the bottom distance safety redundancy. The crosswind drift risk vector is calculated by performing sine decomposition on the real-time wind speed vector and the target ship's heading vector to the ground, and extracting the crosswind component perpendicular to the ship's pitch direction, which is used to quantify the yaw drift risk under complex weather conditions.

5. The method for coordinated scheduling of ship dynamic speed limits and locks based on environmental feature fusion according to claim 4, characterized in that, The dynamic margin of water depth, which integrates real-time water level elevation, basic geographic information of the navigation segment, and vessel draft, quantifies the risk of bottoming out and towing resistance. The formula is as follows: in, The theoretical reference water depth for the navigation segment, This refers to the real-time water level compensation amount transmitted back and interpolated by the waterway sensing system. Real-time draft of the ship; The relative flow velocity and crosswind drift risk vector are used to extract the crosswind component that has the greatest impact on the longitudinal maneuvering of the ship through vector projection. The formula is as follows: in, The wind speed after spatiotemporal alignment. and These are the wind direction angle and the ship's heading angle relative to the ground, respectively.

6. The method for coordinated scheduling of ship dynamic speed limits and locks based on environmental feature fusion according to claim 1, characterized in that, In step S4, the network structure and reasoning process of the dynamic speed limit decision network model are as follows: using gated recurrent units or temporal convolutional networks to extract time series features of the weather and water flow evolution trends within a continuous time step, capturing the dynamic change characteristics of navigation resistance, and outputting a safe speed limit range containing the lower and upper speed limits through the quantile regression algorithm of the model output layer; when the crosswind drift risk vector exceeds the set critical threshold and the dynamic margin water depth ratio is lower than the preset safety threshold, the network model adaptively activates the penalty weight mapping and actively lowers the upper threshold of the safe speed limit range.

7. The method for coordinated scheduling of ship dynamic speed limits and locks based on environmental feature fusion according to claim 6, characterized in that, The network structure and reasoning process of the dynamic speed limiting decision network model include the following steps: S401: Receive a multidimensional tensor matrix, and use a gated recurrent unit or a temporal convolutional network to backtrack and extract the dynamic evolution trend of meteorological and water flow fields within a continuous time step. S402. Introduce a quantile regression algorithm on the output side to output an adaptive safe speed limit range with confidence intervals. When the model detects that the crosswind drift risk vector exceeds the critical threshold, and the dynamic surplus water depth ratio is extremely low due to the dry season, the neural network adaptively activates the penalty weights and nonlinearly adjusts the upper limit of velocity. And fine-tune the lower limit. ; S403. Extract historical safe navigation records of a large number of ships and use their actual speeds on specific sections as a reference for safe speed limits. During the training phase, use the quantile loss function for backpropagation updates of network weights, enabling the model to fit the mean of historical safe speeds and output a speed limit upper limit that meets the preset safety confidence level. Speed ​​limit lower limit This completes the offline pre-training of the network.

8. The method for coordinated scheduling of ship dynamic speed limits and locks based on environmental feature fusion according to claim 1, characterized in that, In step S5, the method for triggering the collaborative automatic optimization of the scheduling scheme specifically includes: after the lock scheduling system receives the updated estimated arrival time, it matches and recalculates it with the current lock master data queuing sequence and lock chamber resource occupancy status; if it is determined that the target vessel cannot safely arrive at the designated lock according to the original scheduling plan, it is automatically moved to the subsequent feasible lock passage batch using a multi-objective optimization algorithm, and the current lock chamber space capacity is released in advance and redistributed to other queuing vessels that meet the lock passage conditions.

9. The method for coordinated scheduling of ship dynamic speed limits and locks based on environmental feature fusion according to claim 8, characterized in that, The method for triggering collaborative automatic optimization of the scheduling scheme includes the following steps: S501. The calculated safe speed limit range is digitally encrypted and encapsulated into a standard control command for issuance. If the actual speed continues to exceed the limit within the set penalty time window, the system will trigger an audible and visual alarm and record the violation. S502. Divide the remaining channel from the target vessel to the downstream lock into $n$ discrete mileage segments. Combined with the dynamic speed limit upper limit for each mileage segment output by the model. Based on the ship's current speed, the estimated arrival time is recalculated using the discrete integral formula: in, This is the current timestamp. To account for the speed reduction factor due to water flow resistance and navigation density; S503, the updated high-precision estimated arrival time is pushed through to the lock scheduling system; after receiving the estimated arrival time drift signal, the lock scheduling algorithm kernel performs multi-target state matching and recalculation with the current lock chamber physical scale matrix and pre-queue sequence; if the evaluation result shows that the target vessel cannot arrive within the original strict time window, the algorithm will automatically start reorganization optimization, smoothly move it to the subsequent feasible lock passage batch, and release the theoretical space capacity of the current lock chamber originally occupied in advance, and redistribute it to other vessels in the queue.

10. A ship dynamic speed limiting and lock coordinated scheduling system based on environmental feature fusion, characterized in that, include: The multimodal sensing module is used to acquire static and dynamic data of the target vessel and obtain high-frequency real-time hydrological and meteorological data through the standard API interface of the distributed waterway sensing system. The spatiotemporal alignment and feature fusion module is used to perform spatial interpolation and temporal alignment on massive heterogeneous data, and to calculate derived coupled features to construct multidimensional navigation situation feature vectors; The decision-making module is internally deployed with a deep neural network containing a spatiotemporal attention mechanism, which is used to map multidimensional navigation situation feature vectors into safe dynamic speed limit zones. The control and coordination module is used to convert the calculated speed limit range into underlying control commands and send them to the ship traffic service system or shipboard terminal. At the same time, it coordinates with the back-end lock scheduling system to recalculate the estimated arrival time and rearrange the lock sequence.

Images