A method and system for ship navigation optimization control based on artificial intelligence

Abstract

This invention relates to the field of ship navigation control technology, and discloses a ship navigation optimization control method and system based on artificial intelligence. The method includes: acquiring physical observation information of surrounding target vessels from the ship's radar and AIS broadcast information; comparing the physical observation information and the AIS broadcast information to identify whether there is persistent inconsistency; when persistent inconsistency is identified, calculating the fluctuation characteristics of the AIS broadcast information corresponding to the persistent inconsistency; determining the source type of the persistent inconsistency based on the fluctuation characteristics and physical observation information; assigning risk weights to the persistent inconsistency based on the determined source type; and generating navigation avoidance instructions for the ship based on the risk weights. This invention achieves more accurate and safer collision avoidance decisions by intelligently identifying and evaluating the nature of inconsistencies between AIS broadcast information and radar observation data.

Description

Technical Field

[0001] This invention relates to the field of ship navigation control technology, and more specifically, to a ship navigation optimization control method and system based on artificial intelligence. Background Technology

[0002] In modern shipping, autonomous navigation systems (AIS) on ships integrate multiple information sources to enhance navigation safety and efficiency. However, when the equipment of external vessels ages, causing slight and persistent deviations in the Automatic Identification System (AIS) information they broadcast, these systems face challenges in judging obstacle behavior and planning routes. This deviation may not be an actual dangerous maneuver, but rather a problem with the equipment itself. However, if the system cannot accurately identify it, it may lead to overly conservative avoidance, thus impacting navigation efficiency and fuel consumption.

[0003] However, when the AI ​​system receives atypical, persistent minor data inconsistencies caused by aging external ship equipment (e.g., aging GPS receivers causing slight and intermittent drift in Automatic Identification System (AIS) position information, thus affecting broadcast speed), the system detects a persistent, minor inconsistency between the target vessel's reported speed and its actual navigation status observed by radar. Faced with this uncertain obstacle behavior that does not conform to known patterns, the AI ​​system's built-in risk assessment mechanism initiates a more conservative judgment. Since the system cannot accurately attribute this data inconsistency to known physical causes, nor can it precisely predict the impact of this "abnormal" speed fluctuation on the target vessel's future trajectory, it treats this uncertainty as an increased potential risk, thus raising the potential collision risk level of the target vessel. To avoid this overestimated "uncertain" obstacle, the global route planning search process immediately begins looking for a more lenient avoidance path, causing the vessel to deviate from its theoretically optimal route and significantly increasing fuel consumption through unnecessary avoidance maneuvers, thereby reducing overall navigation efficiency. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention discloses a ship navigation optimization control method and system based on artificial intelligence. The aim is to solve the problem that in existing technologies, when the aging of external ship equipment causes continuous minor deviations in AIS information, the autonomous navigation system cannot accurately identify the source of data inconsistency, resulting in overly conservative avoidance, which affects navigation efficiency and fuel consumption.

[0005] The technical solution of the present invention is as follows: In a first aspect, the present invention discloses a ship navigation optimization control method based on artificial intelligence, comprising the following steps: To acquire physical observation information of surrounding target vessels by the ship's radar, as well as AIS broadcast information of the target vessels; Compare physical observation information with AIS broadcast information to identify whether there is persistent inconsistency between the physical observation information and AIS broadcast information. When persistent inconsistency is identified, calculate the fluctuation characteristics of the AIS broadcast information corresponding to the persistent inconsistency. Based on wave characteristics and physical observation information, determine the source type of persistent inconsistency information; Based on the identified source type, risk weights are assigned to persistently inconsistent information; Based on risk weights, navigation avoidance instructions are generated for this vessel.

[0006] Through this technical solution, the present invention can effectively identify and analyze the persistent inconsistency between ship AIS information and radar observation information, and assign different risk weights according to their source types, thereby generating more accurate navigation avoidance instructions, avoiding excessive avoidance caused by information uncertainty, and improving navigation efficiency and safety.

[0007] Furthermore, the physical observation information includes the target vessel's initial position, initial speed, and initial course.

[0008] Furthermore, AIS broadcast information includes the target vessel's reported second position, second speed, second course, and vessel identification code.

[0009] Based on the above, the present invention further proposes that the fluctuation characteristics include at least one of fluctuation frequency, fluctuation amplitude, and fluctuation duration, and the steps for calculating the fluctuation characteristics of the AIS broadcast information corresponding to the persistence inconsistency information include: Calculate at least one of the following based on the AIS broadcast information corresponding to persistent inconsistency information: The fluctuation frequency is obtained by calculating the number of times the target vessel reports changes in its second position or second speed within a preset unit time. Calculate the standard deviation of the second position or second speed from its average value to obtain the fluctuation range; The duration of the fluctuation is obtained by calculating the length of time during which persistent inconsistencies occur.

[0010] Furthermore, based on wave characteristics and physical observation information, the steps for determining the source type of persistent inconsistency information include: Match the fluctuation characteristics with a preset set of equipment aging characteristic patterns; If the fluctuation characteristics match the patterns in the preset equipment aging characteristic pattern set more than the first preset threshold, and the physical observation information shows that the changes in the target ship's course and speed are less than the second preset threshold, then the source of the persistent inconsistency information is determined to be benign data noise caused by equipment aging.

[0011] Based on this, the present invention further proposes that the step of determining the source type of persistent inconsistency information based on wave characteristics and physical observation information also includes: If the matching degree between the fluctuation characteristics and the patterns in the preset equipment aging characteristic pattern set is lower than the first preset threshold, and the clear movement intention of the target vessel cannot be confirmed based on physical observation information, then the source type of the persistent inconsistency information is determined to be an unknown risk.

[0012] More specifically, the steps for determining the source type of persistent inconsistency information based on wave characteristics and physical observation information also include: If physical observation information shows that the target vessel's course or speed changes by more than a second preset threshold within a preset time, and the trend of fluctuation characteristics matches the trend of physical observation information by a third preset threshold, then the source of the persistent inconsistency information is determined to be the actual maneuvering operation of the target vessel.

[0013] Specifically, the steps for assigning risk weights to persistently inconsistent information based on the identified source type include: If the source type is benign data noise caused by equipment aging, then the first risk weight is assigned to persistent inconsistency information; If the source type is uncertainty of unknown origin, then a second risk weight is assigned to persistently inconsistent information, where the second risk weight is greater than the first risk weight; If the source type is the actual maneuvering operation of the target vessel, a third risk weight is assigned to the persistent inconsistency information, where the third risk weight is greater than the second risk weight.

[0014] Based on the above, the present invention further proposes that the steps for generating a navigation avoidance command for the vessel based on risk weights include: The collision risk threshold between this vessel and the target vessel is calculated based on risk weights. If the collision risk threshold does not reach the preset collision threshold, a fine-tuning avoidance command is generated. The fine-tuning avoidance command is used to control the ship to gradually increase the safe distance from the target ship by making one or more fine adjustments to its course or speed. If the collision risk threshold reaches the preset collision threshold, a hazard avoidance command is generated in accordance with the International Maritime Collision Avoidance Regulations. The hazard avoidance command is used to control the vessel's turning or deceleration.

[0015] Secondly, the present invention also discloses an artificial intelligence-based ship navigation optimization control system, comprising: The information acquisition module is used to acquire physical observation information of surrounding target vessels by the ship's radar, as well as AIS broadcast information of the target vessels; The feature calculation module is used to compare physical observation information with AIS broadcast information to identify whether there is persistent inconsistency between the physical observation information and AIS broadcast information. When persistent inconsistency is identified, the fluctuation characteristics of the AIS broadcast information corresponding to the persistent inconsistency are calculated. The source type determination module is used to determine the source type of persistent inconsistency information based on wave characteristics and physical observation information; The risk weight allocation module is used to assign risk weights to persistent inconsistency information based on the determined source type. The instruction generation module is used to generate navigation and avoidance instructions for the ship based on risk weights.

[0016] Through this technical solution, the present invention provides a system for implementing the above-mentioned method. Through modular design, the system functions are clear, easy to implement and maintain, and provides hardware and software support for intelligent navigation of ships.

[0017] In summary, this invention provides an artificial intelligence-based ship navigation optimization control method and system. The method acquires physical observation information of surrounding target vessels from the ship's radar and AIS broadcast information from the target vessels, comparing the two to identify persistent inconsistencies. When persistent inconsistencies are identified, the fluctuation characteristics of the corresponding AIS broadcast information are calculated. Subsequently, based on the fluctuation characteristics and physical observation information, the source type of the persistent inconsistency is determined, and a risk weight is assigned according to the determination result. Finally, a navigation avoidance command for the ship is generated based on the risk weight. This method effectively solves the problem in existing technologies where, when aging equipment on external vessels causes persistent minor deviations in AIS information, the autonomous navigation system cannot accurately identify the source of data inconsistency, leading to overly conservative avoidance, affecting navigation efficiency and fuel consumption. By performing refined analysis and risk assessment of inconsistencies, this invention can distinguish between benign data noise and actual danger, avoiding unnecessary avoidance operations, significantly improving navigation efficiency, reducing fuel consumption, and ensuring navigation safety. Attached Figure Description

[0018] Figure 1 This is a flowchart illustrating an artificial intelligence-based ship navigation optimization control method provided in an embodiment of the present invention.

[0019] Figure 2This is a schematic diagram of a ship navigation optimization control system based on artificial intelligence, provided as an embodiment of the present invention.

[0020] Labeling Explanation: 210, Information Acquisition Module; 220, Feature Calculation Module; 230, Source Type Judgment Module; 240, Risk Weight Allocation Module; 250, Instruction Generation Module. Detailed Implementation

[0021] The technical solutions of this invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are merely some, not all, of the embodiments of this invention. The components of this invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the invention provided in the drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without inventive effort are within the scope of protection of this invention.

[0022] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this invention, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0023] Traditional autonomous navigation systems often face challenges in handling persistent, minor deviations in external Automatic Identification System (AIS) information. These deviations may stem from equipment aging rather than actual dangerous maneuvers, but if the system fails to accurately identify them, it may lead to overly conservative avoidance strategies, impacting navigation efficiency and fuel consumption. For example, when a target vessel's GPS receiver aging causes minor and intermittent drifts in its AIS position information, affecting broadcast speed, existing systems may misinterpret this as an uncertain risk, triggering unnecessary avoidance maneuvers, causing the vessel to deviate from its optimal course and increasing fuel consumption.

[0024] Firstly, please see Figure 1 This invention proposes an artificial intelligence-based ship navigation optimization control method, comprising: S1. Obtain physical observation information of surrounding target vessels by the ship's radar and AIS broadcast information of the target vessels; S2. Compare physical observation information with AIS broadcast information to identify whether there is persistent inconsistency between physical observation information and AIS broadcast information. When persistent inconsistency is identified, calculate the fluctuation characteristics of the AIS broadcast information corresponding to the persistent inconsistency. S3. Based on wave characteristics and physical observation information, determine the source type of persistent inconsistency information; S4. Assign risk weights to persistently inconsistent information based on the identified source type; S5. Based on risk weights, generate navigation avoidance instructions for this vessel.

[0025] The method proposed in this invention achieves optimized control of ship navigation through a series of steps. First, it is necessary to acquire physical observation information of surrounding target vessels from the ship's radar, as well as AIS broadcast information from the target vessels. Physical observation information can be acquired through the ship's onboard radar system, which periodically scans the surrounding sea area to detect information such as the distance, bearing, and speed of target vessels. For example, the radar can transmit a pulse every few seconds and receive the echo, determining the real-time position, speed, and heading of the target vessel by calculating the time delay and Doppler shift of the echo. AIS broadcast information is acquired through the ship's AIS receiver, which can receive AIS signals periodically broadcast by surrounding vessels and extract data such as the target vessel's reported position, speed, heading, and vessel identification code. For example, the AIS receiver can continuously monitor the VHF band, decoding AIS data packets upon receipt to extract the required information.

[0026] Next, physical observation information is compared with AIS broadcast information to identify any persistent inconsistencies. When persistent inconsistencies are identified, the fluctuation characteristics of the corresponding AIS broadcast information are calculated. This comparison can be achieved by synchronizing and comparing the two sets of information in real time. For example, the position of the target vessel observed by radar can be compared with the position reported by AIS, and the Euclidean distance between them can be calculated. Simultaneously, the speed observed by radar and the speed reported by AIS, as well as the heading observed by radar and the heading reported by AIS, can be compared. If these differences persist for a period of time within a preset threshold range, persistent inconsistencies can be identified. Once persistent inconsistencies are identified, the fluctuation characteristics of the corresponding AIS broadcast information need to be calculated. For example, time series analysis can be performed on the position or speed data reported by AIS to calculate the frequency of change, maximum deviation, and duration of inconsistency within a certain time window. Calculating these fluctuation characteristics helps quantify the dynamic characteristics of the inconsistencies.

[0027] Subsequently, based on the fluctuation characteristics and physical observation information, the source type of the persistent inconsistency is determined. This step is one of the core aspects of this invention, aiming to distinguish between different types of data inconsistencies. For example, a machine learning model can be used, taking the calculated fluctuation characteristics as input and combining them with physical observation information (such as the actual course and speed changes of the target vessel) to train the model to identify different source types. Specifically, if the fluctuation characteristics exhibit small-amplitude, high-frequency random fluctuations, and the physical observation information shows that the target vessel's course and speed remain stable, it may be judged as benign data noise caused by equipment aging. Conversely, if the fluctuation characteristics exhibit trend-like changes, and the physical observation information shows that the target vessel is making significant course or speed adjustments, it may be judged as actual maneuvering operations by the target vessel.

[0028] Furthermore, risk weights are assigned to persistently inconsistent information based on the identified source type. Different source types correspond to different potential risk levels. For example, benign data noise identified as being caused by equipment aging can be assigned a lower risk weight because it has little impact on the ship's navigation safety. Actual maneuvers identified as those of a target vessel can be assigned a medium risk weight because they may require the ship to take appropriate evasive action. Conversely, unknown risks or malicious interference can be assigned a higher risk weight to encourage the ship to adopt a more cautious evasive strategy. The allocation of risk weights can be achieved through a pre-defined rule table or a fuzzy logic system based on expert experience.

[0029] Finally, based on risk weights, the system generates navigation avoidance instructions for the vessel. Risk weights are a key input for generating these instructions. For example, if a low risk weight is assigned, indicating a low risk of collision, the system can generate fine-tuning avoidance instructions, such as gradually increasing the safe distance from the target vessel through small adjustments to course or speed, thereby avoiding unnecessary abrupt collisions and maintaining navigational efficiency. If a high risk weight is assigned, indicating a high risk of collision, the system will generate dangerous avoidance instructions according to international maritime collision avoidance regulations, such as controlling the vessel to make a sharp turn or emergency deceleration to ensure navigational safety.

[0030] The AI-based ship navigation optimization control method proposed in this invention, through the close coordination of the aforementioned technical features, solves the problem of excessive avoidance caused by the inability to accurately identify the source of AIS information deviation in existing technologies. Specifically, this method first acquires radar physical observation information and AIS broadcast information to lay the foundation for subsequent data comparison and analysis. Subsequently, by comparing these two types of information and identifying persistent inconsistencies, potential data anomalies can be detected in a timely manner. Furthermore, by calculating the fluctuation characteristics of AIS broadcast information and combining it with physical observation information to determine the source type of inconsistencies, this method can intelligently distinguish between different situations such as benign data noise caused by equipment aging, actual maneuvering operations, and unknown risks. This refined identification capability is key to solving the pain points of existing technologies. Based on this, different risk weights are assigned according to the determined source type, making risk assessment more accurate and reasonable. Finally, based on these risk weights, corresponding navigation avoidance commands are generated to ensure navigation safety while avoiding unnecessary excessive avoidance, thereby optimizing navigation efficiency and fuel consumption. The entire process forms a closed-loop intelligent control system, enabling ships to cope with complex navigation environments more intelligently and efficiently.

[0031] This invention represents a significant advancement over existing technologies. Traditional systems, when faced with inconsistencies between AIS information and radar observations, often adopt a conservative strategy, treating all inconsistencies as potential risks, leading to frequent and unnecessary avoidance maneuvers. For example, when the target vessel's AIS information experiences minor fluctuations due to equipment aging, existing systems may fail to recognize their benign nature, triggering large-scale avoidance maneuvers, resulting in course deviations and fuel waste. The core innovation of this invention lies in introducing an intelligent judgment mechanism for the type of source of persistent inconsistencies. By analyzing the fluctuation characteristics of AIS broadcast information and combining it with radar physical observation information, this method can distinguish between benign data noise caused by equipment aging, actual maneuvers of the target vessel, and unknown risks. This refined risk identification capability allows the vessel to allocate different risk weights according to the actual risk level and generate more accurate and optimized avoidance commands. For example, for benign data noise caused by equipment aging, this method can assign a lower risk weight, thereby generating fine-tuned avoidance commands to avoid over-avoidance; while for actual maneuvers or unknown risks, a higher risk weight can be assigned, generating more mandatory danger avoidance commands. This differentiated approach significantly enhances the intelligence of navigation decisions, effectively reduces unnecessary avoidance, improves navigation efficiency and economy, and ensures navigation safety.

[0032] Specifically, the aforementioned physical observation information may include the target vessel's first position, first speed, and first heading.

[0033] Physical observation information refers to real-time dynamic data about surrounding target vessels directly acquired through the ship's own radar and other physical detection equipment. Specifically, physical observation information may include the target vessel's first position, representing its geographical coordinates at a given moment; its first speed, representing its speed relative to the water or land surface; and its first heading, representing its direction of motion. This information is derived from the ship's radar system through detection, tracking, and calculation of the target vessels, with the aim of providing objective motion status data independent of the target vessels' own reports.

[0034] The present invention explicitly includes the target vessel's first position, first speed, and first heading in the physical observation information, enabling subsequent comparisons between the physical observation information and AIS broadcast information to be based on specific and quantifiable parameters. It is precisely because of this detailed physical observation data that the system can have a clear and real-time understanding of the target vessel's actual motion state, providing a solid data foundation for identifying persistent inconsistencies.

[0035] Specifically, the aforementioned AIS broadcast information can be understood as the target vessel broadcasting its own information to the surrounding area through the Automatic Identification System (AIS).

[0036] AIS broadcast information includes the target vessel's reported secondary position, secondary speed, secondary course, and vessel identification code. The secondary position refers to the target vessel's geographic coordinates at a given moment, such as longitude and latitude, used to accurately indicate its location on a nautical chart. The secondary speed refers to the target vessel's speed on the water, usually measured in knots, reflecting its dynamic state. The secondary course refers to the target vessel's direction of travel, usually expressed in degrees based on true north, indicating its trend of movement. The vessel identification code, such as the International Maritime Organization (IMO) number or the Maritime Mobility Services Identifier (MMSI), is a combination of numbers or letters used to uniquely identify the target vessel, ensuring accurate identification and tracking.

[0037] The solution of this invention, by clearly defining the specific content contained in the AIS broadcast information, enables the system to acquire more detailed and standardized dynamic and static data of the target vessel. This data, including the second position, second speed, second course, and vessel identification code, provides a rich and structured data foundation for subsequent comparison of physical observation information and AIS broadcast information. Therefore, when identifying persistent inconsistencies, a refined comparison can be performed based on these specific parameters, thereby improving the accuracy and reliability of inconsistency identification.

[0038] Specifically, in the above method, the calculation method of fluctuation characteristics is further refined.

[0039] The aforementioned fluctuation characteristics include at least one of fluctuation frequency, fluctuation amplitude, and fluctuation duration. The steps for calculating the fluctuation characteristics of the AIS broadcast information corresponding to persistent inconsistency information include: Calculate at least one of the following based on the AIS broadcast information corresponding to persistent inconsistency information: The fluctuation frequency is obtained by calculating the number of times the target vessel reports changes in its second position or second speed within a preset unit time. Calculate the standard deviation of the second position or second speed from its average value to obtain the fluctuation range; The duration of the fluctuation is obtained by calculating the length of time during which persistent inconsistencies occur.

[0040] The fluctuation characteristic refers to the dynamic change characteristics exhibited by the AIS broadcast information itself when there is a persistent inconsistency between AIS broadcast information and physical observation information. This fluctuation characteristic can be quantified as at least one of fluctuation frequency, fluctuation amplitude, and fluctuation duration. Specifically, fluctuation frequency refers to the number of times the target vessel's reported second position or second speed changes within a preset unit of time, reflecting the density of data updates or jumps. Fluctuation amplitude refers to the standard deviation of the second position or second speed from its mean, quantifying the severity or dispersion of data fluctuations. Fluctuation duration refers to the length of time from the first identification of the persistent inconsistency information to the current moment, indicating the persistence of the inconsistency.

[0041] The solution of this invention, through multi-dimensional quantitative analysis of AIS broadcast information corresponding to persistent inconsistencies, can reveal the inherent dynamic characteristics of inconsistencies more deeply. By calculating the fluctuation frequency, the rate of change of AIS data can be understood; by calculating the fluctuation amplitude, the degree to which AIS data deviates from normal values ​​can be assessed; and by calculating the fluctuation duration, the duration of the inconsistency can be determined. This refined extraction of fluctuation features enables the system to move beyond simply identifying inconsistencies to understanding their nature and patterns, providing richer and more reliable data support for subsequent determination of the source type of inconsistencies.

[0042] In some of the above implementations, the determination of the source type of persistent inconsistency information is based on fluctuation characteristics and physical observation information. However, the lack of specific judgment criteria and detailed differentiation mechanisms may lead to inaccurate judgment of the source of inconsistency information, especially in distinguishing between benign data noise caused by equipment aging and actual maneuvering operations or unknown risks, thus affecting the accuracy of subsequent risk assessments.

[0043] In response, this invention further proposes a specific method for determining the source type of persistent inconsistency information, the steps of which include: Match the fluctuation characteristics with a preset set of equipment aging characteristic patterns; If the fluctuation characteristics match the patterns in the preset equipment aging characteristic pattern set more than the first preset threshold, and the physical observation information shows that the changes in the target ship's course and speed are less than the second preset threshold, then the source of the persistent inconsistency information is determined to be benign data noise caused by equipment aging.

[0044] Specifically, fluctuation characteristics can be understood as parameters such as fluctuation frequency, fluctuation amplitude, and fluctuation duration calculated by analyzing AIS broadcast information. These parameters quantify the instability of AIS data. The preset set of equipment aging characteristic patterns refers to typical data fluctuation patterns related to aging and failure of shipboard AIS or radar equipment, obtained through analysis of large amounts of historical data, combined with expert experience, or trained using machine learning algorithms. For example, these patterns may include periodic small-amplitude fluctuations, data loss at specific frequencies, or data jumps. The matching process can be implemented using various pattern recognition algorithms, such as those based on correlation analysis, distance metrics (e.g., Euclidean distance), or machine learning classifiers (e.g., support vector machines, neural networks). Pattern matching degree refers to the degree of similarity or conformity between the fluctuation characteristic and any pattern in the pattern set; a higher value indicates a better match. The first preset threshold is a configurable parameter used to define whether the matching degree is high enough to determine if it indicates equipment aging.

[0045] Physical observation information refers to direct observation data of the target vessel by the ship's radar and other sensors, which is generally considered relatively reliable. Changes in the target vessel's course and speed are both less than the second preset threshold, meaning the target vessel has not performed any significant physical maneuvers; for example, its course change does not exceed 5 degrees within a preset time, and its speed change does not exceed 0.5 knots. The second preset threshold is also a parameter that can be adjusted according to the actual application scenario and safety requirements. When both of the above conditions (i.e., high matching degree and stable physical observation information) are simultaneously met, it can be determined that the source of the persistent inconsistency information is benign data noise caused by equipment aging.

[0046] The present invention matches the fluctuation characteristics of AIS broadcast information with a preset set of equipment aging characteristic patterns, enabling the identification of abnormal patterns related to equipment aging at the data level. Simultaneously, it verifies the actual motion state of the target vessel by combining physical observation information. It is precisely because both conditions—AIS data fluctuation and stable physical observation information—are met simultaneously that the system can rule out the possibility of actual maneuvering by the target vessel, thus accurately attributing persistent inconsistencies to benign data noise caused by equipment aging. This dual verification mechanism effectively improves the accuracy and reliability of determining the source of inconsistencies.

[0047] In some embodiments of the present invention, a method is proposed to determine the source of persistent inconsistency information as benign data noise caused by equipment aging based on the matching degree between wave characteristics and a preset set of equipment aging characteristic patterns, as well as physical observation information. However, in actual navigation environments, some complex situations may exist. That is, when the matching degree between wave characteristics and the set of equipment aging characteristic patterns is not high, and the vessel cannot clearly determine the target vessel's motion intention from physical observation information, the above scheme may not be able to accurately identify the specific source of persistent inconsistency information, which may lead to an underestimation or misjudgment of risk. In response, the present invention further proposes a more comprehensive judgment on the source type of persistent inconsistency information to cover more uncertain scenarios.

[0048] The steps described above for determining the source type of persistent inconsistency information based on wave characteristics and physical observation information also include: If the matching degree between the fluctuation characteristics and the patterns in the preset equipment aging characteristic pattern set is lower than the first preset threshold, and the clear movement intention of the target vessel cannot be confirmed based on physical observation information, then the source type of the persistent inconsistency information is determined to be an unknown risk.

[0049] Specifically, when the system detects persistent inconsistencies, it first assesses the degree of matching between the fluctuation characteristics and a preset set of equipment aging characteristic patterns. If the matching degree is lower than a preset first threshold, it indicates that the inconsistency is unlikely to be benign data noise caused by equipment aging. Simultaneously, if the ship's radar and other physical observation equipment cannot provide sufficiently clear or stable information to infer the target vessel's course, speed changes, or other intentions, i.e., the target vessel's explicit intention cannot be confirmed, then the source of the persistent inconsistency will be classified as an unknown risk. The inability to confirm the target vessel's explicit intention can be understood as the inability to identify significant changes consistent with a specific maneuvering pattern (such as turning, acceleration, or deceleration) through analysis of physical observation information, or the observation data itself contains significant uncertainty, making it difficult for the system to reliably predict the target vessel's future behavior.

[0050] This invention effectively overcomes the limitations of relying solely on equipment aging characteristic patterns by introducing a judgment on the source type of "unknown risks." When the fluctuation characteristics of persistently inconsistent information do not conform to the equipment aging pattern, and physical observation information is insufficient to reveal the target vessel's clear movement intentions, the system no longer forcibly classifies or ignores it, but instead identifies it as an unknown risk. This approach allows the system to remain highly vigilant about abnormal data whose causes cannot be clearly attributed, avoiding misjudgments or omissions due to insufficient information. In this way, the invention can more comprehensively cover the complex situations that may occur in actual navigation, improving the ability to identify uncertain risks.

[0051] In some embodiments of the present invention described above, it is proposed that the source of persistent inconsistencies in information be identified as benign data noise or unknown risks caused by equipment aging, based on wave characteristics and physical observation information. However, during actual navigation, the target vessel may perform actual maneuvers, such as turning or accelerating / decelerating. In such cases, its AIS broadcast information may exhibit persistent inconsistencies with the physical observation information observed by the ship's radar and other physical observation equipment due to transmission delays, sensor response speeds, or system update frequencies. If such inconsistencies caused by the target vessel's actual maneuvers are not accurately identified and are misjudged as benign data noise or unknown risks, it may lead to untimely or inaccurate avoidance commands from the ship, thereby increasing the risk of collision.

[0052] In this regard, the present invention further proposes that the step of determining the source type of persistent inconsistency information includes: If physical observation information shows that the target vessel's course or speed changes by more than a second preset threshold within a preset time, and the trend of fluctuation characteristics matches the trend of physical observation information by a third preset threshold, then the source of the persistent inconsistency information is determined to be the actual maneuvering operation of the target vessel.

[0053] Specifically, when physical observation information acquired by the ship's radar or other physical observation equipment shows a significant change in the target vessel's course or speed within a preset time period—that is, when the change exceeds a second preset threshold—this typically indicates that the target vessel is performing an active maneuver. Simultaneously, if the fluctuation characteristics of the AIS broadcast information corresponding to this physical change, such as at least one of the fluctuation frequency, amplitude, or duration, show a high degree of consistency with the trend of the physical observation information—that is, the matching degree reaches a third preset threshold—it can be further confirmed that this persistent inconsistency is not simply random noise or equipment malfunction, but rather that the AIS system is attempting to report the target vessel's actual maneuvering behavior, despite potential delays or instabilities. In this case, the source of the persistent inconsistency information is determined to be the target vessel's actual maneuvering operation.

[0054] This invention effectively identifies persistent inconsistencies caused by the actual maneuvering of a target vessel by combining the direct reflection of the target vessel's actual motion state from physical observation information with the matching degree between the fluctuation characteristics in AIS broadcast information and the changing trends of physical observation information. Specifically, when physical observation information clearly indicates that the target vessel is undergoing a significant change in course or speed, this provides the first layer of evidence that the target vessel is maneuvering. Based on this, if the AIS broadcast information, although fluctuating, has a fluctuation trend that highly matches the maneuvering trend observed physically, it indicates that the AIS system is attempting to reflect this actual maneuver, although there may be instability in data updates or transmission. This dual verification mechanism—the directness of physical observation and the correlation between AIS fluctuation trends—allows the system to distinguish such inconsistencies from benign data noise caused by equipment aging or uncertainties of unknown origin, thereby avoiding misjudgment.

[0055] In some of the aforementioned embodiments, while it is possible to determine the source type of persistent inconsistency information based on wave characteristics and physical observation information, and to assign risk weights to persistent inconsistency information according to the determined source type, in practical applications, how to specifically and differentiate the risk weights assigned to inconsistency information of different source types to more accurately reflect their potential collision risks remains a problem that needs further refinement. Failure to address this problem may lead to the adoption of equal or inappropriate avoidance strategies for inconsistency information of different risk levels, thereby affecting the effectiveness and safety of avoidance commands. To address this, the present invention further proposes specific steps for assigning risk weights to persistent inconsistency information based on the determined source type, aiming to more accurately assess and respond to potential navigation risks through differentiated risk weight allocation.

[0056] The steps described above for assigning risk weights to persistently inconsistent information based on the identified source type include: If the source type is benign data noise caused by equipment aging, then the first risk weight is assigned to persistent inconsistency information; If the source type is uncertainty of unknown origin, then a second risk weight is assigned to persistently inconsistent information, where the second risk weight is greater than the first risk weight; If the source type is the actual maneuvering operation of the target vessel, a third risk weight is assigned to the persistent inconsistency information, where the third risk weight is greater than the second risk weight.

[0057] Specifically, the first risk weight is assigned to benign data noise caused by equipment aging. This type of noise typically manifests as slight, non-continuous, or predictable fluctuations in AIS broadcast information, while physical observations show minimal changes in the target vessel's course and speed. Since this inconsistency has a relatively small impact on actual navigation safety, it is assigned a low risk weight, for example, a value range of 1-3. The second risk weight is assigned to uncertainties of unknown origin. When the fluctuation characteristics match a pattern in a preset set of equipment aging characteristic patterns less than a first preset threshold, and the target vessel's clear intention to move cannot be confirmed based on physical observations, this inconsistency is classified as an unknown risk. Because its potential risk level is higher than benign data noise, but it lacks a clear danger signal, it is assigned a medium risk weight, for example, a value range of 4-7, and this second risk weight is greater than the first risk weight. In practical applications, the third risk weight is assigned to the target vessel's actual maneuvering operations. When physical observation information shows that the target vessel's course or speed changes by more than a second preset threshold within a preset time period, and the trend of the fluctuation characteristics matches the trend of the physical observation information to a third preset threshold, it indicates that the target vessel is performing actual maneuvering. In this case, inconsistent information may mean that the target vessel's intentions deviate from the ship's expectations, posing a high risk of collision. Therefore, it is assigned the highest risk weight, for example, a value range of 8-10, and this third risk weight is greater than the second risk weight.

[0058] The present invention achieves a refined assessment of potential navigation risks by assigning differentiated risk weights to persistent inconsistencies from different sources. Specifically, when inconsistencies are determined to be benign data noise caused by equipment aging, they are assigned a low first risk weight because their impact on navigation safety is relatively small. This allows the subsequent avoidance command generation process to avoid overreaction, thereby reducing unnecessary course or speed adjustments. When inconsistencies are determined to be uncertainties of unknown origin, they are assigned a moderate second risk weight because their potential risk is higher than benign noise but not a clear danger. This prompts the system to remain vigilant while still allowing for some observation and assessment. When inconsistencies are determined to be actual maneuvers of the target vessel, they are assigned the highest third risk weight because they are directly related to changes in the target vessel's intentions and may lead to a higher collision risk. This prompts the system to generate more proactive and decisive avoidance commands. This hierarchical weighting mechanism enables the system to adopt more reasonable and effective avoidance strategies based on the actual nature and degree of the risk.

[0059] In some embodiments of the present invention described above, although navigation avoidance commands for the vessel are generated based on risk weights, in actual navigation, different levels of collision risk require different intensities and types of avoidance measures. Generating only a single type of avoidance command may not effectively cope with the complex and ever-changing navigation environment, leading to overly conservative or insufficient avoidance operations, thereby affecting navigation efficiency or safety. To address this, the aforementioned AI-based ship navigation optimization control method further proposes a step of generating navigation avoidance commands for the vessel based on risk weights. This involves calculating a collision risk threshold and, based on a comparison between this threshold and a preset collision threshold, generating fine-tuned or dangerous avoidance commands to achieve more refined and intelligent avoidance control.

[0060] Specifically, the steps for generating a navigation avoidance instruction for this vessel based on risk weights include: The collision risk threshold between this vessel and the target vessel is calculated based on risk weights. If the collision risk threshold does not reach the preset collision threshold, a fine-tuning avoidance command is generated. The fine-tuning avoidance command is used to control the ship to gradually increase the safe distance from the target ship by making one or more fine adjustments to its course or speed. If the collision risk threshold reaches the preset collision threshold, a hazard avoidance command is generated in accordance with the International Maritime Collision Avoidance Regulations. The hazard avoidance command is used to control the vessel's turning or deceleration.

[0061] Upon receiving the assigned risk weights, the system uses these weights to calculate the collision risk threshold between the vessel and the target vessel. This collision risk threshold can be understood as a quantitative indicator measuring the likelihood of a collision between the two vessels. Its calculation comprehensively considers factors such as the target vessel's relative position, relative speed, heading, and the uncertainties reflected by the risk weights. For example, parameters such as DCPA (Distance to Nearest Encounter) and TCPA (Time to Nearest Encounter) can be used, combined with the risk weights, to perform a weighted calculation and obtain a comprehensive collision risk assessment value.

[0062] Furthermore, the preset collision threshold refers to a safety limit set internally by the system to distinguish between low-risk and high-risk collision scenarios. When the calculated collision risk threshold does not reach the preset threshold, it indicates that the current collision risk is relatively low, and the system will generate a fine-tuning avoidance command. The fine-tuning avoidance command aims to gradually increase the safe distance between the vessel and the target vessel through small, gradual adjustments to course or speed, avoiding unnecessary drastic maneuvers, thereby maintaining navigational stability and economy.

[0063] Furthermore, when the calculated collision risk threshold reaches or exceeds the preset collision threshold, it indicates a high collision risk, requiring more decisive avoidance measures. In this case, the system will generate a hazard avoidance instruction based on the International Regulations for Preventing Collisions at Sea (COLREGs). The COLREGs are globally recognized regulations for preventing ship collisions, specifying the avoidance actions that ships should take in different encounter situations. The hazard avoidance instruction strictly follows these rules; for example, instructing the vessel to make a sharp turn to avoid a collision, or to immediately reduce speed or even stop to ensure navigational safety.

[0064] This invention achieves refined management of navigation avoidance commands by introducing a collision risk threshold calculation and judgment mechanism. First, based on assigned risk weights, the system can more accurately assess the actual collision risk between the vessel and the target vessel, rather than relying solely on a single risk weight. It is this quantitative risk assessment that enables the system to distinguish between different levels of dangerous scenarios. Second, by comparing the calculated collision risk threshold with a preset collision threshold, the system can intelligently determine whether the current situation is a low-risk state requiring fine-tuning or a high-risk state requiring emergency avoidance. This tiered judgment mechanism avoids excessive avoidance measures in low-risk situations, thereby improving navigation efficiency and fuel economy; simultaneously, it ensures timely and decisive effective avoidance in accordance with international rules in high-risk situations, greatly enhancing navigation safety.

[0065] Secondly, see Figure 2 The present invention also discloses an artificial intelligence-based ship navigation optimization control system, comprising: The information acquisition module 210 is used to acquire the physical observation information of the ship's radar of surrounding target ships and the AIS broadcast information of the target ships. The feature calculation module 220 is used to compare physical observation information with AIS broadcast information to identify whether there is persistent inconsistency between physical observation information and AIS broadcast information. When persistent inconsistency is identified, the fluctuation characteristics of the AIS broadcast information corresponding to the persistent inconsistency are calculated. The source type determination module 230 is used to determine the source type of persistent inconsistency information based on wave characteristics and physical observation information; The risk weight allocation module 240 is used to assign risk weights to persistent inconsistency information based on the determined source type. The instruction generation module 250 is used to generate navigation and avoidance instructions for the ship based on risk weights.

[0066] The system proposed in this invention aims to achieve intelligent processing and risk assessment of multi-source heterogeneous information during ship navigation through modular design. This effectively distinguishes between benign data noise caused by equipment aging and actual navigation risks, thereby optimizing avoidance strategies and improving navigation safety and economy. The system forms a closed-loop intelligent control circuit through the collaborative work of its various functional modules, enabling ships to cope more intelligently and efficiently with complex navigation environments and avoid excessive avoidance due to information uncertainty.

[0067] The AI-based ship navigation optimization control system proposed in this invention represents a significant advancement over existing technologies. Traditional systems, when faced with inconsistencies between AIS information and radar observations, often adopt a conservative strategy, treating all inconsistencies as potential risks, leading to frequent and unnecessary avoidance maneuvers. For example, when the target vessel's AIS information experiences minor fluctuations due to equipment aging, existing systems may fail to recognize their benign nature, triggering significant avoidance maneuvers, resulting in course deviations and fuel waste. The core innovation of this invention lies in its modular intelligent processing architecture, particularly through the collaborative work of the feature calculation module 220, the source type judgment module 230, and the risk weight allocation module 240, which achieves an intelligent judgment mechanism for the source type of persistently inconsistent information. This system can analyze the fluctuation characteristics of AIS broadcast information and combine it with radar physical observation information to distinguish between benign data noise caused by equipment aging, the target vessel's actual maneuvering, and unknown risks. This refined risk identification capability allows the vessel to allocate different risk weights according to the actual risk level, and the command generation module 250 generates more accurate and optimized avoidance commands. For example, for benign data noise caused by equipment aging, this system can assign a lower risk weight to generate fine-tuned avoidance instructions, thus avoiding excessive avoidance; while for actual maneuvering operations or unknown risks, a higher risk weight can be assigned to generate more mandatory danger avoidance instructions. This differentiated approach significantly improves the intelligence level of navigation decision-making, effectively reduces unnecessary avoidance, improves navigation efficiency and economy, and ensures navigation safety.

[0068] The above description is merely an embodiment of the present invention and is not intended to limit the scope of protection of the present invention. For those skilled in the art, the present invention can have various modifications and variations. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A ship navigation optimization control method based on artificial intelligence, characterized in that, include: Acquire physical observation information of surrounding target vessels by the ship's radar, as well as AIS broadcast information of the target vessels; The physical observation information is compared with the AIS broadcast information to identify whether there is any persistent inconsistency between the physical observation information and the AIS broadcast information. When the persistent inconsistency is identified, the fluctuation characteristics of the AIS broadcast information corresponding to the persistent inconsistency are calculated. Based on the wave characteristics and the physical observation information, determine the source type of the persistent inconsistency information; Based on the determined source type, risk weights are assigned to the persistent inconsistency information; Based on the aforementioned risk weights, a navigation avoidance instruction is generated for this vessel.

2. The ship navigation optimization control method based on artificial intelligence according to claim 1, characterized in that, The physical observation information includes the target vessel's first position, first speed, and first heading.

3. The ship navigation optimization control method based on artificial intelligence according to claim 1, characterized in that, The AIS broadcast information includes the target vessel's reported second position, second speed, second course, and vessel identification code.

4. The ship navigation optimization control method based on artificial intelligence according to claim 3, characterized in that, The fluctuation characteristics include at least one of fluctuation frequency, fluctuation amplitude, and fluctuation duration. The step of calculating the fluctuation characteristics of the AIS broadcast information corresponding to the persistent inconsistency information includes: Calculate at least one of the following based on the AIS broadcast information corresponding to the persistent inconsistency information: The fluctuation frequency is obtained by calculating the number of times the target vessel reports changes in its second position or second speed within a preset unit time. Calculate the standard deviation of the second position or the second speed from its average value to obtain the fluctuation range; The duration of the fluctuation is obtained by calculating the length of time during which the persistent inconsistency information occurs.

5. The ship navigation optimization control method based on artificial intelligence according to claim 1, characterized in that, The step of determining the source type of the persistent inconsistency information based on the wave characteristics and the physical observation information includes: The fluctuation characteristics are matched with a preset set of equipment aging characteristic patterns; If the fluctuation characteristics match the patterns in the preset equipment aging characteristic pattern set more than a first preset threshold, and the physical observation information shows that the changes in the target vessel's heading and speed are less than a second preset threshold, then the source type of the persistent inconsistency information is determined to be benign data noise caused by equipment aging.

6. The ship navigation optimization control method based on artificial intelligence according to claim 5, characterized in that, The step of determining the source type of the persistent inconsistency information based on the wave characteristics and the physical observation information further includes: If the degree of matching between the fluctuation characteristics and the patterns in the preset equipment aging characteristic pattern set is lower than the first preset threshold, and the clear movement intention of the target vessel cannot be confirmed based on the physical observation information, then the source type of the persistent inconsistency information is determined to be an unknown risk.

7. The ship navigation optimization control method based on artificial intelligence according to claim 6, characterized in that, The step of determining the source type of the persistent inconsistency information based on the wave characteristics and the physical observation information further includes: If the physical observation information shows that the target vessel's course or speed changes by more than the second preset threshold within a preset time period, and the matching degree between the change trend of the fluctuation characteristics and the change trend of the physical observation information reaches the third preset threshold, then the source type of the persistent inconsistency information is determined to be the actual maneuvering operation of the target vessel.

8. The ship navigation optimization control method based on artificial intelligence according to claim 7, characterized in that, The step of assigning risk weights to the persistent inconsistency information based on the determined source type includes: If the source type is benign data noise caused by equipment aging, then a first risk weight is assigned to the persistent inconsistency information; If the source type is uncertainty of unknown origin, then a second risk weight is assigned to the persistent inconsistency information, wherein the second risk weight is greater than the first risk weight; If the source type is the actual maneuvering operation of the target vessel, then a third risk weight is assigned to the persistent inconsistency information, wherein the third risk weight is greater than the second risk weight.

9. The ship navigation optimization control method based on artificial intelligence according to claim 1, characterized in that, The steps for generating navigation avoidance instructions for the vessel based on the risk weights include: Calculate the collision risk threshold between this vessel and the target vessel based on the risk weights; If the collision risk threshold does not reach the preset collision threshold, a fine-tuning avoidance command is generated. The fine-tuning avoidance command is used to control the ship to gradually increase the safe distance from the target ship by making one or more fine adjustments to its course or speed. If the collision risk threshold reaches the preset collision threshold, a hazard avoidance command is generated in accordance with the International Maritime Collision Avoidance Code. The hazard avoidance command is used to control the ship to turn or decelerate.

10. A ship navigation optimization control system based on artificial intelligence, characterized in that, include: The information acquisition module is used to acquire the physical observation information of the ship's radar of surrounding target vessels and the AIS broadcast information of the target vessels; The feature calculation module is used to compare the physical observation information with the AIS broadcast information to identify whether there is persistent inconsistency information between the physical observation information and the AIS broadcast information. When persistent inconsistency information is identified, the module calculates the fluctuation characteristics of the AIS broadcast information corresponding to the persistent inconsistency information. The source type determination module is used to determine the source type of the persistent inconsistency information based on the fluctuation characteristics and the physical observation information. The risk weight allocation module is used to allocate risk weights to the persistent inconsistency information based on the determined source type. The instruction generation module is used to generate navigation and avoidance instructions for the ship based on the risk weights.

Images