CN-121981550-A - Ship entry and exit intelligent organization and risk deduction system based on big data service

CN121981550ACN 121981550 ACN121981550 ACN 121981550ACN-121981550-A

Abstract

The invention discloses a ship arrival and departure intelligent organization and risk deduction system based on big data service, which relates to the field of ship arrival and departure intelligent organization and risk deduction, and comprises a data acquisition module, a data processing module and a data processing module, wherein the data acquisition module is used for constructing a dynamic digital twin scene of a harbor water area; the system comprises a rule making module, a map construction module, an instruction output module and a model verification module, wherein the rule making module is used for making ship berthing rules from three aspects of resource space-time conflict, traffic streamline conflict and cargo safety attribute conflict, the map construction module is used for constructing a ship arrival/departure intelligent organization and risk deduction knowledge map, the instruction output module is used for outputting berthing instructions of a new berthing ship based on a graph neural network model, and the model verification module is used for outputting berthing instructions of the berthing ship under historical data through the graph neural network model to verify the feasibility of the model. The method has the advantages that the intelligent optimization conversion from the manual experience-dependent data and rule dual-drive of the ship arrival and departure scheduling is realized, and the port navigation efficiency and the operation toughness are effectively improved.

Inventors

BI XUEJUN
XIA QIHANG
SHENG YONGXIN
ZHANG JIANG
WU HAN
YIN XIANMING
JIN YE
LUO YUWEI

Assignees

南京汇海交通科技有限公司
中华人民共和国天津海事局

Dates

Publication Date: 20260505
Application Date: 20260126

Claims (6)

1. The utility model provides a ship business turn over port intelligence organization and risk deduction system based on big data service which characterized in that includes: The data acquisition module is used for receiving and fusing the structured port leaning information chain and the real-time dynamic environment data stream to construct a dynamic digital twin scene of the port water area; the rule making module is used for making ship berthing rules from three aspects of resource space-time conflict, traffic streamline conflict and cargo safety attribute conflict based on expert experience according to different ship basic attribute information; The map construction module is used for constructing a ship arrival intelligent organization and risk deduction knowledge map based on a dynamic digital twin scene of a harbor water area and a ship berthing rule; the instruction output module is used for deducing a knowledge graph according to the intelligent organization and risk of the ship entering and exiting ports and outputting a berthing instruction of a new port-leaning ship based on a graph neural network model; The model verification module is used for outputting berthing instructions of the port-leaning ship under the historical data through the graphic neural network model based on the historical data of the port-approaching ship, and verifying feasibility of the model.
2. The intelligent organization and risk deduction system for ship departure based on big data service according to claim 1, wherein the data acquisition module specifically comprises: The port leaning information unit is used for receiving port leaning information chains of the ships in multiple channels, and the port leaning information chains comprise ship basic attribute information, port leaning operation information and time information; The ship basic attribute information at least comprises a ship type, a draft and a cargo type, the port-leaning operation information at least comprises port-leaning, off-berthing, passing and anchoring berthing, and the time information at least comprises a ship predicted and actual stay time period and a predicted port-entering and exiting time; The environment data unit is used for collecting meteorological data and hydrological data in the field of the port, the meteorological data at least comprise wind speed, wind direction, visibility and precipitation, and the hydrological data at least comprise tide level, water flow speed and water depth; The port resource state unit is used for collecting the occupation condition and the expected accommodation amount of the port and at least comprises a berth occupation condition, berth water depth, a channel traffic state and a loading and unloading equipment state; The unknown ship attribute unit is used for recording and collecting data through a manual verification form when the ship port information chain received by the port information unit is incomplete; The data fusion unit is used for carrying out structural storage on multisource data output by the port information unit, the environment data unit, the port resource state unit and the unknown ship attribute unit, and constructing a dynamic digital twin scene of a port water area.
3. The intelligent organization and risk deduction system for ship arrival and departure based on big data service according to claim 2, wherein the formulating the ship berthing rule from three aspects of resource space-time conflict, traffic flow line conflict and cargo safety attribute conflict based on expert experience according to different ship basic attribute information specifically comprises: According to historical ship berthing data, organizing expert evaluation groups, and evaluating from three aspects of resource space-time conflict, traffic streamline conflict and cargo safety attribute conflict; The resource space-time conflict refers to that different ship plans occupy the same berth and anchor ground at the same moment, or space path overlapping is formed in a limited water area of a channel and a harbor pool; The traffic streamline conflict refers to a meeting point for predicting ship tracks in a future period based on ship AIS dynamic data and a planned route and detecting collision, overtaking or meeting risks among the tracks; the cargo safety attribute conflict refers to the potential conflict between ships carrying incompatible dangerous cargos and the safety interval conflict between dangerous cargo ships and sensitive targets in surrounding open fire operation and densely populated areas according to ship basic attribute information; And formulating a ship berthing rule according to the grading result of the expert group.
4. The intelligent organization and risk deduction system for ship departure based on big data service according to claim 3, wherein the constructing the knowledge graph of intelligent organization and risk deduction for ship departure based on the dynamic digital twin scene and the ship berthing rules in the harbor waters specifically comprises: abstracting a physical entity and a logical entity in the dynamic digital twin scene into map nodes; defining static association and dynamic attribute between entities to form map edge and node attribute; according to the ship berthing rule, converting the ship berthing rule into a logic rule and/or constraint condition which can be automatically inferred in the knowledge graph; according to the map nodes, map edges and node attributes, combining logic rules and/or constraint conditions, constructing a ship arrival/departure intelligent organization and risk deduction knowledge map; and updating the knowledge graph according to the harbor information chain of the new harbor ship and the harbor information of the ship.
5. The intelligent organization and risk deduction system for ship entering and exiting port based on big data service according to claim 4, wherein outputting the berthing instruction of the ship to be newly leaned against port based on the graph neural network model specifically comprises: based on the knowledge graph, generating a structured candidate instruction set comprising berth allocation, channel routes, time windows and recommended speeds for the new port-leaning ship; Encoding nodes, edges and node attributes of the knowledge graph into an input data structure of a graph neural network model, and preprocessing, wherein the input data structure comprises a node characteristic matrix and an adjacent matrix; Inputting the preprocessed knowledge graph data into a pre-trained graph neural network model, wherein the model aggregates information among entities through a graph attention mechanism and outputs a comprehensive evaluation vector about a new port-leaning ship; Defining a core optimization target for generating a berthing instruction, wherein the target at least comprises maximizing port total throughput efficiency, minimizing global risk conflict value and maximizing resource utilization rate; Feature fusion is carried out on the comprehensive evaluation vector and each candidate instruction, and the comprehensive utility score of each candidate instruction under the optimization target is calculated through a lightweight grading network; carrying out fast forward deduction on Top-K candidate instructions with highest scores in a digital twin scene, calling logic rules in a knowledge graph to carry out conflict rechecking, and screening out any instruction causing rule violation; selecting a candidate instruction with highest comprehensive utility score from candidate instructions passing conflict verification, and outputting the candidate instruction as a final executable parking instruction; And if the generated candidate instruction set is empty or all the candidate instructions cannot pass the conflict verification, triggering the manual verification command flow.
6. The intelligent organization and risk deduction system for ship entering and exiting ports based on big data service according to claim 5, wherein the step of outputting the berthing instruction of the port approaching ship under the historical data through the graphic neural network model based on the historical data of ship entering and exiting ports, and the step of verifying the feasibility of the model specifically comprises the following steps: acquiring historical data of a ship entering and exiting port and preprocessing the historical data to form a standardized training data set and a standardized verification data set; taking the port approaching ship berthing instruction actually executed by the history as a comparison group, and marking the instruction with excellent performance as the history fact optimal instruction based on the post effect evaluation; Reconstructing a knowledge graph state at the historical moment according to the training data set, inputting the reconstructed knowledge graph into a pre-trained graph neural network model, and outputting a port leaning ship berthing instruction corresponding to the historical moment as a prediction group; According to the control group and the prediction group, respectively calculating the instruction compliance rate, the conflict resolution rate and the instruction matching accuracy rate as model feasibility quantitative indexes; Screening rule violation, security risk omission and emergency instruction adaptation of a prediction group based on ship berthing rules by adopting a random selection or full detection mode and a manual verification mode; Respectively calculating rule violation rate, security risk omission factor and emergency instruction adaptation rate of the prediction group as artificial quality control indexes of model feasibility; And combining the quantitative index with an artificial quality control result to form a comprehensive evaluation report of the feasibility of the model.

Description

Ship entry and exit intelligent organization and risk deduction system based on big data service Technical Field The invention relates to the field of ship arrival and departure intelligent organization and risk deduction, in particular to a ship arrival and departure intelligent organization and risk deduction system based on big data service. Background With the continuous growth of global trade and the development of large and high-speed ships, ports are used as core nodes of maritime hubs, and the operation efficiency and the safety level face unprecedented pressure. The traditional ship arrival and departure organization highly depends on personal experiences of a ship traffic management system (VTS) attendant and a port dispatcher, and is manually coordinated through a very high frequency wireless telephone (VHF), a ship Automatic Identification System (AIS) and a paper or simple electronic schedule, so that the mode can cope with conventional and low-frequency ship flow fashion, but is free from inherent limitations when faced with complex working conditions such as severe weather, sudden mechanical faults, and dangerous ship concentration to the port. In one aspect, the port water data exhibits multi-source heterogeneous characteristics including vessel dynamic trajectory from AIS, real-time environmental data from the weather hydrology department, vessel planning and cargo information submitted by agent companies, and port facility status data. These data are scattered in different systems, lack of effective fusion, and are difficult to form unified real-time situation awareness, so that "information blind areas" exist in scheduling decisions. On the other hand, port scheduling is essentially a complex resource optimization problem under strong space-time constraint and safety rules, and relates to dynamic allocation of scarce resources such as berths, channels, anchor lands and the like, and environmental boundary conditions such as physical properties (such as captain and draft), cargo risk levels, traffic streamline conflicts, tides, visibility and the like of ships need to be comprehensively considered. The problem of optimizing the high dimensionality and the multiple targets is manually handled, the calculation load is huge, the generation of the global approximate optimal solution in a short time is difficult, the systematic deduction and early warning of the potential risks cannot be carried out, and the problem becomes a key bottleneck for limiting the throughput capacity and the safety upper limit of the port. The prior art scheme is mainly divided into two types, namely an automatic scheduling system based on fixed rules and simple priority ordering, wherein the system usually solidifies part of scheduling rules (such as 'large ship priority', 'first come first serve') in a program, can realize limited automatic scheduling, but has the defects of stiff rules and lack of intelligence and adaptability, and the second type is a method which utilizes a traditional operation study optimization model (such as mixed integer programming), can formally describe scheduling problems, has high computational complexity and poor instantaneity, is difficult to integrate into complex domain knowledge and dynamic rules, is difficult to modify once the model is established, and cannot flexibly incorporate new regulations issued temporarily by maritime departments or operation strategy adjustment of ports. Both types of prior art have the defect of capability of carrying out 'hypothesis analysis' and risk deduction in a digital space, so that the fault tolerance of a decision scheme in actual execution is low, and the spanning from passive response to active early warning and prospective optimization cannot be realized. Disclosure of Invention In order to solve the technical problems, the technical scheme provides a ship arrival/departure intelligent organization and risk deduction system based on big data service, and solves the crossing problem that the decision scheme has low fault tolerance in actual execution and cannot realize the crossing from passive response to active early warning and prospective optimization due to the lack of the capability of carrying out 'hypothesis analysis' and risk deduction in a digital space in the background technology. In order to achieve the above purpose, the invention adopts the following technical scheme: a ship entry and exit intelligent organization and risk deduction system based on big data service comprises: The data acquisition module is used for receiving and fusing the structured port leaning information chain and the real-time dynamic environment data stream to construct a dynamic digital twin scene of the port water area; the rule making module is used for making ship berthing rules from three aspects of resource space-time conflict, traffic streamline conflict and cargo safety attribute conflict based on expert experience according to different ship basi