Offshore wind power construction navigation risk assessment method and system

Abstract

The invention discloses an offshore wind power construction navigation risk assessment method and system, and belongs to the field of construction risk assessment. The method comprises the following steps: firstly, identifying risk sources, such as channel obstruction, large hoisting collision risk involved in assembly operation and the like; then, according to a risk identification result, constructing a Bayesian network model, taking the identified risk events as nodes, taking a causal relationship between the risk events as edges, collecting historical data, and using expert assessment to determine a conditional probability table; and then, reasoning is carried out by using the established Bayesian network model, the risk probability of each risk event is obtained, and the risk is evaluated according to the evaluation of the risk probability and possible consequences. And finally, according to a risk assessment result, precautionary measures and emergency plans are made for events with relatively high risks, so that the risk level is reduced. The method is systematic and scientific, an effective risk management mode can be provided for offshore wind power construction, and the safety and efficiency of the construction process can be improved.

Description

Technical Field

[0001] The present invention belongs to the field of construction risk assessment, and more specifically relates to a method and system for assessing navigation risk in offshore wind power construction. Background Art

[0002] Offshore wind power construction is a complex engineering activity involving the mutual influence and restriction of multiple factors, such as process design, equipment performance, human resources, environmental conditions, etc. During the construction process, due to the complex and changeable marine environment, various risks are prone to occur, such as equipment failure, casualties, delays in construction schedule, etc., which may cause a large amount of economic losses to the project and even threaten the safety of personnel lives.

[0003] Therefore, it is particularly important to carry out navigation risk assessment during the construction of offshore wind power. The main task of navigation risk assessment is to identify and analyze possible risks and provide necessary prevention and response strategies to ensure the smooth progress of construction and the safety of personnel and equipment. In addition, to control risks, it is also necessary to assess the risk level so that corresponding management measures can be taken.

[0004] Traditional offshore wind power construction navigation risk assessment methods are mainly based on qualitative risk analysis techniques, such as risk matrix, risk prevention analysis, etc. However, these methods have weak qualitative descriptions of risks and often cannot accurately assess the risk level and the effectiveness of response strategies. In addition, with the development and application of wind power construction technology, the complexity and uncertainty of the construction environment are also increasing, and these traditional methods are often difficult to meet the actual needs of modern offshore wind power construction.

[0005] As an advanced probabilistic graphical model, Bayesian network has been successfully applied in many fields, such as medical diagnosis, intelligent decision-making, etc. It can effectively deal with problems involving uncertainty and complexity by graphically representing the causal and probabilistic relationships between variables. Therefore, the use of Bayesian network to assess the risk of navigation in offshore wind power construction has great practical value and broad prospects. Summary of the invention

[0006] The present invention aims to solve the limitations of traditional offshore wind power construction navigation risk assessment methods, especially the accuracy and efficiency issues in risk identification, risk analysis and risk control. This requires us to construct a risk assessment method based on the Bayesian network model, which aims to deeply identify and comprehensively assess the risks in engineering construction through a scientific probability model, so as to effectively respond to and control the risks, reduce possible losses, ensure the safety and reliability of the construction process, and improve the overall construction efficiency and economic benefits.

[0007] In order to achieve the above object, the present invention is implemented by adopting the following technical scheme: the method comprises:

[0008] Risk identification: Based on the actual construction situation, risk sources are identified during the construction phase, such as obstruction of waterways and collision risks caused by large hoisting involved in assembly operations.

[0009] Establishing a Bayesian network model: Based on the risk identification results, a Bayesian network model is constructed, with the identified risk events as nodes and the causal relationships between risk events as edges. Parameter learning: Collecting historical data or determining the conditional probability table of each node in the network through expert evaluation is the basis for Bayesian network learning and reasoning.

[0010] Risk analysis and assessment: Use the constructed Bayesian network model to reason and derive the risk probability of each risk event, and assess the risk based on the risk probability and possible consequences. Risk control: Based on the risk assessment results, formulate preventive measures and emergency plans for events with higher risks to reduce the risk level.

[0011] In one embodiment, the risk identification includes:

[0012] (1) Entities involved in the construction of offshore wind farms;

[0013] (2) Dangerous situations and accidents that may occur during the construction of offshore wind farms;

[0014] (3) Direct or potential risk sources that may cause accidents during the construction of offshore wind farms.

[0015] In one scheme, the establishment of the Bayesian network model includes: using a dynamic Bayesian network to probabilistically model the dependency relationship between different features corresponding to the variables, reflecting the temporal relationship between the features, and reflecting the probabilistic dependency relationship between the variables and the changes over time through the network topology structure. It is suitable for modeling complex features such as offshore wind power construction that have both feature correlation and temporal correlation.

[0016] In one embodiment, the parameter learning includes:

[0017] Determine the conditional probability table for each node in the Bayesian network; based on historical data or expert experience;

[0018] Use statistical learning methods to estimate these values, maximum likelihood estimation methods;

[0019] There is a data set D, in which each row of records represents an assignment of all nodes, and n is the total number of rows in D. Then for the case of A=a and B=b, the sample frequency is calculated as:

[0020] f_A,B(a,b)=n(A=a,B=b) / n

[0021] Here, n(A=a,B=b) is the number of rows in data set D where A=a and B=b occur simultaneously;

[0022] Then according to the maximum likelihood estimation method, the parameter theta is calculated as:

[0023] theta(A=a|B=b)=f_A,B(a,b) / f_B(b)

[0024] In this formula, when the parent node set B is valued as b, the probability of node A taking the value of a is equal to the joint frequency of A taking the value of a and B taking the value of b divided by the frequency of B taking the value of b.

[0025] In one embodiment, the risk analysis and assessment includes:

[0026] Risk probability inference: Based on the Bayesian network model, the known information is used to calculate the conditional probability of each risk event occurring;

[0027] Risk impact assessment: For each risk event, assess the impact or loss it will cause after it occurs, through historical data analysis or expert experience;

[0028] Risk value calculation: For each risk event, calculate its risk value based on its probability of occurrence and the consequences it brings; if the probability of risk event A occurring is p, and the impact or loss caused is L, its risk value R is calculated as R=p*L.

[0029] In one embodiment, the risk control includes:

[0030] Determine the risk events that need to be focused on: rank the risk events according to their probability and the consequences they bring, and focus on the risk events with greater risk value;

[0031] Develop emergency plans: Emergency plans detail how various resources will be deployed, the authority and process for key decisions, and the methods of coordination and linkage after a risk event occurs.

[0032] In one embodiment, a navigation risk assessment system for offshore wind power construction is provided, and the system is applicable to the method described, and the system comprises:

[0033] Risk identification module, Bayesian network model building module, parameter learning module, risk analysis and assessment module, risk control module;

[0034] Among them, the risk identification module identifies the risk sources in the construction phase according to the actual construction situation;

[0035] The Bayesian network model building module builds a Bayesian network model based on the risk identification results;

[0036] The parameter learning module uses historical data or expert evaluation to determine the conditional probability table for each node in the network;

[0037] The risk analysis and assessment module uses the constructed Bayesian network model to make inferences and derive the risk probability of each risk event;

[0038] The risk control module formulates protective measures and emergency plans for high-risk events based on the assessment results.

[0039] In one embodiment, the parameter learning module performs parameter learning using maximum likelihood estimation or Bayesian estimation, and determines the conditional probability table of each node based on historical data or expert evaluation.

[0040] In one solution, the risk control module formulates protective measures and emergency plans for high-risk events based on risk assessment results, including determining key risk events to be controlled, formulating preventive measures and emergency plans for different risk events, executing and testing preventive measures and emergency plans, and updating risk levels.

[0041] Beneficial effects of the present invention:

[0042] The offshore wind power construction navigation risk assessment method based on the Bayesian network model can deeply identify and comprehensively evaluate the risks in engineering construction, conduct effective quantitative analysis of risks, and provide corresponding risk control strategies, thereby significantly improving the safety and economy of construction projects.

[0043] The Bayesian network model can effectively deal with the uncertainty and complexity in offshore wind power construction, and can better capture and describe the complex relationship between risk events, including the mutual influence relationship between risk sources, risk events, and risk consequences.

[0044] The present invention can not only intuitively display the risk path and help understand the source and transmission mechanism of risk, but also provide decision makers with more accurate risk level judgment basis and risk control decision support through risk quantification results, thereby improving the scientific level of risk decision-making. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 is a flow chart of the method of the present invention;

[0046] Figure 2 The dynamic Bayesian network diagram of the present invention;

[0047] Figure 3 This is a Bayesian network model for risk assessment during the construction phase of the present invention. DETAILED DESCRIPTION

[0048] In order to facilitate understanding of the present invention, the present invention will be described more fully below with reference to the relevant drawings. Typical embodiments of the present invention are provided in the drawings. However, the present invention can be implemented in many different forms and is not limited to the embodiments described in the present invention. On the contrary, the purpose of providing these embodiments is to make the disclosure of the present invention more thorough and comprehensive.

[0049] Unless otherwise defined, all technical and scientific terms used in the present invention have the same meaning as those commonly understood by those skilled in the art to which the present invention belongs. The terms used in the present invention in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. In order to facilitate the understanding of the present invention, the present invention will be described more fully below with reference to the relevant drawings. Typical embodiments of the present invention are shown in the drawings. However, the present invention can be implemented in many different forms and is not limited to the embodiments described in the present invention. On the contrary, the purpose of providing these embodiments is to make the disclosure of the present invention more thorough and comprehensive.

[0050] Embodiment 1:

[0051] like Figure 1 As shown, the method for assessing the risk of navigation in offshore wind power construction includes:

[0052] S1. Risk identification: Based on the actual construction situation, risk sources are identified during the construction phase, such as obstruction of waterways and collision risks caused by large hoisting involved in assembly operations.

[0053] The entities involved in the construction of offshore wind farms are:

[0054] (1) Personnel: mainly construction workers

[0055] (2) Ship: the ship itself and the equipment on board

[0056] (3) Wind towers, wind turbines, and booster stations: Wind power facilities and equipment with wind towers, wind turbines, and booster stations as the main components

[0057] 2. Dangerous accidents that may occur during the construction of offshore wind farms:

[0058] (1) Casualties: Construction workers collide with each other when getting on and off the ship, fall into the water, or slip;

[0059] (2) Ship accidents: collision (between construction ships, between construction ships and wind turbines, between construction ships and nearby fishing boats), grounding (mainly occurs in offshore intertidal zone wind farm construction), fire, wind disaster, and dragging anchor;

[0060] (3) Installation accidents of wind towers, wind turbines and booster stations: the hook falls when lifting piles, the sliding bearing is damaged, etc.

[0061] 3. Direct or potential risk sources that may cause accidents during the construction of offshore wind farms:

[0062] Table 1 Main risk sources during the construction phase

[0063]

[0064]

[0065] S2. Establish a Bayesian network model: Based on the risk identification results, a Bayesian network model is constructed, with the identified risk events as nodes and the causal relationships between risk events as edges.

[0066] The establishment of the Bayesian network model includes: using a dynamic Bayesian network to perform probabilistic modeling on the dependency relationship between different features corresponding to the variables, reflecting the temporal relationship between the features, and reflecting the probabilistic dependency relationship between the variables and the changes over time through the network topology structure. It is suitable for modeling complex features such as offshore wind power construction that have both feature correlation and temporal correlation.

[0067] Dynamic Bayesian Network (DBN) is an extension of Bayesian Network. It is a Bayesian Network that connects different variables with adjacent time steps. It is an extension of static network in time series. It can not only express the causal relationship between variables like Bayesian Network, but also describe the evolution of the state of each variable in time series. That is, dynamic Bayesian Network can model and analyze dynamic events.

[0068] Figure 2 The variable X in t ={C t ,M t ,O t} has a probability dependency that changes with time t=1,2,…n, Variable Set X i The joint probability distribution of is:

[0069] P(X t )=P(C t ,M t ,O t )=P(C t )P(M t / C t )P(O t / C t ,M t )

[0070] Consider O t and C t The conditional probability between:

[0071]

[0072] Between time t-1 and t, the variable set C t The state of has been transferred, so the variable set X t The transition probability is:

[0073] P(X t |X t-1 )=P(C t |C t-1 )

[0074] The dynamic Bayesian network can not only perform probabilistic modeling of the dependency relationship between different features corresponding to the variables, but also better reflect the temporal relationship between the features. It reflects the probabilistic dependency relationship between variables and its changes over time through the network topology structure. Therefore, it is suitable for modeling complex features such as offshore wind power construction that have both feature correlation and temporal correlation.

[0075] In addition to the advantages of static Bayesian networks, dynamic Bayesian networks also have time characteristics, which allows the influence of time to be considered in the process of analyzing problems, ensuring continuity in the process of reasoning, and making the entire reasoning method appear to be more in line with the development laws of objective things. Dynamic Bayesian networks can automatically retain information when inputting it, and can well combine previous data and current evidence information. The storage of information is cumulative over time, which effectively reduces the uncertainty during information fusion reasoning and improves the accuracy of information fusion.

[0076] The dynamic Bayesian network is based on a probabilistic network. On the basis of the static Bayesian network, time information is added, so the dynamic model based on this has the ability to process time series data. However, it is relatively complicated to establish a general dynamic Bayesian network model. Generally, assumptions and simplifications need to be made before establishing the model. This article makes the following assumptions about the dynamic Bayesian network:

[0077] Assumption 1: Within a finite time, for all t, the change process of conditional probability is stable and consistent;

[0078] Assumption 2: The dynamic probability process is Markovian, and the probability of an event occurring in the future has nothing to do with the past, but only with the present moment;

[0079] Assumption 3: The conditional probability process remains stationary at adjacent times, that is, probability has no relationship with time.

[0080] The four stages of the full life cycle of offshore wind power construction and production (planning stage, construction stage, operation and maintenance stage, and abandonment stage) are continuous, with an obvious time sequence, and are dynamically connected on the entire timeline. However, considering that the goals of each stage are different, there are certain differences in the composition of the risk source network. Therefore, it is difficult to achieve risk prediction at each stage of offshore wind power construction and production through simple network probability transfer. Therefore, in the risk study of each stage of the full life cycle of offshore wind power construction and production, this study selects one node as a representative according to the sequence of each stage on the timeline, and according to the risk structure and characteristics of each stage, and simplifies the analysis of the risk characteristics of the dynamic Bayesian network with the idea of ​​static Bayesian network and risk source replacement.

[0081] According to historical accident data and expert questionnaire survey, the risk factors in the construction and production of offshore wind power are as follows: ① casualties; ② people falling into the water; ③ collisions between construction workers getting on and off the ship; ④ people slipping; ⑤ collision; ⑥ stranding; ⑦ fire; ⑧ wind disaster; ⑨ anchor dragging; ⑩ installation accidents of wind towers, wind turbines, and booster stations; The access standards for construction workers are unclear; Construction workers were not certified to work; Inadequate training of construction personnel; Overall, the overall quality of the crew is poor compared to professional crew members; The crew is unfit for duty; Unclear job responsibilities; Low physical and psychological quality; Operational errors; Lookout negligence; Failure to comply with navigation regulations; The access standards for construction vessels are unclear; The construction company is quite casual in choosing the ship type; The competent authorities do not strictly control the construction permits for construction vessels; The construction vessels are lacking in professional equipment; High requirements for ship seaworthiness; Transport vessels and other transportation ships are seriously overloaded; The life-saving facilities are inadequate; Fishing boats roam freely in the work area; Accidents caused by strong winds and waves; The source of weather forecasts is inaccurate; Weather forecasts are not timely; The depth of the intertidal zone near the coast varies greatly; Sudden severe weather; The sheltered anchorage is far away; Construction workers have a weak sense of safety; The emergency response plan is not perfect; The emergency plan is not very operational; The emergency response plan is not comprehensive; Failure to respond in time turns a dangerous situation into an accident; Lifesaving measures were not timely; The conditions for limiting construction work are not clear; Lack of necessary warning signs; No effective monitoring system; The communication effect at sea is not ideal; Search and rescue forces far offshore are relatively weak; There are many shipping companies and their management is chaotic; Lack of effective communication and cooperation between the construction unit and the construction unit; The maritime supervision force and facilities are not in place. The Bayesian network model for risk assessment in the construction and production phase of offshore wind power generation is as follows: Figure 3 shown.

[0082] S3. Parameter learning: Collect historical data or determine the conditional probability table of each node in the network through expert evaluation. This step is the basis of Bayesian network learning and reasoning.

[0083] Determine the conditional probability table for each node in the Bayesian network; based on historical data or expert experience.

[0084] In the Bayesian network, the Bayesian conditional probability table is the link between the nodes. When the node status is unknown, the basic node probability can be determined by hydrological and meteorological data and prior probability, while the intermediate nodes and top nodes need to be determined by the conditional probability table.

[0085] There are generally two methods to obtain the conditional probability table. One is to use a large amount of accident data to learn the Bayesian network, thereby obtaining the conditional probability table. Using historical accident data to learn can be very persuasive and authentic, and is an indispensable means of assessing risks. However, historical accident data is often not detailed and comprehensive enough. If accident data is used as the only means to obtain the conditional probability table (CPT), the lack of accident data may cause the loss of important nodes in the Bayesian network, thereby affecting the accuracy of the CPT. Another method is to use expert questionnaires to obtain the conditional probability table. Although expert experience can make up for the lost nodes to a certain extent, if only expert experience is used to determine the conditional probability table, it will also cause the entire Bayesian network to be too subjective, and the loss of objective data support will reduce the credibility of the entire Bayesian network. In order to make up for the defects of the two methods, this paper will use historical accident data to optimize the conditional probability table obtained from the expert questionnaire after verifying its reliability, and verify the effect of the optimization.

[0086] This paper collected 2057 accident records in recent years at home and abroad. After a detailed analysis of the accident process, the nodes in the Bayesian network of the sinking accident risk reflected in the accident process were counted. Considering that there are some unreasonable parts in the conditional probability table obtained based solely on historical accident data, this is because the complete cause of the accident cannot be obtained from the accident data, which leads to missing basic nodes and intermediate nodes in the Bayesian network, affecting the accuracy of the conditional probability table. It is of great significance to optimize and improve the conditional probability table by integrating the experience of a large number of experts in the field to make up for the errors that may be caused by simple accident data learning.

[0087] In order to make the expert experience more reliable, it is necessary not only to collect as many expert opinions as possible, but also to ensure that the experts surveyed are people with good knowledge and rich experience in the shipping industry. It is worth mentioning that the experts surveyed in this article include professors and doctors in shipping disciplines, captains and first mates with many years of sailing experience, and front-line staff engaged in wind farm construction and operation and maintenance. The combination of professionalism and practicality, the questionnaire survey results can be used as a consensus in the field.

[0088] Use statistical learning methods to estimate these values, maximum likelihood estimation methods;

[0089] There is a data set D, in which each row of records represents an assignment of all nodes, and n is the total number of rows in D. Then for the case of A=a and B=b, the sample frequency is calculated as:

[0090] f_A,B(a,b)=n(A=a,B=b) / n

[0091] Here, n(A=a,B=b) is the number of rows in data set D where A=a and B=b occur simultaneously;

[0092] Then according to the maximum likelihood estimation method, the parameter theta is calculated as:

[0093] theta(A=a|B=b)=f_A,B(a,b) / f_B(b)

[0094] In this formula, when the parent node set B is valued as b, the probability of node A taking the value of a is equal to the joint frequency of A taking the value of a and B taking the value of b divided by the frequency of B taking the value of b.

[0095] S4. Risk analysis and assessment: Use the constructed Bayesian network model to perform reasoning to derive the risk probability of each risk event, and assess the risk based on the risk probability and possible consequences.

[0096] In the risk analysis and assessment stage, the risk probability of each risk event is obtained mainly through reasoning (graph model inference) on the constructed Bayesian network model, and the risk is assessed based on these probabilities and the possible consequences.

[0097] Risk probability inference: Based on the Bayesian network model, known information (such as current environmental conditions, risk events that have occurred, etc.) can be used to calculate the conditional probability of each risk event. This step is usually completed through probabilistic reasoning algorithms, such as the variable elimination algorithm and belief propagation algorithm for exact reasoning, and the resampling algorithm and variational reasoning algorithm for approximate reasoning.

[0098] Risk impact assessment: For each risk event, it is necessary to assess the impact or loss that may be caused after it occurs, which can be done through historical data analysis or expert experience.

[0099] Risk value calculation: For each risk event, its risk value can be calculated based on its probability of occurrence and possible consequences. Common risk value calculation methods include expected value method, worst case method, etc. If the probability of risk event A is p, and the impact or loss caused is L, its risk value R can be calculated as R = p*L.

[0100] Risk sorting and prioritization: Based on the risk value of all risk events, they can be sorted and prioritized. Risk events with high risk value (i.e., high risk probability and / or high potential impact) are prioritized.

[0101] Through this process, we can not only know the probability of each risk event, but also understand the possible impact of these risk events, so as to make targeted risk assessment and risk control decisions.

[0102] S5. Risk control: Based on the risk assessment results, formulate preventive measures and emergency plans for events with higher risks to reduce the risk level. The risk control process based on the risk assessment results specifically includes the following steps:

[0103] Identify risk events that require key control: Sort risk events by probability and possible consequences, and prioritize risk events with greater risk value.

[0104] Formulate preventive measures: For each risk event that needs to be focused on, formulate effective preventive measures based on the specific causes and conditions of its occurrence. Preventive measures can include controlling the conditions that cause the risk (such as improving processes, improving equipment performance, etc.), mitigating the impact of the risk after it occurs (such as installing emergency shutdown equipment, improving maintenance capabilities, etc.), etc.

[0105] Develop an emergency plan: An emergency plan refers to how to respond quickly and effectively after a risk event occurs to minimize the impact of the risk. The emergency plan can detail how various resources (including personnel, equipment, materials, etc.) are deployed, the authority and process for key decisions, and the coordination and linkage methods after a risk event occurs.

[0106] Implementation and testing of preventive measures and emergency plans: After the preventive measures and emergency plans are formulated, they need to be implemented and tested and rehearsed regularly. Through testing and rehearsals, the shortcomings and deficiencies of the preventive measures and emergency plans can be discovered in time, and timely improvements can be made.

[0107] Update of risk level: After implementing preventive measures and emergency plans, the risk will be reassessed and the risk level will be updated based on the actual effect of these measures. This is an ongoing process that needs to be repeated to ensure the effectiveness of risk control.

[0108] Embodiment 2:

[0109] A navigation risk assessment system for offshore wind power construction, the system is applicable to the method described, and the system comprises:

[0110] Risk identification module, Bayesian network model building module, parameter learning module, risk analysis and assessment module, risk control module;

[0111] Among them, the risk identification module identifies the risk sources in the construction phase according to the actual construction situation;

[0112] The Bayesian network model building module builds a Bayesian network model based on the risk identification results;

[0113] The parameter learning module uses historical data or expert evaluation to determine the conditional probability table for each node in the network;

[0114] The risk analysis and assessment module uses the constructed Bayesian network model to make inferences and derive the risk probability of each risk event;

[0115] The risk control module formulates protective measures and emergency plans for high-risk events based on the assessment results.

[0116] The parameter learning module uses maximum likelihood estimation or Bayesian estimation to perform parameter learning, and determines the conditional probability table of each node based on historical data or expert evaluation.

[0117] The risk control module formulates protective measures and emergency plans for high-risk events based on risk assessment results, including determining key risk events to be controlled, formulating preventive measures and emergency plans for different risk events, executing and testing preventive measures and emergency plans, and updating risk levels.

[0118] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program, and the program can be stored in a computer-readable storage medium, and when the program is executed, it can include the processes of the embodiments of the above-mentioned methods. The storage medium can be a disk, an optical disk, a read-only memory (ROM) or a random access memory (RAM).

[0119] It should be understood that the detailed description of the technical solutions of the present invention by means of the preferred embodiments is illustrative rather than restrictive. A person skilled in the art may modify the technical solutions described in the embodiments, or replace some of the technical features by equivalents, based on reading the specification of the present invention; and these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for assessing navigation risk during offshore wind power construction, characterized in that: The method includes: Risk identification: Based on the actual construction situation, identify the risk sources during the construction phase; Establish a Bayesian network model: Based on the risk identification results, a Bayesian network model is constructed, with the identified risk events as nodes and the causal relationships between risk events as edges; Parameter learning: Collect historical data or determine the conditional probability table for each node in the network through expert evaluation; Risk analysis and assessment: Use the constructed Bayesian network model to reason, derive the risk probability of each risk event, and assess the risk based on the risk probability and possible consequences; Risk control: Based on the risk assessment results, formulate preventive measures and emergency plans for events with higher risks.

2. A method for assessing navigation risk during offshore wind power construction according to claim 1, characterized in that: Features: The risk identification includes: (1) Entities involved in the construction of offshore wind farms; (2) Dangerous situations and accidents that may occur during the construction of offshore wind farms; (3) Direct or potential risk sources that may cause accidents during the construction of offshore wind farms.

3. The method for assessing navigation risk during offshore wind power construction according to claim 1, characterized in that: The establishment of the Bayesian network model includes: using a dynamic Bayesian network to perform probabilistic modeling on the dependency relationship between different features corresponding to the variables, reflecting the temporal relationship between the features, and reflecting the probabilistic dependency relationship between the variables and the changes over time through the network topology structure. It is suitable for modeling complex features such as offshore wind power construction that have both feature correlation and temporal correlation.

4. The method for assessing navigation risk during offshore wind power construction according to claim 1, characterized in that: The parameter learning includes: Determine the conditional probability table for each node in the Bayesian network; based on historical data or expert experience; Use statistical learning methods to estimate these values, maximum likelihood estimation methods; There is a data set D, in which each row of records represents an assignment of all nodes, and n is the total number of rows in D. Then for the case of A=a and B=b, the sample frequency is calculated as: f_A,B(a,b)=n(A=a,B=b) / n Here, n(A=a,B=b) is the number of rows in data set D where A=a and B=b occur simultaneously; Then according to the maximum likelihood estimation method, the parameter theta is calculated as: theta(A=a|B=b)=f_A,B(a,b) / f_B(b) In this formula, when the parent node set B is valued as b, the probability of node A taking the value of a is equal to the joint frequency of A taking the value of a and B taking the value of b divided by the frequency of B taking the value of b.

5. The method for assessing navigation risk during offshore wind power construction according to claim 1, characterized in that: The risk analysis and assessment described include: Risk probability inference: Based on the Bayesian network model, the known information is used to calculate the conditional probability of each risk event occurring; Risk impact assessment: For each risk event, assess the impact or loss it will cause after it occurs, through historical data analysis or expert experience; Risk value calculation: For each risk event, calculate its risk value based on its probability of occurrence and the consequences it brings; if the probability of risk event A occurring is p, and the impact or loss caused is L, its risk value R is calculated as R=p*L.

6. The method for assessing navigation risk during offshore wind power construction according to claim 1, characterized in that: The risk controls described include: Determine the risk events that need to be focused on: rank the risk events according to their probability and the consequences they bring, and focus on the risk events with greater risk value; Develop emergency plans: Emergency plans detail how various resources will be deployed, the authority and process for key decisions, and the methods of coordination and linkage after a risk event occurs.

7. A navigation risk assessment system for offshore wind power construction, the system being applicable to the method according to any one of claims 1 to 6, characterized in that: The system comprises: Risk identification module, Bayesian network model building module, parameter learning module, risk analysis and assessment module, risk control module; Among them, the risk identification module identifies the risk sources in the construction phase according to the actual construction situation; The Bayesian network model building module builds a Bayesian network model based on the risk identification results; The parameter learning module uses historical data or expert evaluation to determine the conditional probability table for each node in the network; The risk analysis and assessment module uses the constructed Bayesian network model to make inferences and derive the risk probability of each risk event; The risk control module formulates protective measures and emergency plans for high-risk events based on the assessment results.

8. The offshore wind power construction navigation risk assessment system according to claim 7 is characterized in that: The parameter learning module uses maximum likelihood estimation or Bayesian estimation to perform parameter learning, and determines the conditional probability table of each node based on historical data or expert evaluation.

9. The offshore wind power construction navigation risk assessment system according to claim 7 is characterized in that: The risk control module formulates protective measures and emergency plans for high-risk events based on risk assessment results, including determining key risk events to be controlled, formulating preventive measures and emergency plans for different risk events, executing and testing preventive measures and emergency plans, and updating risk levels.

Images