A ship automatic berthing safety control system

Abstract

This invention relates to an automatic berthing safety control system for ships, belonging to the field of port vessel operation safety control technology. The system comprises four core modules connected sequentially to form a closed-loop control system: a situational synchronization perception module generates full-element dynamic attitude information through spatiotemporal synchronization and data fusion of ship-shore inertial units; a digital twin simulation module performs real-time simulation based on a six-degree-of-freedom motion model of the ship, proactively predicting the ship's trajectory and berth forces; a safety decision and early warning module compares the predicted trajectory with the safe berthing corridor using a trajectory-berth force coupling algorithm, outputting a collaborative control strategy; and a multi-terminal collaborative execution module converts the strategy into instructions and synchronously distributes them to tugboats, the bridge, and the terminal system. This achieves fully automatic closed-loop control from perception, prediction, decision-making to execution, effectively improving the safety, accuracy, and operational efficiency of ship berthing.

Description

Technical Field

[0001] This invention belongs to the field of port vessel operation safety control technology, specifically relating to an automatic berthing safety control system for ships. Background Technology

[0002] With the continued growth of global maritime trade, port terminals, as hubs of sea and land transportation, have seen their operational efficiency and safety management levels become core industry demands. Ship berthing, a crucial link in port operations, requires the precise berthing of commercial vessels ranging from tens of thousands to hundreds of thousands of tons to designated berths within limited water space. This process necessitates addressing multiple complex factors simultaneously, including ship inertia, wind and current disturbances, and limitations imposed by terminal facilities. The precision and safety of this operation directly determine port operational efficiency, equipment lifespan, and personnel safety.

[0003] Currently, ship berthing operations still primarily rely on the traditional model of "manual supervision + auxiliary equipment": captains use their navigation experience to determine the ship's position and attitude, coordinate tugboat operations via walkie-talkie, and adjust the ship's speed and direction using visual guidance from shore-based personnel. While some ports have introduced single sensors or simple automated equipment, a systematic management and control capability has not yet been established. This traditional model has significant technical limitations:

[0004] Manual operation relies on experience and has a low margin for error. The captain's judgment of the ship's attitude is easily affected by subjective factors such as visibility, blind spots, and psychological state. Especially in bad weather or at night, problems such as deviations in track judgment and delays in tugboat coordination instructions are likely to occur, increasing the risk of safety accidents such as collisions between the ship and the dock fenders and cable breakage.

[0005] The data from multiple sources is asynchronous and has low fusion. The data collected by the existing auxiliary equipment belongs to independent systems, and there is no unified time reference at both the ship and shore ends, resulting in misalignment of data spatiotemporal labels. At the same time, no effective data fusion mechanism has been established, and single data is easily interfered with, making it impossible to generate continuous and accurate dynamic attitude information of all ship elements, which makes it difficult to support reliable berthing decisions.

[0006] Digital simulation is disconnected from actual operations. Some ports have attempted to introduce digital models to simulate the berthing process, but these models are mostly static parameter configurations and cannot receive dynamic data such as tugboat thrust and instantaneous wind and current disturbances in real time. This results in a large deviation between the simulation results and the actual ship motion, making them only suitable for pre-operation drills and unable to provide decision support for real-time berthing.

[0007] Multi-terminal collaborative response is lagging. In the traditional mode, berthing decisions require multiple steps: "captain's judgment - instruction transmission - tugboat execution - dock coordination". Instructions are transmitted via walkie-talkie or manual announcement, which is prone to information loss or delay. Especially in emergency scenarios, the delay in manual instruction transmission, combined with the inertia of the ship, can easily cause the best adjustment opportunity to be missed, requiring emergency braking or additional tugboat support, which not only increases operating costs but may also cause secondary safety risks. Summary of the Invention

[0008] To address the aforementioned problems in the existing technology, this invention provides an automatic berthing safety control system for ships. The objective of this invention can be achieved through the following technical solution:

[0009] An automatic berthing safety control system for ships includes four core modules connected in sequence to form a closed-loop control:

[0010] The situational awareness module is used to build a ship-shore integrated collaborative awareness network. By synchronizing and fusing the ship-end inertial measurement unit and the dock-end inertial navigation unit under a unified time base, it integrates real-time collected ship position, speed, direction of travel and wind / current disturbance data to generate dynamic attitude information of all ship elements with a unified spatiotemporal reference.

[0011] The digital twin simulation module is connected to the situational awareness module and receives the dynamic attitude information of all elements. It has a built-in six-degree-of-freedom motion model of the ship and a berth environment model. It takes real-time data and tugboat thrust as inputs for calculation and performs real-time simulation of the ship berthing process in virtual space. It also uses a built-in prediction algorithm to make forward-looking predictions of the ship's future trajectory, motion attitude and interaction forces with the berth.

[0012] The safety decision and early warning module is connected to the digital twin simulation module and embeds a trajectory-berth force coupling prediction algorithm based on spatiotemporal synchronization. By comparing the ship's motion trajectory predicted by the digital twin simulation module with the preset safe berthing corridor, the module assesses the ship's impact force on the berth, dynamically determines the safety status of the berthing process, and outputs a collaborative control strategy based on the prediction results.

[0013] The multi-terminal collaborative execution module, connected to the safety decision and early warning module, converts the collaborative control strategy into executable instructions and synchronously distributes them to the corresponding tugboats, ship bridges, and dock systems.

[0014] As a preferred embodiment of the present invention, in the situational awareness synchronization module, the synchronization of the unified time base is achieved through a network time protocol and timing satellite signals, specifically including:

[0015] The network time protocol includes deploying time servers at both the ship and dock ends, exchanging time synchronization messages through redundant network links including redundant Ethernet and 5G industrial private network, and compensating for message transmission delays.

[0016] The timing satellite signal is received by a dual-mode timing receiver configured at both the ship and the dock, and multi-satellite joint calculation is used to eliminate signal interference. After synchronization is completed, the spatiotemporal labels of all sensor data at both the ship and the dock are based on the unified time base and associated with the dock reference point.

[0017] Specifically, the situational awareness module uses Kalman filtering to fuse and complement the dynamic data from the ship-side inertial measurement unit and the static reference data from the dock-side inertial navigation unit. The data fusion process includes:

[0018] Using static reference data at the dock as a reference, dynamic data at the ship end is predicted; the measured values ​​of the dynamic data are compared with the predicted values, and the predicted values ​​are corrected by the Kalman gain matrix to compensate for sensor errors; the fused data is smoothed to generate the full-element dynamic attitude information.

[0019] Specifically, in the digital twin simulation module, the ship's six-degree-of-freedom motion model is a parameterized mathematical model, which is further coupled with the ship's hydrodynamic coefficients obtained from ship design data and mooring tests, as well as the ship's load parameters acquired in real time through the draft sensors deployed on the ship; wherein, the ship's hydrodynamic coefficients are stored and retrieved in segments according to the ship's speed range, and the ship's load parameters are used to correct the buoyancy center position and inertia matrix in the model in real time.

[0020] Specifically, the time window for the digital twin simulation module to predict the future trajectory and attitude of the ship is configurable and can be manually or automatically adjusted according to the current berthing stage, ship handling performance, or external environmental conditions.

[0021] Specifically, in the safety decision and early warning module, the boundary of the safe berthing corridor is dynamically adjusted by a calculation model based on real-time input environmental data, ship tonnage, and windward area; the environmental data includes wind speed and direction, and current speed and direction.

[0022] Specifically, the safety decision and early warning module is preset with at least two levels of safety thresholds, including a warning level and an alarm level. When the predicted deviation from the flight path or the impact force reaches the warning level threshold, the system issues a warning to the operator. When the predicted value exceeds the alarm level threshold, the system automatically triggers the collaborative control strategy and hands it over to the multi-terminal collaborative execution module for execution. At the same time, the warning status is upgraded, including upgrading the visual warning from yellow flashing to red flashing and the audible warning from intermittent beeping to continuous beeping.

[0023] Specifically, the optimal cooperative control strategy is generated based on model predictive control algorithm. Under the premise of satisfying ship dynamics constraints and berth space geometric constraints, a sequence of commands with the optimal objective function is solved through mathematical programming. The objective function is used to comprehensively optimize berthing time, energy consumption and safety risks.

[0024] Specifically, the instructions distributed to the tugboats by the multi-terminal collaborative execution module are specific thrust magnitude and azimuth angle instructions calculated by the thrust distribution algorithm for each participating tugboat.

[0025] Specifically, the instructions distributed by the multi-terminal collaborative execution module to the terminal system include status querying and linkage control of the terminal's collision avoidance and mooring facilities; the linkage control instructions are triggered based on the prediction results of the safety decision and early warning module at different berthing distances.

[0026] Specifically, the multi-terminal collaborative execution module uses a communication protocol and link with redundancy design for instruction transmission. The redundancy design includes at least one wired communication link and one wireless communication link. The transmitted instructions include a sequence number and a check code for verifying integrity.

[0027] Specifically, the system constructs a learning and optimization unit based on historical berthing data. The unit uses machine learning algorithms to self-correct and optimize specific parameters in the ship's six-degree-of-freedom motion model and the boundary thresholds of the safe berthing corridor.

[0028] The beneficial effects of this invention are as follows:

[0029] The situational awareness module employs a dual synchronization mechanism of "Network Time Protocol (NTP) + timing satellite signal," coupled with a redundant Ethernet and 5G industrial private network link design. Through message delay compensation and multi-satellite joint calculation, the ship-shore clock deviation is controlled at the microsecond level, ensuring that the spatiotemporal labels of all sensor data are consistent and associated with the dock reference point. Kalman filtering is then used to fuse the ship-side IMU dynamic data and the dock-side INS static data, effectively compensating for sensor drift errors and ultimately generating full-element dynamic attitude information. This design completely solves the problems of spatiotemporal data misalignment and susceptibility to interference with single data points in traditional models, providing continuous and accurate basic data support for subsequent simulations and decisions, and avoiding trajectory judgment errors caused by data deviations.

[0030] The six-degree-of-freedom motion model of the digital twin deduction module for ships couples the hydrodynamic coefficients obtained from ship design data and mooring tests, and combines the load conditions parameters collected in real time by the draft sensors, making the model calculation results close to the actual motion state of the ship. At the same time, the configurable advanced deduction time window can prospectively predict the future track, attitude of the ship and the interaction force with the berth. This design upgrades digital simulation from "pre-event rehearsal" to "real-time decision support", can identify potential yaw risks in advance, reserve sufficient adjustment time for safety decisions, and avoid the passive situation of "discovering risks means approaching risks" in the traditional mode.

[0031] The safety decision-making and early warning module dynamically adjusts the boundaries of the safe berthing corridor through real-time environmental data, ship tonnage, and windward area, ensuring that the corridor boundaries always match the actual operation scenario. Then, combined with two-level safety thresholds, it realizes hierarchical response - the warning level triggers audible and visual warnings to remind the operator to intervene, and the alarm level automatically triggers collaborative control strategies and upgrades the warnings. Compared with the traditional manual experience judgment mode, this design greatly improves the risk identification efficiency, and accurately evaluates the impact force through the track-berth force coupling prediction algorithm, which can effectively avoid dock collision accidents caused by artificial visual blind spots and misjudgments in bad weather, and significantly improve the safety of berthing operations.

[0032] The multi-terminal collaborative execution module outputs precise thrust magnitude and direction angle commands for each tugboat through the thrust distribution algorithm to ensure the efficiency of tugboat assistance operations; for the dock system, it triggers linkage control according to the berthing distance to achieve coordinated protection of ship-shore facilities; at the same time, the redundant communication link and the design of instruction sequence numbers and CRC32 check codes greatly improve the instruction response speed compared with the traditional VHF intercom transmission. This design completely打通 the "decision - execution" link, avoiding improper adjustments caused by instruction delays or deviations. Especially in emergency scenarios, it can quickly link tugboats, the bridge, and the dock system to seize the best adjustment opportunity. Brief Description of the Drawings

[0033] For the convenience of those skilled in the art to understand, the present invention will be further described below in conjunction with the drawings.

[0034] Figure 1 It is a schematic flow diagram of a ship automatic berthing safety control system of the present invention;

[0035] Figure 2 It is an architecture diagram of the digital twin deduction module of the present invention;

[0036] Figure 3 It is an architecture diagram of the safety decision-making and early warning module of the present invention. Detailed Embodiments

[0037] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided.

[0038] Please see Figure 1 An automatic berthing safety control system for ships includes four core modules connected in sequence to form a closed-loop control:

[0039] The situational awareness module is used to build a ship-shore integrated collaborative awareness network. By synchronizing and fusing the ship-end inertial measurement unit and the dock-end inertial navigation unit under a unified time base, it integrates real-time collected ship position, speed, direction of travel and wind / current disturbance data to generate dynamic attitude information of all ship elements with a unified spatiotemporal reference.

[0040] The digital twin simulation module is connected to the situational awareness module and receives the dynamic attitude information of all elements. It has a built-in six-degree-of-freedom motion model of the ship and a berth environment model. It takes real-time data and tugboat thrust as inputs for calculation and performs real-time simulation of the ship berthing process in virtual space. It also uses a built-in prediction algorithm to make forward-looking predictions of the ship's future trajectory, motion attitude and interaction forces with the berth.

[0041] The safety decision and early warning module is connected to the digital twin simulation module and embeds a trajectory-berth force coupling prediction algorithm based on spatiotemporal synchronization. By comparing the ship's motion trajectory predicted by the digital twin simulation module with the preset safe berthing corridor, the module assesses the ship's impact force on the berth, dynamically determines the safety status of the berthing process, and outputs a collaborative control strategy based on the prediction results.

[0042] The multi-terminal collaborative execution module, connected to the safety decision and early warning module, converts the collaborative control strategy into executable instructions and synchronously distributes them to the corresponding tugboats, ship bridges, and dock systems.

[0043] Specifically, in the situational awareness synchronization module, the synchronization of the unified time base is achieved through network time protocols and timing satellite signals, specifically including:

[0044] The network time protocol includes deploying time servers at both the ship and dock ends, exchanging time synchronization messages through redundant network links including redundant Ethernet and 5G industrial private network, and compensating for message transmission delays.

[0045] The timing satellite signal is received by a dual-mode timing receiver configured at both the ship and the dock, and multi-satellite joint calculation is used to eliminate signal interference. After synchronization is completed, the spatiotemporal labels of all sensor data at both the ship and the dock are based on the unified time base and associated with the dock reference point.

[0046] In this embodiment, one of the core tasks of the situational awareness module is to build a unified spatiotemporal reference for ship-shore integration.

[0047] The unified time base synchronization is achieved through two parallel or selectively chosen technical paths to form redundancy backups and ensure synchronization reliability under any operating conditions:

[0048] Path 1: Synchronization based on high-precision network time protocol

[0049] Hardware deployment: A secondary master clock time server is deployed on both the ship's bridge and the dock control center. This server supports the IEEE 1588 (PTP) precision time protocol.

[0050] Network Links: The ship and dock are connected via a dual-link redundant network. The primary link is an industrial Ethernet network composed of fiber optic cables, and the backup link is a 5G industrial private network based on 5G CPE devices. Both links transmit PTP synchronization messages simultaneously.

[0051] Synchronization Process and Delay Compensation: The master clock at the dock serves as the time source for the entire system, periodically sending Sync and Follow-Up messages to the slave clock on the ship. The slave clock on the ship records the message arrival time and calculates the network transmission delay. Through the delay request-response mechanism defined by the PTP protocol, combined with message timestamps, the system can accurately calculate and compensate for message transmission delays in redundant Ethernet and 5G networks.

[0052] Path 2: Synchronization based on precise timing satellite signals

[0053] Hardware deployment: Beidou / GPS dual-mode timing receivers are installed on unobstructed areas of the ship's compass deck and at dock reference points.

[0054] Synchronization Process and Interference Resistance: The timing receivers at both the ship and dock ends simultaneously receive BeiDou-2 B1I signals and GPS L1 C / A signals. The receivers employ a multi-satellite joint calculation algorithm, receiving signals from multiple navigation satellites and using Kalman filtering technology to fuse the calculations. This effectively suppresses multipath effects and signal attenuation caused by large port cranes, buildings, or temporary obstructions, ensuring that even in complex port environments, single-station timing accuracy can reach the 100 nanosecond level.

[0055] Cold backup mechanism: When the synchronization accuracy of the network time protocol decreases due to network congestion or failure, the system automatically switches to satellite timing signals as the highest reference.

[0056] After synchronization, all data collected by sensors at the ship's end (IMU, GPS / BDS receiver, anemometer) and at the dock end (INS, current profiler, etc.) are assigned a millisecond-precision timestamp based on the aforementioned unified time base, and their spatial coordinates are unified to a local coordinate system (such as the UTM coordinate system) with the dock's reference point as the origin. This ensures that all data are fully aligned in both spatiotemporal dimensions, laying a solid foundation for subsequent data fusion.

[0057] After obtaining spatiotemporally unified sensor data, the system uses the Extended Kalman Filter (EKF) algorithm to fuse and complement the dynamic data from the ship-side IMU and the static reference data from the dock-side INS. The specific fusion process is as follows:

[0058] Using the static reference (fixed position and attitude) provided by the high-precision INS at the dock as the initial value of the system state, and combining the angular rate and linear acceleration measured by the IMU at the ship, the ship's attitude, velocity and position at the next moment are predicted through the ship's kinematic equations.

[0059] The actual position / velocity measurements provided by the GPS / BeiDou positioning system are compared with the predicted values ​​to obtain the residuals. These residuals are then used to optimally correct the predicted values ​​by calculating the Kalman gain matrix. The core function of this process is to estimate and compensate for inherent sensor errors of the IMU in real time, such as the accelerometer's zero bias and the gyroscope's random walk error, thereby significantly suppressing the cumulative drift of inertial navigation.

[0060] The fused state estimation data is subjected to moving average filtering or low-pass filtering to remove impulse noise caused by transient electromagnetic interference or vibration.

[0061] This embodiment employs a redundant synchronization scheme combining hardware and software to ensure that the data from the ship-shore sensing network achieves spatiotemporal uniformity at its source. Furthermore, through advanced data fusion algorithms, dynamic and static, absolute and relative positioning data are combined to complement each other, ultimately outputting a continuous, stable, and high-precision set of ship dynamic attitude information. This provides a reliable data foundation for subsequent digital twin simulations, preventing model inaccuracies and decision-making errors caused by inconsistent data benchmarks from the outset.

[0062] Specifically, the situational awareness module uses Kalman filtering to fuse and complement the dynamic data from the ship-side inertial measurement unit and the static reference data from the dock-side inertial navigation unit. The data fusion process includes:

[0063] Using static reference data at the dock as a reference, dynamic data at the ship end is predicted; the measured values ​​of the dynamic data are compared with the predicted values, and the predicted values ​​are corrected by the Kalman gain matrix to compensate for sensor errors; the fused data is smoothed to generate the full-element dynamic attitude information.

[0064] Specifically, in the digital twin simulation module, the ship's six-degree-of-freedom motion model is a parameterized mathematical model, which is further coupled with the ship's hydrodynamic coefficients obtained from ship design data and mooring tests, as well as the ship's load parameters acquired in real time through the draft sensors deployed on the ship; wherein, the ship's hydrodynamic coefficients are stored and retrieved in segments according to the ship's speed range, and the ship's load parameters are used to correct the buoyancy center position and inertia matrix in the model in real time.

[0065] In this embodiment, the core of the digital twin simulation module is to establish a six-degree-of-freedom motion model that can accurately reflect the real motion characteristics of the ship. This includes the construction method of the parameterized mathematical model, the acquisition methods of key parameters, and the real-time correction mechanism.

[0066] The six-degree-of-freedom motion model of the ship is based on the Newton-Euler equations, describing the ship's linear motion (pitch, sway, heave) and angular motion (roll, pitch, bow) in the hull coordinate system. Its general vector form is:

[0067] M · ̇ν + C(ν) · ν + D(ν) · ν + G(η) = τ

[0068] in:

[0069] M is the system inertia matrix, which includes the ship's inherent mass inertia and the fluid-added mass inertia.

[0070] C(ν) is the Coriolis centripetal force matrix.

[0071] D(ν) is the fluid damping matrix.

[0072] G(η) is the restoring force / torque vector (generated by gravity and buoyancy).

[0073] τ is the vector of external environmental forces / torques (including tug thrust, wind, flow, etc.).

[0074] ν is the velocity vector in the ship's coordinate system, and η is the pose vector in the Earth's coordinate system.

[0075] These coefficients are fundamental to the model's accuracy and are precisely obtained and structured for storage through the following methods:

[0076] The hull geometry is extracted from the ship's lines drawing, and the preliminary added mass coefficient and linear damping coefficient are calculated using the slicing method or the surface element method.

[0077] The hydrostatic test provides parameters related to restoring force, such as the position of the center of buoyancy, the height of the epicenter, and the tons per centimeter of draft.

[0078] Based on constrained model tests, complex hydrodynamic parameters such as nonlinear damping coefficient and fluid memory effect are accurately obtained by simulating the forced oscillation and dragging of ships in the flow field.

[0079] Because hydrodynamic coefficients (especially damping coefficients) exhibit significant nonlinearity at different speeds, the system stores key coefficients in three main ranges based on the ship's speed relative to ground:

[0080] Range I (low-speed range: 0-2 knots): Damping is mainly linear with relatively small coefficient values, suitable for the final stage of berthing.

[0081] In section II (low-speed zone: sections 2-5): the nonlinear damping effect begins to appear, and the coefficient value increases significantly.

[0082] Interval III (Control Zone: Sections 5-8): Damping force is correlated with the square or even higher powers of the velocity, with the largest coefficient value.

[0083] During each real-time simulation step, the model automatically calls the coefficient matrix of the corresponding speed range for interpolation calculation based on the calculated instantaneous speed, ensuring the accuracy of the dynamic response.

[0084] To achieve dynamic consistency between the model and the actual ship, the system acquires onboard information in real time through a sensor network deployed on the ship:

[0085] Six high-precision draft sensors are deployed on both the bow, midship, and stern sides of the ship to measure the draft at each point in real time.

[0086] Average draft T_avg: Calculated from the draft at the bow, midships and stern, used to query the hydrostatic curve and determine the ship's displacement Δ.

[0087] Trim: Calculated from the difference between the forward draft and the aft draft.

[0088] Heel: Calculated from the difference in draft between the port and starboard sides.

[0089] The ship's loading parameters are used in each simulation step to dynamically correct the model's center of buoyancy position and inertia matrix:

[0090] The system uses real-time average draft T_avg and trim value Trim to interpolate the precise three-dimensional coordinates (x_b, y_b, z_b) of the center of buoyancy in the current state from a pre-stored hydrostatic data table. These coordinates are directly used to calculate the restoring moment G(η). When the ship's draft changes due to loading / unloading or ballast water adjustments, this correction ensures that the point of application of buoyancy and gravity is accurately simulated.

[0091] Mass inertia matrix update: Based on the real-time displacement Δ, recalculate the ship's total mass and moments of inertia about the three coordinate axes.

[0092] Additional mass matrix update: The additional mass coefficient is directly related to the displacement. The system updates the additional mass item in the M matrix by scaling the stored hydrodynamic coefficients proportionally according to the current Δ.

[0093] Workflow example:

[0094] The system reads a draft of 8.5 meters at the front, 9.5 meters at the rear, and an average draft of 9.0 meters, with no list.

[0095] Based on this, the model calculates that the current displacement is 85,000 tons and the center of buoyancy has moved backward by 0.8 meters compared to the previous moment;

[0096] The model then calls the inertial parameters corresponding to 85,000 tons, and calls the hydrodynamic coefficients of section II based on the current speed of 2.5 knots;

[0097] The digital twin engine uses these updated parameters to perform dynamic calculations for the next time step, thus predicting the ship's motion.

[0098] This embodiment solves the key problems of model rigidity and disconnect from the actual ship state in traditional berthing simulations by constructing a deeply parameterized ship motion model with dynamic self-correction capabilities. Through speed interval management of hydrodynamic coefficients and online correction of model parameters based on real-time draft, it ensures that the ship's dynamic characteristics in the digital twin virtual space remain highly consistent with the physical entity in the real world, providing a real and reliable physical basis for subsequent safety decisions and precise control.

[0099] Specifically, the time window for the digital twin simulation module to predict the future trajectory and attitude of the ship is configurable and can be manually or automatically adjusted according to the current berthing stage, ship handling performance, or external environmental conditions.

[0100] Specifically, in the safety decision and early warning module, the boundary of the safe berthing corridor is dynamically adjusted by a calculation model based on real-time input environmental data, ship tonnage, and windward area; the environmental data includes wind speed and direction, and current speed and direction.

[0101] Specifically, the safety decision and early warning module is preset with at least two levels of safety thresholds, including a warning level and an alarm level. When the predicted deviation from the flight path or the impact force reaches the warning level threshold, the system issues a warning to the operator. When the predicted value exceeds the alarm level threshold, the system automatically triggers the collaborative control strategy and hands it over to the multi-terminal collaborative execution module for execution. At the same time, the warning status is upgraded, including upgrading the visual warning from yellow flashing to red flashing and the audible warning from intermittent beeping to continuous beeping.

[0102] Specifically, the optimal cooperative control strategy is generated based on model predictive control algorithm. Under the premise of satisfying ship dynamics constraints and berth space geometric constraints, a sequence of commands with the optimal objective function is solved through mathematical programming. The objective function is used to comprehensively optimize berthing time, energy consumption and safety risks.

[0103] In this embodiment, the core function of the safety decision and early warning module is to generate the optimal cooperative control strategy and, based on the model predictive control algorithm, solve a multi-objective optimization instruction sequence under multiple constraints.

[0104] Model predictive control (MMC) is a model-based closed-loop optimization control strategy. Its core principle can be summarized as "rolling time domain, iterative optimization." In this system, its workflow is as follows:

[0105] The ship's six-degree-of-freedom motion model in the digital twin simulation module is used as the internal prediction model. This model can accurately predict the ship's future trajectory based on the current state and control input.

[0106] The system performs a complete optimization calculation every control cycle (e.g., every 0.5 seconds). The optimization is performed for a future finite time domain (e.g., the next 30 seconds), generating a sequence of control commands, but only the first command in the sequence is executed. In the next cycle, the system re-optimizes based on the latest ship status, and this process is repeated.

[0107] In each control cycle, MPC transforms a multi-objective optimization problem into a mathematical programming problem for solution. This problem comprises three core components: the objective function, dynamic constraints, and geometric constraints.

[0108] The objective function J is a comprehensive performance metric used to quantitatively evaluate the merits of different control strategies. It is designed to minimize the weighted sum of the following three key metrics:

[0109] J = w_t * J_time + w_e * J_energy + w_s * J_safety

[0110] in:

[0111] J_time (berthing time): proportional to the average distance between the vessel and the target berth within the prediction time domain. A weighting coefficient w_t = 0.3 is used to ensure berthing efficiency.

[0112] J_energy (energy consumption): Primarily calculates the sum of the squares of the thrust of all tugboats in the prediction time domain (reflecting energy consumption). The weighting coefficient w_e = 0.2 is intended to save operating costs.

[0113] J_safety (Safety Risk): Consists of two sub-items: (1) the integral of the deviation between the predicted trajectory and the boundary of the safe berthing corridor; (2) the difference between the predicted ship-berth impact force and its safety threshold. The weighting coefficient w_s = 0.5 assigns the highest priority to the safety target.

[0114] Optimization must meet the ship's own physical maneuvering limits, including:

[0115] Main unit speed constraint: Main unit speed_min ≤ N ≤ Main unit speed_max (wherein, the main unit speed_max is usually not higher than 110% of the rated speed).

[0116] Rudder angle range constraint: -35° ≤ δ ≤ +35°.

[0117] Tug thrust constraint: For each tug i, 0 ≤ T_i ≤ T_i_max (rated thrust).

[0118] Optimization must ensure the safety of the ship in physical space, including:

[0119] Berth angle constraint: The angle α between the bow of the vessel and the quay line is ≤10° to ensure a smooth approach.

[0120] Safety distance constraint: The distance D between the vessel and the adjacent berth or obstacle is ≥ 1.2 times the vessel's width.

[0121] The constrained optimization problem described above is typically transformed into a nonlinear programming problem. The system employs an efficient numerical solver (e.g., interior-point method or sequential quadratic programming algorithm) for online solution. The variables to be solved are a series of control quantities (main engine speed, rudder angle, and thrust of each tugboat) over a future period.

[0122] The solver outputs an optimal sequence of control commands, for example:

[0123] [t0 t5 seconds]: The rudder angle is maintained at +3°, the main engine speed is reduced to slow speed (40% of rated speed), the thrust of tugboat A is 25%, and the thrust of tugboat B is 15%.

[0124] [t6 t10 seconds]: The rudder angle returns to zero, the main engine speed drops to extremely slow (20% of rated speed), and the thrust of tugboat A increases to 40%.

[0125] The system sends the first instruction in the sequence (i.e., the instruction at [t0 t5 seconds]) to the multi-terminal collaborative execution module. In the next control cycle (0.5 seconds later), the system receives the latest ship status provided by the situation synchronization and awareness module, and restarts the entire prediction and optimization process based on this new status.

[0126] This embodiment transforms berthing control from a passive, reactive operation into a proactive, forward-looking optimization process through model predictive control algorithms. It fully considers future ship dynamics, environmental disturbances, and safety boundaries at every step, systematically balancing the often conflicting objectives of efficiency, economy, and safety. Ultimately, it generates a mathematically rigorous optimal cooperative control strategy, achieving intelligent and precise automatic control of ship berthing.

[0127] Specifically, the instructions distributed to the tugboats by the multi-terminal collaborative execution module are specific thrust magnitude and azimuth angle instructions calculated by the thrust distribution algorithm for each participating tugboat.

[0128] Specifically, the instructions distributed by the multi-terminal collaborative execution module to the terminal system include status querying and linkage control of the terminal's collision avoidance and mooring facilities; the linkage control instructions are triggered based on the prediction results of the safety decision and early warning module at different berthing distances.

[0129] In this embodiment, the multi-terminal collaborative execution module is deeply integrated with the terminal system, realizing intelligent status monitoring and forward-looking linkage control of the terminal's collision avoidance and mooring facilities.

[0130] Hardware Integration Fundamentals:

[0131] Rubber fender system: Each fender unit has built-in pressure and deformation sensors and is connected to the terminal PLC via an industrial bus (such as PROFIBUS-DP).

[0132] Mooring winch system: Each winch is equipped with a high-precision encoder (to measure the length of cable winding and unwinding), a tension sensor, and a drive motor status monitoring module.

[0133] Central communication gateway: The dock PLC exchanges high-speed data with the server of the multi-terminal collaborative execution module via industrial Ethernet.

[0134] The multi-terminal collaborative execution module broadcasts a status query command to the dock PLC once per second. After receiving the command, the dock PLC summarizes the status data of each device and returns a status data packet containing the following information within 100 milliseconds:

[0135] Real-time pressure values ​​(unit: kN) and compression (unit: %) of each fender unit; real-time cable tension (unit: tons), cable length (unit: meters), and motor operating status (running / stopping / faulting) of each mooring winch; health status indicators of all sensors.

[0136] Based on the predicted ship trajectory and predicted distance to shore provided by the safety decision and early warning module, the system triggers corresponding linkage control strategies at different stages:

[0137] Phase 1: Early Warning Preparation Phase (Predicted distance 50 meters from the dock)

[0138] Triggering condition: The digital twin module predicts that the ship will enter within 50 meters of the dock within 30 seconds.

[0139] Send the CMD_PRE_STRESS command to the fender system to increase the pressure monitoring sampling frequency of all smart fenders from 1Hz to 10Hz, entering high-precision monitoring mode. Send the CMD_PREPARE command to the mooring system to start the winch motors of all designated berths, putting them in standby tensioned state while maintaining the brakes.

[0140] Phase Two: Pre-tensioning Phase (Predicted distance 10 meters from the dock)

[0141] Triggering condition: It is predicted that a vessel will enter within 10 meters of the dock within 10 seconds.

[0142] The CMD_PRE_TENSION command is sent to the mooring system to control the winch to execute an intelligent pre-tensioning program: based on the ship's tonnage and current wind and current conditions, 30%-50% of the rated tension is calculated and applied as pre-tension. The winch drum pre-releases 2-3 meters of cable based on the predicted ship approach speed, ensuring sufficient safety margin for personnel handling the cable. Tension control employs a PID algorithm to ensure smooth application.

[0143] Phase 3: Berthing and docking phase (estimated distance from the pier: 2 meters)

[0144] Trigger condition: The ship is predicted to make contact with the fender within 5 seconds.

[0145] Send the CMD_IMPACT_READY command to the fender system to activate the collision energy absorption monitoring mode. Continuously monitor the fender pressure gradient; if the pressure change rate exceeds the safety threshold, immediately send an early warning to the safety decision module.

[0146] Mooring stabilization phase (ship fully berthed)

[0147] Triggering conditions: The ship's actual position is stable, and all fender pressures tend to stabilize.

[0148] Send the CMD_FULL_TENSION command to the mooring system to execute a progressive tensioning procedure: within 60 seconds, gradually increase the cable tension from pre-tension to 80% of the rated tension according to a preset "S"-shaped tension curve. Monitor the vessel's position in real time for fine-tuning to ensure even tension distribution. Send the CMD_STABLE_MONITOR command to the fender system to restore the normal 1Hz monitoring frequency, while maintaining the pressure over-limit alarm function.

[0149] Security protection mechanism

[0150] Command verification: Before executing any control command, the dock PLC must verify the validity of the command's digital signature and timestamp.

[0151] Timeout rollback: If communication is interrupted for more than 3 seconds during instruction execution, the system will automatically roll back to the previous safe state.

[0152] Emergency stop priority: An emergency stop instruction (ESTOP) from any party has the highest priority and immediately terminates all control sequences.

[0153] This embodiment achieves the following by deeply integrating the terminal facilities into the automatic berthing control closed loop:

[0154] Active protection: Based on predicted distances, facility status is configured in advance, transforming passive acceptance into proactive adaptation.

[0155] Process optimization: Intelligent pre-tensioning avoids the risks associated with sudden and forceful tensioning in traditional mooring.

[0156] System synergy: The ship's power, tugboat thrust, and terminal facilities form a unified and controlled system, significantly improving berthing safety and efficiency.

[0157] Specifically, the multi-terminal collaborative execution module uses a communication protocol and link with redundancy design for instruction transmission. The redundancy design includes at least one wired communication link and one wireless communication link. The transmitted instructions include a sequence number and a check code for verifying integrity.

[0158] Specifically, the system constructs a learning and optimization unit based on historical berthing data. The unit uses machine learning algorithms to self-correct and optimize specific parameters in the ship's six-degree-of-freedom motion model and the boundary thresholds of the safe berthing corridor.

[0159] In this embodiment, the core function of the learning and optimization unit is to achieve continuous self-evolution of system performance. Machine learning algorithms are used to automatically correct and optimize core model parameters and safety thresholds.

[0160] During each berthing operation, the system automatically records a complete berthing data packet, which includes:

[0161] Environmental data: time series of wind speed, wind direction, current speed, and current direction.

[0162] Ship status data: The ship's actual position, speed, and attitude (roll, pitch, bow) sequence from the situational awareness module.

[0163] Control command data: All control command sequences sent to the host, servo motor, and tugboat.

[0164] Intermediate model data: A copy of the parameters of the six-degree-of-freedom motion model of the ship used in the simulation by the digital twin module and its prediction results.

[0165] Final results data: actual berthing position after berthing, deviation from the plan, maximum impact force, etc.

[0166] All time-series data are aligned under a unified time base, and filtering algorithms are used to remove obvious outliers and noise. Key features are extracted from the raw data; for example, environmental data are synthesized into equivalent wind pressure and flow pressure acting on the ship's hull; the ship's average speed and acceleration at different stages are calculated. Labels are generated for supervised learning. For example, the deviation between the model's predicted trajectory and the actual trajectory, and the deviation between the predicted attitude and the actual attitude, are used as labels for optimizing the motion model; the actual safe corridor boundary used in a high-scoring berthing without warning or emergency braking is used as a reference label for optimizing the boundary threshold.

[0167] This unit performs periodic or trigger-based optimization on the following two types of core parameters:

[0168] A. Parameter optimization of a ship's six-degree-of-freedom motion model:

[0169] Hydrodynamic derivatives in ship motion models, especially nonlinear damping coefficients and added mass coefficients, are difficult to obtain accurately through experiments. Long Short-Term Memory (LSTM) networks or gated recurrent units (GRUs) are employed because they can effectively process time-series data.

[0170] The system uses the "control command sequence" and "environmental data sequence" from a berthing data packet as input features, and the "actual ship motion state sequence" as the target output. This dataset is then fed into an LSTM network for training. The network learns a residual model that aims to predict the trajectory and attitude errors generated during simulations based on the current model parameters. The trained LSTM network can predict the errors of the current model parameters based on real-time input. The system then fine-tunes specific hydrodynamic coefficients in the original motion model based on this error, using backpropagation or gradient descent. For example, if the model consistently shows a delay in predicting large rudder angles, the corresponding nonlinear bow damping coefficient is automatically increased. The corrected parameters are then sent to a digital twin module for offline simulation verification. If, in the simulation with the new parameters, the average prediction error of the past 50 berthings falls within a set range, the system automatically replaces the old parameters with the new ones, completing the model update.

[0171] B. Optimization of safe berthing corridor boundary thresholds:

[0172] The width of the safe berthing corridor and the impact force thresholds for each warning level are determined. Ensemble learning algorithms such as Random Forest or XGBoost are employed.

[0173] The system correlates the "environmental data (features)" of each berthing with the "final safety score (label) of that berthing." The safety score is calculated by comprehensively considering multiple indicators such as berthing efficiency, smoothness, and maximum impact force. The random forest model can output the degree of influence of different environmental features (such as wind speed, current speed, and ship tonnage) on the safety score. Based on the analysis results, the system dynamically adjusts the safety corridor boundary. For example, the model finds that for container ships of a specific tonnage, when the wind speed exceeds 10 m / s, the actual tracks of all high-scoring berthings in historical data are 15% wider than the original corridor. The system will then automatically generate a new rule: when a similar ship is identified and the wind speed is >10 m / s, the baseline value of the corridor width will automatically increase by 15%. Each successful berthing experience (such as a berthing that was completed safely under stronger winds and currents) is added to the dataset as a new positive sample, making the setting of the safety boundary increasingly precise and adaptive, gradually evolving from a conservative fixed value to a dynamic intelligent boundary based on actual capabilities.

[0174] This embodiment, by introducing data-driven and machine learning mechanisms, transforms the automated berthing system from a static system dependent on the initial design into an adaptive intelligent system with continuous learning capabilities. It can learn from each actual operation, continuously revise its internal model to better reflect the real physical world, and dynamically adjust its safety strategies to balance efficiency and risk. This achieves a long-term evolutionary effect, becoming more accurate and safer with use, thus solving the performance bottleneck problem of traditional control systems caused by model inaccuracies and rigid strategies.

[0175] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A ship automatic berthing safety control system, characterized in that, It includes four core modules that are connected in sequence and form a closed-loop control: The situational awareness module is used to build a ship-shore integrated collaborative awareness network. By synchronizing and fusing the ship-end inertial measurement unit and the dock-end inertial navigation unit under a unified time base, it integrates real-time collected ship position, speed, direction of travel and wind / current disturbance data to generate dynamic attitude information of all ship elements with a unified spatiotemporal reference. The digital twin simulation module is connected to the situational awareness module and receives the dynamic attitude information of all elements. It has a built-in six-degree-of-freedom motion model of the ship and a berth environment model. It takes real-time data and tugboat thrust as inputs for calculation and performs real-time simulation of the ship berthing process in virtual space. It also uses a built-in prediction algorithm to make forward-looking predictions of the ship's future trajectory, motion attitude and interaction forces with the berth. The safety decision and early warning module is connected to the digital twin simulation module and embeds a trajectory-berth force coupling prediction algorithm based on spatiotemporal synchronization. By comparing the ship's motion trajectory predicted by the digital twin simulation module with the preset safe berthing corridor, the module assesses the ship's impact force on the berth, dynamically determines the safety status of the berthing process, and outputs a collaborative control strategy based on the prediction results. The multi-terminal collaborative execution module, connected to the safety decision and early warning module, converts the collaborative control strategy into executable instructions and synchronously distributes them to the corresponding tugboats, ship bridges, and dock systems.

2. The system according to claim 1, characterized in that, In the situational awareness synchronization module, the synchronization of the unified time base is achieved through network time protocols and timing satellite signals, specifically including: The network time protocol includes deploying time servers at both the ship and dock ends, exchanging time synchronization messages through redundant network links including redundant Ethernet and 5G industrial private network, and compensating for message transmission delays. The timing satellite signal is received by a dual-mode timing receiver configured at both the ship and the dock, and multi-satellite joint calculation is used to eliminate signal interference. After synchronization is completed, the spatiotemporal labels of all sensor data at both the ship and the dock are based on the unified time base and associated with the dock reference point.

3. The system according to claim 1, characterized in that, The situational awareness module uses Kalman filtering to fuse and complement the dynamic data from the ship-side inertial measurement unit and the static reference data from the dock-side inertial navigation unit. The data fusion process includes: Using static reference data at the dock as a reference, dynamic data at the ship end is predicted; the measured values ​​of the dynamic data are compared with the predicted values, and the predicted values ​​are corrected by the Kalman gain matrix to compensate for sensor errors; the fused data is smoothed to generate the full-element dynamic attitude information.

4. The system according to claim 1, characterized in that, In the digital twin simulation module, the ship's six-degree-of-freedom motion model is a parameterized mathematical model, which is further coupled with the ship's hydrodynamic coefficients obtained from ship design data and mooring tests, as well as the ship's load parameters acquired in real time by the draft sensors deployed on the ship. The ship's hydrodynamic coefficients are stored and retrieved in segments according to the ship's speed range, and the ship's load parameters are used to correct the buoyancy center position and inertia matrix in the model in real time.

5. The system according to claim 1, characterized in that, The time window for the digital twin simulation module to predict the future trajectory and attitude of the ship is configurable and can be manually or automatically adjusted according to the current berthing stage, ship handling performance, or external environmental conditions.

6. The system according to claim 1, characterized in that, In the safety decision-making and early warning module, the boundary of the safe berthing corridor is dynamically adjusted by a calculation model based on real-time input environmental data, ship tonnage, and windward area; the environmental data includes wind speed and direction, and current speed and direction.

7. The system according to claim 6, characterized in that, The safety decision and early warning module is preset with at least two levels of safety thresholds, including a warning level and an alarm level. When the predicted deviation from the flight path or the impact force reaches the warning level threshold, the system issues a warning to the operator. When the predicted value exceeds the alarm level threshold, the system automatically triggers the collaborative control strategy and hands it over to the multi-terminal collaborative execution module for execution. At the same time, the warning status is upgraded, including upgrading the visual warning from yellow flashing to red flashing and the audible warning from intermittent beeping to continuous beeping.

8. The system according to claim 1, characterized in that, The optimal cooperative control strategy is generated based on the model predictive control algorithm. Under the premise of satisfying the ship dynamics constraints and the berth space geometric constraints, a sequence of commands with the optimal objective function is obtained through mathematical programming. The objective function is used to comprehensively optimize berthing time, energy consumption and safety risks.

9. The system according to claim 1, characterized in that, The instructions distributed to the tugboats by the multi-terminal collaborative execution module are specific thrust magnitude and azimuth angle instructions calculated by the thrust distribution algorithm for each participating tugboat.

10. The system according to claim 1, characterized in that, The instructions distributed by the multi-terminal collaborative execution module to the terminal system include status query and linkage control of the terminal's collision avoidance and mooring facilities; the linkage control instructions are triggered at different berthing distances based on the prediction results of the safety decision and early warning module.

11. The system according to claim 1, characterized in that, The multi-terminal collaborative execution module uses a communication protocol and link with redundancy design for instruction transmission. The redundancy design includes at least one wired communication link and one wireless communication link. The transmitted instructions include a sequence number and a check code for verifying integrity.

12. The system according to claim 1, characterized in that, The system includes a learning and optimization unit based on historical berthing data. This unit uses machine learning algorithms to self-correct and optimize specific parameters in the ship's six-degree-of-freedom motion model and the boundary thresholds of the safe berthing corridor.

Images