Method and device for preventing touch accident of autonomous ship and medium

Abstract

The invention relates to an autonomous ship touch accident prevention method and device and a medium, and the method comprises the following steps: obtaining ship operation data, and employing a rooting theory and system theory process analysis method to recognize and obtain a plurality of autonomous ship touch risk factors; constructing a plurality of autonomous ship touch risk typical scenes based on the ship operation data, dynamically simulating an evolution process of autonomous ship touch risk factors through a fuzzy cognitive map method, and iteratively calculating a steady-state value of each autonomous ship touch risk factor in each autonomous ship touch risk typical scene, corresponding collision accident prevention and control countermeasures are made according to the autonomous ship touch risk factors with the steady-state values ranked top N, and an interaction matrix used in the steady-state value iterative calculation process is determined according to a comprehensive influence matrix obtained through calculation based on a decision test and evaluation laboratory method. Compared with the prior art, the navigation safety of the autonomous ship can be further improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of water traffic safety, and in particular relates to a method, device and medium for preventing autonomous ship collision accidents. Background Art

[0002] With the rapid development of ship automation and intelligent technologies, autonomous ships, as the future development direction of shipping, have attracted extensive attention from the academic and industrial circles at home and abroad. Although autonomous ships are expected to reduce the accident risks caused by human factors, the new technologies and new systems they adopt may also bring new safety hazards. Collision accidents are one of the most common types of accidents during navigation and an important safety challenge faced by autonomous ships. These accidents may not only cause casualties and environmental damage but also result in huge economic losses and business interruptions. In view of the above practical problems, to ensure the safe operation of autonomous ships, it is necessary to deeply analyze the risk factors and accident causes of their collision accidents and formulate targeted safety prevention and control strategies.

[0003] The existing ship risk identification and assessment schemes include two aspects: 1) One is the ship risk identification system and assessment method under the influence of specific waters or specific risk factors, such as for key waters and key ships (CN118194073A), inland river ship risks (CN111951607A), inland river multi-scenario ship navigation risks (CN115510644A), ship risks in restricted waters (CN116029554A), the risk of ship self-sinking affected by inland river wind and waves (CN114655387A), the regional ship collision risk based on fuzzy logic (CN114464014A), ship risks under the influence of typhoons (CN117789529A); 2) The other is the ship risk identification system and assessment method using specific methods or systems, such as for the navigation risks of all ships (CN115310673A), ship management business data and AIS data analysis (CN116777218A), big data analysis (CN116151625A), fuzzy neural network analysis (CN118410344A), virtual simulation training system (CN118571101A), ship-shore collaborative multi-source data (CN117953729A), edge computing (CN117854324A).

[0004] It should be noted that the research object of the above improved technical solution is relatively single, and it does not start from the perspective of autonomous ships and touch accidents. There are gaps in its risk identification system and assessment method. The accident data analysis at the regional scale helps to grasp the risk situation of local waters, but it is difficult to reveal the overall accident rules of global ships, especially for touch accidents with a relatively high occurrence frequency; the overall solution lacks consideration of the dynamics of the risk evolution of the assessment object and has not fully revealed the coupling relationship of risk factors. In summary, there is an urgent need to design a new type of touch accident prevention plan to comprehensively examine the risk factors and accident causes of autonomous ships touching accidents under different environmental conditions, and provide reliable decision-making support for government agencies, shipping companies, rescue organizations, etc. Summary of the Invention

[0005] The purpose of the present invention is to overcome the defects of the above-mentioned existing technologies and provide a method, device and medium for preventing autonomous ship touch accidents, so as to further improve the safety of autonomous ship navigation.

[0006] The purpose of the present invention can be achieved through the following technical solutions:

[0007] The present invention provides a method for preventing autonomous ship touch accidents, including the following steps:

[0008] Obtain ship operation data, and identify multiple autonomous ship touch risk factors by using grounded theory and system theory process analysis methods. The ship operation data includes historical touch accident data, multi-source maritime accident investigation reports and autonomous ship risk research literature;

[0009] Based on the ship operation data, construct multiple typical scenarios of autonomous ship touch risks, dynamically simulate the evolution process of autonomous ship touch risk factors by using the fuzzy cognitive map method, iteratively calculate the steady-state values of each autonomous ship touch risk factor in each typical scenario of autonomous ship touch risks, and formulate corresponding collision accident prevention and control countermeasures according to the top N autonomous ship touch risk factors ranked by the steady-state values. The interaction matrix used in the steady-state value iterative calculation process is determined according to the comprehensive influence matrix calculated by the decision-making trial and evaluation laboratory method.

[0010] Furthermore, the specific process of identifying multiple autonomous ship touch risk factors is as follows:

[0011] Extract the characteristics of ship touch accidents according to the historical touch accident data;

[0012] Analyze the multi-source maritime accident investigation reports by using grounded theory, and extract traditional ship touch accident risk factors in combination with the characteristics of ship touch accidents;

[0013] Adopt the systematic theory process analysis method, construct a safety control structure model for autonomous ships according to the research literature on autonomous ship risks, and extract multiple autonomous ship collision risk factors by combining the risk factors of traditional ship collision accidents.

[0014] Further, the risk factors of traditional ship collision accidents include human / organizational factors, ship equipment factors, and environmental factors.

[0015] Further, the safety control structure model for autonomous ships includes a supervision layer, a company layer, a shore-based control layer, a ship layer, and a physical process layer, and corresponding safety control behaviors are provided between each layer; compare and integrate the risk factors of traditional ship collision accidents with the safety control behaviors between each layer of the safety control structure model for autonomous ships to obtain multiple autonomous ship collision risk factors.

[0016] Further, multiple typical scenarios of autonomous ship collision risks include berthing and unberthing scenarios in port and terminal waters and navigation scenarios in bridge waters. Each typical scenario of autonomous ship collision risk includes several types of unsafe control behaviors and the corresponding collision risk scenarios for each type of unsafe control behavior.

[0017] Further, the specific process of iteratively calculating the steady-state values of each autonomous ship collision risk factor in each typical scenario of autonomous ship collision risk is as follows:

[0018] Take the autonomous ship collision risk factors as nodes, use fuzzy logic to represent the causal relationships between the nodes, and construct a fuzzy cognitive map;

[0019] Iteratively update the state values of each node in the fuzzy cognitive map according to the interaction matrix until the change value of the state value is less than the set threshold. At this time, the state values of each node are the steady-state values of the corresponding autonomous ship collision risk factors.

[0020] Further, count the frequencies of each autonomous ship collision risk factor in each typical scenario of autonomous ship collision risk, calculate the direct influence matrix based on the decision-making trial and evaluation laboratory method, and then obtain the comprehensive influence matrix.

[0021] Further, the comprehensive influence matrix includes the influence degree value, the influenced degree value, the centrality value, and the weight of each autonomous ship collision risk factor.

[0022] The present invention also provides an electronic device, including a memory, a processor, and a program stored in the memory. When the processor executes the program, the above method is implemented.

[0023] The present invention also provides a computer-readable storage medium, on which a computer program is stored. When the program is executed by a processor, the above method is implemented.

[0024] Compared with the prior art, the present invention has the following beneficial effects:

[0025] 1. The present invention proposes a method for preventing autonomous ship collision accidents. First, by using two methods, namely grounded theory and system theory process analysis, risk factors for autonomous ship collisions are identified based on ship operation data, which can ensure the comprehensiveness and effectiveness of the obtained risk factors for autonomous ship collisions. Then, multiple typical scenarios of autonomous ship collision risks are constructed, and the steady-state values of each risk factor for autonomous ship collisions in each typical scenario are iteratively calculated by the fuzzy cognitive map method. Corresponding collision accident prevention and control countermeasures are formulated according to the top N risk factors for autonomous ship collisions ranked by the steady-state values. The interaction matrix used in the iterative calculation process of the steady-state values is determined based on the comprehensive influence matrix calculated by the decision-making trial and evaluation laboratory method. In the above process, the decision-making trial and evaluation laboratory method can start from the frequency of risk factors and reveal the static importance degree of risk factors through matrix operations, while the fuzzy cognitive map method can dynamically simulate the evolution process of risk factors for autonomous ship collisions, quantitatively evaluate the importance degree of risk factors and their dynamic change laws. Combining these two methods can overcome the limitation of subjective assignment in causal modeling, accurately analyze the static and dynamic interaction relationships between risk factors, identify the core causal factors in autonomous ship collision accidents, and thus more pertinently formulate prevention and control countermeasures, fundamentally strengthening the systematicness, perception generalization ability, emergency response level and risk pre-control quality of autonomous navigation safety, and further effectively reducing the risk of collision accidents.

[0026] 2. The present invention extracts the characteristics of ship collision accidents from a macroscopic level based on historical collision accident data, then analyzes the multi-source maritime accident investigation reports by using grounded theory to extract the risk factors of traditional ship collision accidents from a microscopic level, and finally uses the system theory process analysis method to construct an autonomous ship safety control structure model on the basis of the research literature on autonomous ship risks. The model includes a supervision layer, a company layer, a shore-based control layer, a ship layer and a physical process layer, and corresponding safety control behaviors are provided between each layer. By comparing and integrating the risk factors of traditional ship collision accidents with the safety control behaviors between each layer, multiple risk factors for autonomous ship collisions are obtained, providing support for subsequent screening of key risk factors. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 is a schematic flow chart of the method of the present invention;

[0028] Figure 2 is a schematic structural diagram of the autonomous ship safety control structure model;

[0029] Figure 3 is a schematic flow chart of the calculation of the steady-state value of the risk factor for autonomous ship collision;

[0030] Figure 4 is the steady-state value of each autonomous ship collision risk factor in the berthing and unberthing scenarios in the waters of port terminals;

[0031] Figure 5 is the steady-state value of each autonomous ship collision risk factor in the navigation scenarios in bridge waters. Specific implementation manners

[0032] The present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. This embodiment is implemented on the premise of the technical solution of the present invention, and gives detailed implementation manners and specific operation processes, but the protection scope of the present invention is not limited to the following embodiments.

[0033] Embodiment:

[0034] This embodiment provides a method for preventing autonomous ship collision accidents, as Figure 1 shown, mainly including the following three steps:

[0035] S1. Identification of autonomous ship collision risk factors: Using two methods, namely grounded theory and system theory process analysis, analyze and identify autonomous ship collision risk factors from the perspectives of accident experience and system safety.

[0036] S2. Construction of typical scenarios for autonomous ship collision risks: Conduct in-depth research on the construction of collision risk scenarios of autonomous ships in complex water environments.

[0037] S3. Analysis of the core causes of autonomous ship collision accidents: Calculate the comprehensive influence matrix of risk factors through the decision-making trial and evaluation laboratory method, and then dynamically and iteratively calculate the steady-state values of each autonomous ship collision risk factor in each typical scenario of autonomous ship collision risks through the fuzzy cognitive map method. Formulate corresponding countermeasures for preventing collision accidents according to the autonomous ship collision risk factors ranked top N in the steady-state values.

[0038] The specific processes of each step are as follows:

[0039] S1. Identification of autonomous ship collision risk factors.

[0040] S21. Identification of traditional ship collision risk factors based on grounded theory.

[0041] Obtain ship operation data, including historical collision accident data, multi-source maritime accident investigation reports, and autonomous ship risk research literature. In this embodiment, 6,572 global ship collision accident data from 1977 to 2022 in the Llyod’s Intelligence database are selected as the historical collision accident data, and the characteristics of ship collision accidents are extracted from them.

[0042] In this embodiment, 184 typical ship collision accident investigation reports are selected as multi-source maritime accident investigation reports, including 127 reports of ships colliding with docks and their ancillary facilities and 57 reports of ships colliding with bridges. They come from different countries and organizations, cover different waters such as inland rivers and offshore areas, and different ship types such as passenger ships and cargo ships, and have a certain degree of representativeness and diversity. The grounded theory is used to deeply analyze the ship collision accident investigation reports, focusing on the content such as the process of the accident, direct causes and indirect causes in the reports, especially the elements related to ship collision risks such as human behavior, ship status, organizational factors and environmental conditions, and extract traditional ship collision accident risk factors from the micro level. When a risk factor is identified, it is labeled with a concise word or phrase to form an initial concept, such as "not reducing speed in a timely manner according to berthing conditions", "the captain ignoring the crew's warning", etc. Through the open coding of multi-source maritime accident investigation reports, more than 700 initial concepts are refined, covering multiple aspects such as crew operation, ship equipment, environmental conditions, and management systems, laying a foundation for the subsequent axial coding. Excerpts of the original statements of some accident reports and the corresponding open coding are shown in Table 1.

[0043] Table 1 Correspondence between partial original statements and initial concepts

[0044]

[0045]

[0046] Through repeated comparison of the initial concepts, several initial categories are identified. For example, concepts such as "not reducing speed in a timely manner according to berthing conditions" and "excessive speed approaching the dock" are classified into the category of "unsafe speed"; concepts such as "the captain ignoring the crew's warning" and "poor communication between the pilot and the tugboat" are classified into the category of "insufficient communication and cooperation".

[0047] Then through axial coding, more than 700 initial concepts are summarized into 42 initial categories, covering different aspects such as human factors, ship equipment factors, environmental factors and organizational management factors. These initial categories reveal the general risk factors that cause ship collision accidents. Some open coding concepts and the corresponding categories are shown in Table 2.

[0048] Table 2 Correspondence between partial open coding concepts and corresponding categories

[0049]

[0050] Based on the three aspects of "human unsafe behavior", "unsafe state of objects", and "adverse environmental impacts", the core categories of influencing factors are determined, forming 32 main categories, including 24 human / organizational factors such as unsafe speed, insufficient communication and collaboration, fatigue, and insufficient crew training, 3 ship equipment factors such as propulsion control system or equipment failure, power system failure, and steering control system or equipment failure, and 5 environmental factors such as narrow waterways, tidal influence, wind influence, wave influence, and current influence. These core categories and main categories represent the key risk factors of traditional ship collision accidents, and the specific results are shown in Table 3.

[0051] Table 3 Risk Factors of Traditional Ship Collision Accidents

[0052]

[0053]

[0054] S22. Combine the systems-theoretic process analysis method and the risk factors of traditional ship collision accidents to identify the risk factors of autonomous ship collisions.

[0055] Autonomous ship collision accidents may cause serious consequences such as casualties, ship damage, cargo loss, damage to wharves and bridges, and environmental pollution. Therefore, in this embodiment, it is defined as a system-level accident. To systematically analyze the risk factors of collision accidents, it is necessary to determine a reasonable analysis boundary, mainly including the autonomous ship and its on-board subsystems (such as the autonomous ship bridge system, propulsion and steering systems, on-board sensors, etc.), the shore-based control center, the external objects with which the ship interacts (such as port wharves, navigation aids, other ships, etc.), communication and navigation facilities, and ship management and supervision departments, etc. These elements constitute the safety control structure of the autonomous ship operation through control and feedback relationships, and are the objects that need to be focused on in the collision risk analysis.

[0056] Based on the technical characteristics and operation mode of autonomous ships, as well as the basic principles of Systems-Theoretic Process Analysis (STPA), an autonomous ship safety control structure model is constructed according to the autonomous ship risk research literature, and multiple autonomous ship collision risk factors are obtained by combining the extraction of traditional ship collision accident risk factors. The autonomous ship safety control structure model is as Figure 2 shown, and vertically includes the supervision layer (corresponding to Figure 2 the competent authorities and coastal state management departments in Figure 2 ), the company layer (corresponding to Figure 2 the shipping company in Figure 2the autonomous integrated bridge system, propulsion system, steering system, generator set, and other ship auxiliaries and various sensors), and the physical process layer (corresponding to Figure 2 the ship navigation process and auxiliary processes such as anchoring and weighing anchor in

[0057] Table 4 Explanation of Some Safety Control Behaviors

[0058]

[0059] A total of 32 key safety control behaviors were identified, involving control links such as information perception, intelligent decision-making, instruction execution, monitoring and feedback, as well as organizational levels such as company management, ship-shore interaction, and human-machine collaboration, as shown in Table 5. The autonomous ship safety control structure model comprehensively depicts the control constraint mechanism in the operation process of autonomous ships, laying a foundation for carrying out collision risk analysis.

[0060] Table 5 Safety Control Behaviors

[0061]

[0062]

[0063]

[0064] After clarifying the safety control structure of autonomous ship operation, further analyze the possible unsafe states and their manifestations of each control behavior. Compare and integrate the 32 traditional ship collision accident risk factors identified by grounded theory with the autonomous ship safety control structure model, and finally screen out 28 collision risk control behaviors applicable to autonomous ships. Among them, the unsafe states in perception include failure to accurately identify environmental information, navigation and positioning system failure, etc.; in decision-making include defects in route planning and collision avoidance algorithms, insufficient berthing and unberthing plans, etc.; in control include pilot error, improper emergency measures, etc.; in execution include delayed response of propulsion and steering, improper operation of the anchor, etc.; in human-machine interaction include failure to switch to manual control in time, neglect of observation, etc.; in ship-shore collaboration include communication interruption, insufficient communication and cooperation, etc. On this basis, considering the possible consequences of each control behavior under different failure modes (such as not provided, provided but incorrect, provided at the wrong time, improper duration, etc.), 28 key unsafe control behaviors of autonomous ship collision accidents, that is, autonomous ship collision risk factors, are summarized as shown in Table 6.

[0065] Table 6 Autonomous Ship Collision Risk Factors

[0066]

[0067]

[0068] S2. Construction of Typical Scenarios for the Risk of Autonomous Ship Collision

[0069] Based on the Unsafe Control Actions (UCA) in autonomous ship collision accidents, some typical accident scenarios can be further identified. For example, when an autonomous ship approaches or leaves a dock, it may collide due to the failure to correctly perceive the position and distance of the dock, or when passing through a bridge area, it may hit a bridge pier due to the failure of the decision-making system. These scenarios may cause varying degrees of losses such as casualties, ship damage, and damage to docks and bridges. At the same time, by tracing the causes of each scenario, some common problems can be found, such as the unreliability of the perception system, the limitations of decision-making algorithms, the timeliness of control instructions, and the complexity of human-machine interaction. If these problems are not effectively controlled, it may lead to the occurrence of collision accidents. In this embodiment, the following two typical scenarios for the risk of autonomous ship collision are constructed:

[0070] (1) Berthing and Unberthing Scenarios in Port and Dock Waters

[0071] By systematically analyzing the investigation reports of a large number of ship collisions with docks and combining the analysis of the design and operation characteristics of autonomous ships, ten main types of unsafe control actions and the resulting dock collision risk scenarios during the berthing and unberthing process in dock waters are identified, covering aspects such as environmental perception, communication transmission, decision-making and planning, power control, and human-machine collaboration, as shown in Table 7.

[0072] Table 7 Risk Scenarios of Autonomous Ship Collision in Dock Waters

[0073]

[0074] Due to the immaturity of autonomous ship technology, the lack of perfect technical standards and sufficient operation experience, there are certain uncertainties and vulnerabilities in its perception, decision-making, execution and other systems. In a complex berthing environment, it may face risks such as inaccurate perception, information transmission interruption, unreasonable decision-making, power out of control, and poor human-machine collaboration, which may lead to the ship colliding with the dock.

[0075] (2) Navigating Scenarios in Bridge Waters

[0076] During the navigation process in bridge waters, based on the in-depth analysis of the investigation reports of ship collisions with bridges in the bridge area and considering the analysis of the design and operation characteristics of autonomous ships, nine main types of unsafe control actions and the resulting bridge collision risk scenarios are identified, involving aspects such as bridge and waterway environment perception, ship motion perception, route planning and collision avoidance decision-making, power control response, communication interruption, and manual takeover, as shown in Table 8.

[0077] Table 8 Risk Scenarios of Autonomous Ship Collision in Bridge Waters

[0078]

[0079]

[0080] Due to the complex structure of the bridge area and the small navigable safety margin, higher requirements are put forward for the perception ability, intelligent decision-making and operation control of autonomous ships. Once failures such as perception blind spots, decision-making mistakes or control failures occur, it is extremely easy to cause accidents such as hitting bridges or navigational obstacles.

[0081] S3. Analysis of the core causes of autonomous ship collision accidents.

[0082] By using the Fuzzy Cognitive Map (FCM) method, dynamically simulate the evolution process of the risk factors of autonomous ship collision, iteratively calculate the steady-state values of each risk factor of autonomous ship collision in each typical scenario of autonomous ship collision risk, and formulate corresponding prevention and control countermeasures for collision accidents according to the risk factors of autonomous ship collision ranked top N in the steady-state values. The interaction matrix used in the iterative calculation process of the steady-state values is determined according to the comprehensive influence matrix obtained by the Decision Making Trial and Evaluation Laboratory (DEMATEL) method. The DEMATEL method starts from the frequency of risk factors, reveals the complex direct and indirect influence relationships between risk factors through matrix operations, and defines a series of indicators to quantitatively evaluate the importance of risk factors, providing an analysis framework for understanding the formation mechanism of ship collision accidents and identifying key risks; combined with the FCM method to further dynamically simulate the evolution process of risk factors, and through the risk state evolution curve obtained by simulation reasoning, the importance degree and its dynamic change law of risk factors can be quantitatively evaluated, providing a basis for formulating targeted safety management strategies.

[0083] The cause analysis of ship collision accidents based on DEMATEL-FCM can overcome the limitation of subjective assignment in causal modeling, and its process is specifically as Figure 3 shown. First, count the frequencies of each risk factor of autonomous ship collision in each typical scenario of autonomous ship collision risk, as shown in Table 9 specifically. Then, use the DEMATEL method to calculate the comprehensive influence matrix T. The calculation process of the comprehensive influence matrix T is as follows: First, calculate the direct influence matrix A. The direct influence degree D ij is the element of the direct influence matrix A, reflecting the direct influence relationship between risk factors x i and x j , and the calculation formula is as follows:

[0084]

[0085] In formula (1), f xi∩xj is the co-occurrence frequency; f xi is the marginal frequency of risk factor x i .

[0086] However, the direct influence matrix A ignores the possible indirect influences between factors. To reveal the complete influence network of risk factors, it is necessary to calculate the comprehensive influence matrix on the basis of the direct influence matrix by normalizing the direct influence matrix.

[0087]

[0088] T = D * (E - D * ) -1 (3)

[0089] The normalized direct influence matrix A* can be obtained through formula (2); on this basis, the comprehensive influence matrix T is calculated through formula (3), where E is the identity matrix.

[0090] To quantitatively analyze the status and role of risk factors, the DEMATEL method introduces three indicators: "influence degree", "affected degree" and "centrality". According to the comprehensive influence matrix T, the influence degree B of the touched risk factor is calculated as shown in formula (4); according to the comprehensive influence matrix T, the affected degree C of the factor is calculated as shown in formula (5).

[0091]

[0092]

[0093] In the above formula, the influence degree B i represents the comprehensive influence of factor x i on all factors; C i represents the comprehensive influence of factor x i by all factors.

[0094] Adding the influence degree B and the affected degree C, the centrality B + C is obtained. The centrality B + C is the sum of the i-th row and the i-th column of the comprehensive influence matrix T, indicating the role of risk factor x i in the ship collision accident. The greater the centrality B + C, the greater the influence of the risk factor on the accident. Dividing the centrality B + C by the sum of the total centralities, the weight of risk factor x i is obtained, so as to more intuitively reflect the relative importance of each risk factor in the accident.

[0095] The influence degree, affected degree, centrality, and weight calculation values of the risk factors for ship-berth collisions are specifically shown in Table 10. It can be seen that insufficient communication and cooperation, wind influence, improper emergency measures, improper ship operation, and insufficient tugboat assistance are the most important risk factors for ship-berth collisions, with centralities of 9.308, 5.729, 5.669, 5.465, and 5.444 respectively. Similarly, the static importance degrees of the risk factors for ship-bridge collisions can be obtained, as shown in Table 11. It can be seen that water flow influence, insufficient communication and cooperation, insufficient risk assessment, insufficient training / experience, and improper collision avoidance measures are the most important risk factors for ship-bridge collisions, with centralities of 8.064, 7.728, 7.241, 7.028, and 5.851 respectively.

[0096] In the risk analysis of ship collision accidents, each risk factor is a concept node in the FCM. Let S i (t) represent the state value of risk factor x i at a certain moment, which reflects the severity of this risk factor. In terms of input, the state values S i of each risk factor x i (0) at the initial moment are given based on historical data. These state values are used as the starting input for calculations and represent the initial severity of the risk factors. In the risk analysis of ship collision accidents, the initial state values can be determined based on historical data such as the frequency of occurrence of this risk factor and the degree of impact caused in previous similar accidents.

[0097] Over time, the state value of each risk factor will change dynamically under the influence of other risk factors. This influence relationship can be described by the interaction matrix W of the FCM, where W ji represents the influence intensity and direction of risk factor x j on risk factor x i (a positive value indicates a positive correlation, and a negative value indicates a negative correlation). At the same time, the elements in the comprehensive influence matrix T obtained by the DEMATEL method determine the weights in the interaction matrix of the FCM. Therefore, at the (t + 1)th moment, the state value of risk factor x i can be updated through Equation (6):

[0098]

[0099] where f is a threshold function in the form of a sigmoid, which maps the concept state value to the interval [0, 1] through Equation (7):

[0100]

[0101]

[0102] In terms of output, by iteratively calculating the above state update formulas (6) and (7) until S i (t+1) reaches stability, the final result of risk evolution can be obtained. This process can be vividly understood as follows: The initial risk state gradually spreads and amplifies through the mutual influence between risk factors, and finally reaches a dynamic balance. The final risk steady-state value R is shown in formula (8). The risk steady-state value is used to measure the magnitude of risk. The larger its value, the greater the risk.

[0103] Table 9 Frequencies of Risk Factors for Each Autonomous Ship in Typical Scenarios of Autonomous Ship Collision

[0104]

[0105]

[0106] Table 10 Static Importance Degrees of Risk Factors for Ship Collision with Wharf

[0107]

[0108]

[0109] Table 11 Static Importance Degrees of Risk Factors for Ship Collision with Bridge

[0110]

[0111]

[0112] By analyzing the influence relationship between elements through the FCM method, the current situation of the system can be diagnosed and analyzed to find out the reasons and influencing factors leading to the current state. After determining the initial concept values (frequency values) and threshold functions of risk factors, the weights of W in the interaction matrix of FCM are determined according to the elements in the comprehensive influence matrix T obtained by DEMATEL calculation. According to the formula mentioned above, after 8 iterative calculations, the final values of each factor reach stability.

[0113] As Figure 4 shown, among the factors of ship collision with wharf, insufficient communication and cooperation (Xq18) has the greatest impact on ship collision with wharf, followed by improper emergency measures (Xq6), improper ship operation (Xq28), wind influence (Xq25), and insufficient training / experience (Xq13). The steady-state values of risk factors are 0.9993, 0.9884, 0.9846, 0.9836, and 0.9834 respectively. As Figure 5As shown, among the factors of ship-bridge collision, water flow influence (Xb19), insufficient communication and cooperation (Xb12), insufficient risk assessment (Xb6), insufficient training / lack of experience (Xb10), and improper collision avoidance measures (Xb7) are the top 5 factors with greater impact on ship-bridge collision. The steady-state values of the risk factors are 0.9974, 0.9964, 0.9955, 0.9945, and 0.9790 respectively.

[0114] Compared with the existing ship risk identification systems and assessment methods for specific waters or specific risk factors, such as those for key waters and key ships (CN118194073A), inland river ship risks (CN111951607A), inland river multi-scenario ship navigation risks (CN115510644A), ship risks in restricted waters (CN116029554A), ship self-sinking risks affected by inland river wind and waves (CN114655387A), regional ship collision risks based on fuzzy logic (CN114464014A), ship risks under typhoon influence (CN117789529A), etc., the method proposed in this embodiment can reveal the overall accident rules of global ships. Especially for collision accidents with a relatively high occurrence frequency, it is more in line with the actual scenario.

[0115] In addition, the existing technologies represented by solutions such as those based on the navigation risks of all ships (CN115310673A), ship management business data, AIS data analysis (CN116777218A), big data analysis (CN116151625A), fuzzy neural network analysis (CN118410344A), virtual simulation training system (CN118571101A), ship-shore collaborative multi-source data (CN117953729A), edge computing (CN117854324A), etc. lack in-depth risk analysis for specific navigation areas and environmental scenarios, and also lack consideration of the dynamics of risk evolution, and have not fully revealed the coupling relationship of risk factors. This embodiment establishes a static network relationship and a dynamic evolution model of collision risk factors, and identifies key causal factors such as insufficient communication and cooperation, wind / water flow interference, improper risk assessment, etc. and their coupling rules through simulation reasoning. On this basis, targeted safety prevention and control strategies such as optimizing the ship-shore collaborative mechanism, strengthening complex environment perception and adaptive decision-making, and improving emergency takeover and personnel training are proposed. These countermeasures focus on the key control behaviors and weak links of autonomous ships, and through systematic design and precise intervention, are expected to fundamentally strengthen the systematicness, perception generalization ability, emergency response level, and risk pre-control quality of autonomous navigation safety, thereby effectively reducing the risk of collision accidents.

[0116] The above method can provide scientific basis and decision-making support for government agencies, shipping companies and rescue organizations, so as to more effectively prevent contact accidents, ensure personnel safety, reduce economic losses and protect the marine environment. In the long run, this has an important impact on promoting the sustainable development of the global shipping industry and improving shipping safety and efficiency.

[0117] The above description of the embodiments is for the convenience of those of ordinary skill in the art to understand and use the invention. It is obvious that those skilled in the art can easily make various modifications to these embodiments and apply the general principles described herein to other embodiments without creative efforts. Therefore, the present invention is not limited to the above embodiments, and the improvements and modifications made by those skilled in the art without departing from the scope of the present invention according to the disclosure of the present invention should be within the protection scope of the present invention.

Claims

1. A method for preventing collision accidents of autonomous ships, characterized in that: The following steps are involved: Obtain ship operation data, and use grounded theory and system theory process analysis methods to identify multiple autonomous ship collision risk factors. The ship operation data includes historical collision accident data, multi-source maritime accident investigation reports, and autonomous ship risk research literature; Based on the ship operation data, multiple typical scenarios of autonomous ship collision risks are constructed, the evolution process of autonomous ship collision risk factors is dynamically simulated through the fuzzy cognitive map method, the steady-state values ​​of each autonomous ship collision risk factor in each typical scenario of autonomous ship collision risk are iteratively calculated, and corresponding collision accident prevention and control countermeasures are formulated according to the autonomous ship collision risk factors ranked top N in steady-state values. The interaction matrix used in the iterative calculation process of the steady-state value is determined according to the comprehensive influence matrix obtained by calculation based on the decision-making test and evaluation laboratory method.

2. The method for preventing collision accidents of autonomous ships according to claim 1, characterized in that: The specific process of identifying and obtaining multiple autonomous ship collision risk factors is as follows: Extracting ship collision accident characteristics according to the historical collision accident data; The grounded theory is used to analyze the multi-source maritime accident investigation report, and the risk factors of traditional ship collision accidents are extracted in combination with the characteristics of the ship collision accidents; The system theory process analysis method was adopted to construct an autonomous ship safety control structure model based on the autonomous ship risk research literature, and multiple autonomous ship collision risk factors were obtained by combining the traditional ship collision accident risk factors.

3. The method for preventing collision accidents of autonomous ships according to claim 2, characterized in that: The traditional risk factors for ship collision accidents include human / organizational factors, ship equipment factors and environmental factors.

4. The method for preventing collision accidents of autonomous ships according to claim 2, characterized in that: The autonomous ship safety control structure model includes a supervision layer, a company layer, a shore-based control layer, a ship layer and a physical process layer, and corresponding safety control behaviors are set between each layer; the traditional ship collision accident risk factors are compared and integrated with the safety control behaviors between the layers of the autonomous ship safety control structure model to obtain multiple autonomous ship collision risk factors.

5. The method for preventing collision accidents of autonomous ships according to claim 1, characterized in that: Multiple typical scenarios of autonomous ship collision risks include berthing and unberthing scenarios in port and terminal waters and navigation scenarios in bridge waters. Each typical scenario of autonomous ship collision risk includes several types of unsafe control behaviors and collision risk scenarios corresponding to each type of unsafe control behavior.

6. The method for preventing collision accidents of autonomous ships according to claim 1, characterized in that: The specific process of iteratively calculating the steady-state value of each autonomous ship collision risk factor in each typical autonomous ship collision risk scenario is as follows: Taking the autonomous ship collision risk factors as nodes, fuzzy logic is used to represent the cause-effect relationship between nodes and a fuzzy cognition diagram is constructed. The state value of each node in the fuzzy recognition diagram is iteratively updated according to the interaction matrix until the change value of the state value is less than the set threshold value. At this time, the state value of each node is the steady-state value of the corresponding autonomous ship collision risk factor.

7. The method for preventing collision accidents of autonomous ships according to claim 1, characterized in that: The frequencies of the autonomous ship collision risk factors in each typical scenario of autonomous ship collision risk are counted, and the direct impact matrix is ​​calculated based on the decision-making test and evaluation laboratory method, and then the comprehensive impact matrix is ​​obtained.

8. The method for preventing collision accidents of autonomous ships according to claim 1, characterized in that: The comprehensive impact matrix includes the impact value, the impact value, the centrality value and the weight of each autonomous ship collision risk factor.

9. An electronic device comprising a memory, a processor, and a program stored in the memory, characterized in that: When the processor executes the program, the method according to any one of claims 1 to 8 is implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method according to any one of claims 1 to 8 is implemented.

Images