A multi-agent collaborative optimization method for vertical large models in the field of ocean

Abstract

The present application relates to the technical field of design optimization, in particular to a multi-agent collaborative optimization method for vertical large models in the field of ocean, which comprises: obtaining a candidate path corresponding to a target marine equipment through a path planning agent; obtaining a sea state forecast record of different sea areas and different water depths of the target ocean through a sea state interpretation agent; obtaining a risk review conclusion description record corresponding to each local voyage section through a risk review agent, and obtaining a sea state forecast record used by the risk review agent when performing risk review on each local voyage section; performing a proofreading operation on the risk review matching record and the target sea state forecast record corresponding to the local voyage section; and performing path optimization according to the record proofreading results corresponding to all local voyage sections. Through the proofreading operation, the data reference dislocation of the risk review agent can be identified to a certain extent, thereby improving the rationality of path optimization and the final ocean operation scheme generation.

Description

Technical Field

[0001] This invention relates to the field of design optimization technology, specifically to a multi-agent collaborative optimization method for large vertical models in the marine domain. Background Technology

[0002] With the development of intelligent marine applications, tasks such as subsea pipeline inspection, offshore wind power foundation inspection, port waterway patrol, and close-in observation by underwater robots increasingly rely on large-scale vertical models for marine-specific decision support. In existing complex marine operations, a multi-agent collaborative architecture is typically employed. For example, a path planning agent generates candidate paths, a sea state interpretation agent outputs objective sea state forecasts, a risk verification agent combines this information to output risk assessment conclusions and safety recommendations, and finally, a coordinating agent synthesizes the operational plan.

[0003] However, existing multi-agent cooperative optimization methods for large-scale vertical models in the marine domain often suffer from technical problems: Because the large model driving the risk verification agent may contain contextual logic fallacies, i.e., the illusion of a large model, it is very easy for underlying data to be misreferenced when conducting safety assessments on specific local segments. For example, the risk verification agent may incorrectly extract low-velocity forecast data from adjacent sea areas, adjacent water depths, or outdated versions, and generate a seemingly logical but actually physically inconsistent "safe / low-risk" conclusion based on this misattribution. This leads to poor rationality in path optimization, and consequently, poor rationality in the final marine operation plan. Summary of the Invention

[0004] To address the technical problem of poor rationality in the final marine operation plan due to poor path optimization, this invention proposes a multi-agent collaborative optimization method for large vertical models in the marine domain.

[0005] In a first aspect, this invention provides a multi-agent cooperative optimization method for large-scale vertical models in the marine domain, the method comprising: The path planning agent obtains candidate paths for the target marine equipment and acquires the traffic environment forecast records for each local segment of each candidate path. The sea state interpretation agent obtains sea state forecast records at different water depths in different sea areas of the target ocean. Based on the navigation environment forecast records, the target sea state forecast records corresponding to each local segment are selected from all sea state forecast records. The risk verification conclusion description record corresponding to each local segment is obtained through the risk verification agent, and the sea state forecast record that the risk verification agent can use when performing risk verification on each local segment is obtained, thus forming a set of candidate sea state forecast records corresponding to each local segment. Based on the risk verification conclusion description record corresponding to each local segment, the risk verification matching record corresponding to each local segment is selected from the candidate sea state forecast record set corresponding to each local segment; The risk verification and matching records and target sea state forecast records corresponding to each local segment are verified to obtain the record verification results for each local segment. Path optimization is performed based on the record verification results of all local segments in all candidate paths corresponding to the target marine equipment.

[0006] In conjunction with the first aspect mentioned above, in one possible implementation, the navigation environment forecast record includes the entry time of the navigation segment, the departure time of the navigation segment, the sea area to which the navigation segment belongs, the water depth to which the navigation segment belongs, the tidal current velocity, and the information version; the sea state forecast record includes the start time, end time, sea area, water depth, tidal current velocity, and the information version of the tidal current window; the tidal current window is a continuous period of time in which the tidal current velocity is lower than the operational safety threshold; the risk verification conclusion description record includes the start time, end time, sea area, water depth, tidal current velocity, and the information version.

[0007] In conjunction with the first aspect above, in one possible implementation, the step of selecting the target sea state forecast record corresponding to each local segment from all sea state forecast records based on the navigation environment forecast records includes: Any local segment is designated as a marked local segment, and the corresponding navigation environment forecast record is designated as a marked navigation environment forecast record. Any sea state forecast record is designated as a marked sea state forecast record. If the entry time of the marked passage environment forecast record is greater than or equal to the start time of the marked sea state forecast record, and the departure time of the marked passage environment forecast record is less than or equal to the end time of the marked sea state forecast record, and the sea area to which the passage included in the marked passage environment forecast record belongs is the same as the sea area included in the marked sea state forecast record, and the water depth to which the passage included in the marked passage environment forecast record belongs is the same as the water depth included in the marked sea state forecast record, and the information version included in the marked passage environment forecast record is the same as the information version included in the marked sea state forecast record, then the marked sea state forecast record is determined to be the target sea state forecast record corresponding to the marked local passage.

[0008] In conjunction with the first aspect above, in one possible implementation, the sea state forecast records that the risk verification agent selects to use when performing risk verification for each local segment constitute a candidate sea state forecast record set for each local segment, including: Any local flight segment is designated as a marked local flight segment, and the traffic environment forecast record corresponding to the marked local flight segment is designated as a marked traffic environment forecast record; The sea area to which the navigation segment belongs and its adjacent sea areas included in the marked navigation environment forecast record constitute the reference sea area set corresponding to the marked local navigation segment; The water depths to which the navigation segment belongs and its adjacent water depths are included in the marked navigation environment forecast records, forming a set of reference water depths corresponding to the marked local navigation segment. The information versions included in the marked passage environment forecast record and their adjacent information versions constitute the reference information version set corresponding to the marked local flight segment; From all sea state forecast records, select sea state forecast records whose sea area belongs to the reference sea area set, whose water depth belongs to the reference water depth set, and whose information version belongs to the reference information version set, and use them as candidate sea state forecast records corresponding to the marked local segment; All candidate sea state forecast records corresponding to the marked local segment are used to form the candidate sea state forecast record set corresponding to the marked local segment.

[0009] In conjunction with the first aspect above, in one possible implementation, the step of selecting the risk verification matching record corresponding to each local segment from the candidate sea state forecast record set corresponding to each local segment based on the risk verification conclusion description record corresponding to each local segment includes: Any local segment is designated as a marked local segment, and the risk verification conclusion description record corresponding to the marked local segment is designated as a marked risk verification conclusion description record. Any candidate sea state forecast record in the candidate sea state forecast record set corresponding to the marked local segment is designated as a pending sea state forecast record. Based on the start and end times included in the marked risk verification conclusion description record and the start and end times included in the pending sea state forecast record, the time difference between the marked risk verification conclusion description record and the pending sea state forecast record is determined. Based on whether the target marine equipment needs to perform operations on the marked local segment, determine the operation characterization time corresponding to the marked local segment; Based on the end time included in the marked risk verification conclusion description record and the pending sea state forecast record, as well as the operation characterization time, the degree of operation difference between the marked risk verification conclusion description record and the pending sea state forecast record is determined; The absolute value of the difference between the tidal current velocity included in the marked risk verification conclusion description record and the tidal current velocity included in the undetermined sea state forecast record is determined as the tidal current velocity difference degree between the marked risk verification conclusion description record and the undetermined sea state forecast record. Based on the time difference, the operational difference, and the tidal current velocity difference, determine whether the pending sea state forecast record is the risk verification matching record corresponding to the marked local segment.

[0010] In conjunction with the first aspect above, in one possible implementation, determining the time difference between the marked risk verification conclusion description record and the pending sea state forecast record based on the start and end times included in the marked risk verification conclusion description record and the start and end times included in the pending sea state forecast record includes: The absolute value of the difference between the start time included in the marked risk verification conclusion description record and the start time included in the pending sea state forecast record is determined as the start difference degree between the marked risk verification conclusion description record and the pending sea state forecast record. The absolute value of the difference between the end time included in the marked risk verification conclusion description record and the end time included in the pending sea state forecast record is determined as the end difference degree between the marked risk verification conclusion description record and the pending sea state forecast record. The sum of the initial difference and the final difference is determined as the time difference between the marked risk verification conclusion description record and the pending sea state forecast record.

[0011] In conjunction with the first aspect above, in one possible implementation, determining the operation characterization time corresponding to the marked local segment based on whether the target marine equipment needs to perform operations on the marked local segment includes: If the target marine equipment needs to operate on the marked local section of the waterway, then the start time of the operation of the target marine equipment on the marked local section of the waterway shall be determined as the operation characterization time corresponding to the marked local section of the waterway. If the target marine equipment does not need to operate on the marked local segment, then half of the sum of the entry time and departure time of the navigation environment forecast record corresponding to the marked local segment shall be determined as the operation characterization time corresponding to the marked local segment.

[0012] In conjunction with the first aspect above, in one possible implementation, determining the operational difference between the marked risk verification conclusion description record and the pending sea state forecast record based on the end time included in the marked risk verification conclusion description record and the pending sea state forecast record, and the operational characterization time, includes: The difference between the end time included in the risk verification conclusion description record and the operation characterization time corresponding to the marked local flight segment is determined as the remaining operable time corresponding to the risk verification conclusion description record. The difference between the end time included in the pending sea state forecast record and the operation characterization time corresponding to the marked local segment is determined as the remaining operable time corresponding to the pending sea state forecast record. The absolute value of the difference between the remaining workable time corresponding to the marked risk verification conclusion description record and the remaining workable time corresponding to the pending sea state forecast record is determined as the workability difference between the marked risk verification conclusion description record and the pending sea state forecast record.

[0013] In conjunction with the first aspect above, in one possible implementation, determining whether the pending sea state forecast record is a risk verification matching record corresponding to the marked local navigation segment based on the time difference, the operational difference, and the tidal current velocity difference includes: If the time difference is less than or equal to a preset time difference threshold, and the operation difference is less than or equal to a preset operation difference threshold, and the tidal current velocity difference is less than or equal to a preset tidal current velocity difference threshold, then the pending sea state forecast record is determined to be the risk verification matching record corresponding to the marked local segment; otherwise, the pending sea state forecast record is determined not to be the risk verification matching record corresponding to the marked local segment.

[0014] In conjunction with the first aspect above, in one possible implementation, the step of performing a verification operation on the risk verification and matching records and the target sea state forecast records corresponding to each local segment to obtain the record verification results corresponding to each local segment includes: Any local segment is designated as a marked local segment. If the risk verification matching record corresponding to the marked local segment is the same as the target sea state forecast record, it is determined that the risk verification agent references the information of the marked local segment normally, and the record verification result with normal information reference is obtained. If the risk verification matching record and the target sea state forecast record corresponding to the marked local segment are different, it is determined that the risk verification agent has an error in referencing the information of the marked local segment, and the record verification result with the error in information reference is obtained.

[0015] Secondly, this invention provides a multi-agent cooperative optimization system for large-scale vertical models in the marine domain, the system comprising: The path and record acquisition module is used to acquire candidate paths corresponding to the target marine equipment through the path planning agent, and to acquire the traffic environment forecast records corresponding to each local segment of each candidate path. The acquisition and filtering module is used to acquire sea state forecast records at different water depths in different sea areas of the target ocean through the sea state interpretation agent, and to filter the target sea state forecast records corresponding to each local segment from all sea state forecast records based on the navigation environment forecast records. The acquisition and construction module is used to acquire the risk verification conclusion description record corresponding to each local segment through the risk verification agent, and to acquire the sea state forecast record that the risk verification agent can use when performing risk verification on each local segment, thus forming a set of candidate sea state forecast records corresponding to each local segment. The record filtering module is used to filter out the risk verification matching records corresponding to each local segment from the candidate sea state forecast record set corresponding to each local segment based on the risk verification conclusion description record corresponding to each local segment; The verification operation module is used to verify the risk verification and matching records and target sea state forecast records corresponding to each local segment, and obtain the record verification results for each local segment. The path optimization module is used to optimize the path based on the record verification results of all local segments in all candidate paths corresponding to the target marine equipment.

[0016] The present invention has the following beneficial effects: This invention presents a multi-agent collaborative optimization method for large-scale vertical models in the marine domain. Through verification, it can, to a certain extent, facilitate the identification of data reference misalignments in the risk verification agent, solving the technical problem of poor rationality in the final marine operation plan generation due to poor path optimization. This improves the rationality of path optimization and, consequently, the rationality of the final marine operation plan generation. Specifically, this invention obtains candidate paths corresponding to the target marine equipment through a path planning agent, acquires the traffic environment forecast records corresponding to each local segment of each candidate path, acquires sea state forecast records at different water depths in different sea areas of the target ocean through a sea state interpretation agent, and acquires risk verification conclusion description records corresponding to each local segment through a risk verification agent. By analyzing these records, a verification operation is performed between the risk verification matching records corresponding to each local segment and the target sea state forecast records. This facilitates the identification of data reference misalignments in the risk verification agent and, based on the verification results of all local segments in all candidate paths corresponding to the target marine equipment, path optimization is performed, improving the rationality of path optimization and, consequently, the rationality of the final marine operation plan generation. Attached Figure Description

[0017] Figure 1 This is a flowchart of a multi-agent cooperative optimization method for a large vertical model in the marine domain, according to the present invention. Figure 2 This is a schematic diagram of the composition structure of a multi-agent cooperative optimization system for a large vertical model in the marine field according to the present invention. Figure 3 This is a schematic diagram of the structure of a computer device according to the present invention. Detailed Implementation

[0018] The multi-agent structure framework of this invention can adopt LangChain or AutoGen multi-agent framework.

[0019] refer to Figure 1 This paper illustrates the flowchart of a multi-agent cooperative optimization method for large-scale vertical models in the marine domain, according to the present invention. The method includes the following steps: Step S1: Obtain candidate paths corresponding to the target marine equipment through the path planning agent, and obtain the traffic environment forecast records corresponding to each local segment of each candidate path.

[0020] The path planning agent, also known as the trajectory planning module, is responsible for planning the robot's trajectory based on the required work location, obstacles, and equipment capabilities. It outputs candidate paths, start and end coordinates of local segments, estimated arrival time, and operating water depth. The target marine equipment can be any marine equipment for which operational path planning is required. This equipment primarily consists of underwater robots and automated underwater operation equipment, such as robots performing subsea pipeline inspections, close-in observations, and ship hull cleaning. It can also include unmanned surface vessels (USVs) or special-purpose vessels. The candidate path, also known as the candidate operational path, is a preliminary operational path generated by the path planning agent for the target marine equipment.

[0021] Operations can be specific engineering or maintenance tasks performed by marine equipment in seawater. An operation path can be the spatial trajectory and time schedule of the marine equipment's navigation to complete the operation, often consisting of a starting point, turning points, detection points, and an endpoint connected in chronological order. It can include the estimated entry and exit times of the equipment at each coordinate point, as well as the operating water depth. In practice, when generating an operation path, the path planning agent often outputs a sequence of spatially discrete points (such as starting point, turning points, detection points, and endpoints). Connecting adjacent nodes in this sequence automatically forms a local segment of the continuous trajectory between them. Therefore, an operation path can consist of multiple local segments.

[0022] The navigation environment forecast record can be a navigation environment forecast generated by a path planning agent when the target marine equipment arrives at a local segment. It can include the entry time, departure time, sea area of ​​the segment, water depth of the segment, tidal current velocity, and information version. The entry time, or the expected entry time of the local segment, represents the time when the marine equipment enters the segment. The departure time, or the expected departure time of the local segment, represents the time when the marine equipment leaves the segment. The sea area of ​​the segment can be the sea area where the local segment is located. The water depth of the segment can be the water depth where the local segment is located. The tidal current velocity included in the navigation environment forecast record can be represented by the average tidal current velocity between the expected entry time and expected departure time of the local segment. In practice, ocean current forecasts are often dynamically updated (e.g., morning and evening forecasts), and each update of the ocean current forecast can be recorded as an information version. Each path planning often requires using the latest information version; therefore, the information version included in the navigation environment forecast record can be the latest information version.

[0023] Step S2: Obtain sea state forecast records at different water depths in different sea areas of the target ocean through the sea state interpretation agent. Based on the navigation environment forecast records, select the target sea state forecast records corresponding to each local segment from all sea state forecast records.

[0024] The sea state interpretation agent is an environmental data parsing module. Its function is to analyze ocean forecast data such as tidal currents and wave heights, identify time periods when sea states meet operational conditions, and output the start and end times of tidal current windows, tidal current velocity, water depth, location, and information version. This data constitutes the sea state forecast record. Sea state forecast records are collected in real time, so the collected records often include not only currently collected data but also historical data. Therefore, the information versions included in the sea state forecast records may be the latest or older. A tidal current window can be a continuous time period where the tidal current velocity is below the operational safety threshold. The tidal current velocity within the tidal current window can be characterized by the average tidal current velocity within the window. The operational safety threshold can be the upper limit of the tidal current velocity that allows marine equipment to operate normally. Exceeding this value often leads to loss of equipment attitude control, drifting with the current, or positioning failure. It can be obtained by directly reading from the equipment's physical manual or by calibrating based on historical operation logs and manually verified samples. A common value is 0.25 m / s to 0.4 m / s; for example, 0.25 m / s can be selected for operations with high stability requirements, while 0.4 m / s can be selected for operations with low passage requirements. The target ocean can be the ocean in which the target marine equipment needs to operate.

[0025] As an example, this step may include the following steps: The first step is to obtain sea state forecast records at different water depths in different areas of the target ocean through a sea state interpretation agent.

[0026] The second step is to designate any local segment as a marked local segment, and to designate the corresponding navigation environment forecast record as a marked navigation environment forecast record, and to designate any sea state forecast record as a marked sea state forecast record.

[0027] Third, if the entry time of the marked passage environment forecast record is greater than or equal to the start time of the marked sea state forecast record, and the departure time of the marked passage environment forecast record is less than or equal to the end time of the marked sea state forecast record, and the sea area to which the passage is included in the marked passage environment forecast record is the same as the sea area to which the marked sea state forecast record is included, and the water depth to which the passage is included in the marked passage environment forecast record is the same as the water depth to which the marked sea state forecast record is included, and the information version included in the marked passage environment forecast record is the same as the information version included in the marked sea state forecast record, then the marked sea state forecast record is determined to be the target sea state forecast record corresponding to the marked partial passage.

[0028] Among them, the target sea state forecast record corresponding to the marked local segment can represent the current window of the latest information version that matches the marked local segment.

[0029] It should be noted that if a marked local segment lacks a corresponding target sea state forecast record, it often indicates that the marked local segment lacks a corresponding tidal flow window. This often means that the tidal flow velocity of the marked local segment within the time frame planned by the path planning agent is likely to exceed the operational safety threshold, and the tidal flow velocity is no longer acceptable. In this case, marine equipment cannot perform operations normally, and subsequent verification steps for the marked local segment are usually unnecessary. If the risk verification agent correctly determines that the marked local segment is impassable, then it is usually not necessary to verify the sea state forecast record used for the risk verification agent's determination. However, if the risk verification agent determines that the marked local segment is passable, it can often be directly determined that the risk verification agent may have referenced an incorrect sea state forecast record, such as referencing sea state forecast records from other safe segments. In this case, subsequent verification steps for the marked local segment are usually unnecessary. To improve operational safety, the incorrectly referenced sea state forecast record for the marked local segment can be updated to the correct one, and personnel can be reminded to verify it. This can, to some extent, prevent marine equipment from performing operations in hazardous environments.

[0030] If a corresponding target sea state forecast record exists for a marked local segment, it often indicates that a corresponding tidal current window exists for the marked local segment. This often means that the tidal current velocity for the marked local segment is lower than the operational safety threshold within the time frame planned by the path planning agent. In this case, marine equipment may be able to operate normally. Regardless of whether the risk verification agent determines that the marked local segment is passable or impassable, it is necessary to verify whether the risk verification agent has referenced an incorrect sea state forecast record to avoid misjudgment.

[0031] Step S3: Obtain the risk verification conclusion description record corresponding to each local segment through the risk verification agent, and obtain the sea state forecast record to be used by the risk verification agent when performing risk verification on each local segment, thus forming a set of candidate sea state forecast records corresponding to each local segment.

[0032] The risk verification agent is a safety assessment module. Its function is to conduct risk reviews of candidate paths, output a safety score for the candidate path, a determination of whether it is executable, and the sea state description text cited in the conclusion. The risk verification conclusion description record can be the data record of the current window that the risk verification agent subjectively relies on and uses as the basis for the assessment when generating the safety assessment conclusion. It can be the descriptive parameters extracted from the description text, specifically including: the start time, end time, sea area, water depth, current velocity, and information version of the current window; the current velocity included in the risk verification conclusion description record can be represented by the average current velocity within the current window.

[0033] As an example, this step may include the following steps: The first step is to obtain the risk verification conclusion description record corresponding to each local flight segment through the risk verification agent.

[0034] The second step is to designate any local segment as a marked local segment and to designate the corresponding traffic environment forecast record as a marked traffic environment forecast record.

[0035] The third step is to construct a set of reference sea areas corresponding to the marked local segments by including the sea areas to which the navigation segments belong and their adjacent sea areas in the above-mentioned marked navigation environment forecast records.

[0036] The set of reference sea areas corresponding to the marked local segment may include: the sea area to which the marked local segment belongs, and the sea area adjacent to that sea area.

[0037] The fourth step is to construct a reference depth set corresponding to the marked local water segment by combining the water depth of the water segment to which the above-mentioned navigation environment forecast record belongs and its adjacent water depth.

[0038] For example, if we divide the water into 6 depth layers, and these 6 depth layers are: Layer 1: 0 meters to 5 meters below the surface; Layer 2: 5 meters to 10 meters below the surface; Layer 3: 10 meters to 20 meters below the surface; Layer 4: 20 meters to 30 meters below the surface; Layer 5: 30 meters to 50 meters below the surface; Layer 6: 50 meters below the surface to the seabed, then for Layer 4, the water depth layers adjacent to it can include Layer 3 and Layer 5.

[0039] The fifth step is to combine the information versions included in the above-mentioned marked traffic environment forecast records and their adjacent information versions to form a set of reference information versions corresponding to the above-mentioned marked local flight segments.

[0040] The set of reference information versions corresponding to the marked local navigation segment may include: the latest information version and the information version of the last ocean current forecast.

[0041] The sixth step is to select from all sea state forecast records the sea areas that belong to the aforementioned reference sea area set, the water depth that belongs to the aforementioned reference water depth set, and the information version that belongs to the aforementioned reference information version set, and these records will be used as candidate sea state forecast records corresponding to the aforementioned marked local navigation segments.

[0042] In practice, when referencing sea state forecast records, the risk verification agent may mistakenly refer to sea state forecast records that are adjacent in sea area, water depth, or information version. Therefore, candidate sea state forecast records can be sea state forecast records that the risk verification agent may refer to.

[0043] Step 7: Combine all the candidate sea state forecast records corresponding to the marked local segments to form a set of candidate sea state forecast records corresponding to the marked local segments.

[0044] Step S4: Based on the risk verification conclusion description record corresponding to each local segment, select the risk verification matching record corresponding to each local segment from the candidate sea state forecast record set corresponding to each local segment.

[0045] Among them, the risk verification matching record corresponding to the marked local segment can be the sea state forecast record actually referenced by the risk verification agent when performing risk verification on the marked local segment.

[0046] As an example, this step may include the following steps: The first step is to designate any local segment as a marked local segment, and to designate the risk verification conclusion description record corresponding to the marked local segment as a marked risk verification conclusion description record. Then, to designate any candidate sea state forecast record from the candidate sea state forecast record set corresponding to the marked local segment as a pending sea state forecast record.

[0047] The second step, determining the time difference between the aforementioned risk verification conclusion description record and the aforementioned pending sea state forecast record based on the start and end times included in the above-mentioned risk verification conclusion description record and the aforementioned pending sea state forecast record, may include the following sub-steps: The first sub-step is to determine the absolute value of the difference between the start time included in the above-mentioned risk verification conclusion description record and the start time included in the above-mentioned pending sea state forecast record as the start difference degree between the above-mentioned risk verification conclusion description record and the above-mentioned pending sea state forecast record.

[0048] The second sub-step involves determining the absolute value of the difference between the end time included in the above-mentioned risk verification conclusion description record and the end time included in the above-mentioned pending sea state forecast record as the end difference degree between the above-mentioned risk verification conclusion description record and the above-mentioned pending sea state forecast record.

[0049] The third sub-step involves determining the sum of the aforementioned start difference and the aforementioned end difference as the time difference between the aforementioned risk verification conclusion description record and the aforementioned pending sea state forecast record.

[0050] It should be noted that the greater the time difference between the risk verification conclusion description record and the pending sea state forecast record, the more likely the start and end times of the tidal current window represented by the risk verification conclusion description record and the pending sea state forecast record may be different, and the more likely the tidal current window represented by the risk verification conclusion description record and the pending sea state forecast record are not the same.

[0051] The third step, determining the operational characterization time for the marked local section of the waterway based on whether the target marine equipment needs to perform operations on the marked local section, may include the following sub-steps: In the first sub-step, if the target marine equipment needs to operate on the marked local section of the waterway, the start time of the operation of the target marine equipment on the marked local section of the waterway is determined as the operation representation time corresponding to the marked local section of the waterway.

[0052] The second sub-step is as follows: if the target marine equipment does not need to operate on the marked local section, then half of the sum of the entry time and departure time of the navigation environment forecast record corresponding to the marked local section is determined as the operation characterization time corresponding to the marked local section.

[0053] The fourth step, based on the end time included in the above-mentioned risk verification conclusion description record and the above-mentioned pending sea state forecast record, as well as the above-mentioned operation characterization time, determines the degree of operational difference between the above-mentioned risk verification conclusion description record and the above-mentioned pending sea state forecast record, which may include the following sub-steps: The first sub-step is to determine the difference between the end time included in the above-mentioned risk verification conclusion description record and the operation characterization time corresponding to the above-mentioned marked local flight segment as the remaining operable time corresponding to the above-mentioned risk verification conclusion description record.

[0054] The second sub-step involves determining the difference between the end time included in the aforementioned pending sea state forecast record and the operation characterization time corresponding to the aforementioned marked local segment as the remaining operable time corresponding to the aforementioned pending sea state forecast record.

[0055] The third sub-step is to determine the absolute value of the difference between the remaining workable time corresponding to the above-mentioned risk verification conclusion description record and the remaining workable time corresponding to the above-mentioned pending sea state forecast record as the workability difference degree between the above-mentioned risk verification conclusion description record and the above-mentioned pending sea state forecast record.

[0056] It should be noted that the greater the operational difference between the risk verification conclusion description record and the pending sea state forecast record, the more likely the remaining operable time of the tidal current window represented by the risk verification conclusion description record and the pending sea state forecast record may be different, and the more likely the tidal current window represented by the risk verification conclusion description record and the pending sea state forecast record is not the same.

[0057] The fifth step is to determine the absolute value of the difference between the tidal current velocity included in the above-mentioned risk verification conclusion description record and the tidal current velocity included in the above-mentioned undetermined sea state forecast record as the tidal current velocity difference degree between the above-mentioned risk verification conclusion description record and the above-mentioned undetermined sea state forecast record.

[0058] It should be noted that the greater the difference in tidal current velocity between the risk verification conclusion description record and the pending sea state forecast record, the more likely the tidal current velocity within the tidal current window represented by the risk verification conclusion description record and the pending sea state forecast record may be different, and the more likely the tidal current window represented by the risk verification conclusion description record and the pending sea state forecast record may not be the same.

[0059] Step 6: Based on the above time difference, the above operational difference, and the above tidal current velocity difference, determine whether the above pending sea state forecast record is the risk verification matching record corresponding to the above marked local navigation segment.

[0060] For example, if the aforementioned time difference is less than or equal to a preset time difference threshold, and the aforementioned operational difference is less than or equal to a preset operational difference threshold, and the aforementioned tidal current velocity difference is less than or equal to a preset tidal current velocity difference threshold, then the aforementioned pending sea state forecast record is determined to be the risk verification matching record corresponding to the aforementioned marked local segment; otherwise, the aforementioned pending sea state forecast record is determined not to be the risk verification matching record corresponding to the aforementioned marked local segment.

[0061] The preset time difference threshold can be used to determine whether the start and end times of the tidal flow window match, and it can be determined by the data resolution of the tidal flow forecast. For example, if the time resolution of the tidal flow forecast is 10 minutes, the preset time difference threshold can be half of the resolution, i.e., 5 minutes. The preset work difference threshold can be used to determine whether the remaining workable time of the tidal flow window matches, and it can be equal to half of the preset time difference threshold. The preset tidal flow velocity difference threshold can be used to determine whether the tidal flow velocities of the tidal flow window match, and it is derived from the tidal flow forecast accuracy limit or the historical allowable velocity error statistics. Its value range can be from 0.03 m / s to 0.08 m / s.

[0062] It should be noted that if a target sea state forecast record exists for a marked local waterway segment, but no corresponding risk verification matching record exists, it often indicates that the risk verification agent is actually referencing a record that does not correspond to the current flow window when performing risk verification on the marked local waterway segment. This often suggests that the risk verification agent is referencing incorrect information for the marked local waterway segment. If a target sea state forecast record exists for a marked local waterway segment, and a corresponding risk verification matching record exists, it is often necessary to further determine whether the target sea state forecast record and the risk verification matching record represent the same current flow window.

[0063] Step S5: Perform a verification operation on the risk verification and matching records and target sea state forecast records corresponding to each local segment to obtain the record verification results for each local segment.

[0064] As an example, this step may include the following steps: The first step is to designate any local segment as a marked local segment. If the risk verification matching record and the target sea state forecast record corresponding to the marked local segment are the same, then it is determined that the risk verification agent has a normal reference to the information of the marked local segment, and the record verification result with normal information reference is obtained.

[0065] The second step is to determine that if the risk verification matching record and the target sea state forecast record corresponding to the marked local segment are different, the risk verification agent is found to have made an error in referencing the information of the marked local segment, and the record verification result with the error in information reference is obtained.

[0066] Step S6: Optimize the path based on the record verification results of all local segments in all candidate paths corresponding to the target marine equipment.

[0067] As an example, for all local segments where erroneous information is identified, the data of these local segments is updated to the data of the corresponding current flow window in the risk verification agent. Based on this correct data, the risk verification agent re-verifies the risk of these local segments and re-outputs the risk assessment conclusions and safety recommendations for these local segments. This achieves the updating of the risk assessment conclusions and safety recommendations for these local segments. Based on the updated information, the route selection is re-performed, and finally, the coordination agent synthesizes the operation plan, thus achieving route optimization.

[0068] refer to Figure 2 Based on the same inventive concept as the above-described method embodiments, this invention provides a multi-agent cooperative optimization system for large-scale vertical models in the marine domain. This system may specifically include: The path and record acquisition module 201 is used to acquire candidate paths corresponding to the target marine equipment through a path planning agent, and to acquire the traffic environment forecast records corresponding to each local segment of each candidate path. The acquisition and filtering module 202 is used to acquire sea state forecast records at different water depths in different sea areas of the target ocean through the sea state interpretation agent, and filter the target sea state forecast records corresponding to each local segment from all sea state forecast records based on the navigation environment forecast records. The acquisition and construction module 203 is used to acquire the risk verification conclusion description record corresponding to each local segment through the risk verification agent, and to acquire the sea state forecast record that the risk verification agent can use when performing risk verification on each local segment, thus forming a set of candidate sea state forecast records corresponding to each local segment. The record filtering module 204 is used to filter out the risk verification matching record corresponding to each local segment from the candidate sea state forecast record set corresponding to each local segment based on the risk verification conclusion description record corresponding to each local segment; The verification operation module 205 is used to perform verification operations on the risk verification and matching records and target sea state forecast records corresponding to each local segment, and obtain the record verification results corresponding to each local segment. The path optimization module 206 is used to optimize the path based on the record verification results of all local segments in all candidate paths corresponding to the target marine equipment.

[0069] Figure 3 This is a schematic diagram of the structure of a computer device provided in an embodiment of the present invention. For example, as shown... Figure 3 As shown, the computer device 300 includes: a memory 301, a processor 302, and a computer program 303 stored in the memory 301 and running on the processor 302. When the processor 302 executes the computer program 303, the computer device can execute any of the aforementioned multi-agent cooperative optimization methods for large vertical models in the marine field.

[0070] In summary, this invention obtains candidate paths for the target marine equipment through a path planning agent, acquires traffic environment forecast records for each local segment of each candidate path, obtains sea state forecast records for different sea areas and depths in the target ocean through a sea state interpretation agent, and obtains risk verification conclusion description records for each local segment through a risk verification agent. By analyzing these records, the invention achieves the verification operation between the risk verification matching records for each local segment and the target sea state forecast records. This facilitates the identification of data reference misalignment by the risk verification agent to a certain extent. Furthermore, based on the verification results of the records for all local segments in all candidate paths for the target marine equipment, the invention optimizes the path and improves the rationality of the path optimization, thereby enhancing the rationality of the final marine operation plan.

Claims

1. A multi-agent cooperative optimization method for large-scale vertical models in the marine domain, characterized in that, Includes the following steps: The path planning agent obtains candidate paths for the target marine equipment and acquires the traffic environment forecast records for each local segment of each candidate path. The sea state interpretation agent obtains sea state forecast records at different water depths in different sea areas of the target ocean. Based on the navigation environment forecast records, the target sea state forecast records corresponding to each local segment are selected from all sea state forecast records. The risk verification conclusion description record corresponding to each local segment is obtained through the risk verification agent, and the sea state forecast record that the risk verification agent can use when performing risk verification on each local segment is obtained, thus forming a set of candidate sea state forecast records corresponding to each local segment. Based on the risk verification conclusion description record corresponding to each local segment, the risk verification matching record corresponding to each local segment is selected from the candidate sea state forecast record set corresponding to each local segment; The risk verification and matching records and target sea state forecast records corresponding to each local segment are verified to obtain the record verification results for each local segment. Path optimization is performed based on the record verification results of all local segments in all candidate paths corresponding to the target marine equipment.

2. The multi-agent cooperative optimization method for a large-scale vertical model in the marine domain according to claim 1, characterized in that, The navigation environment forecast record includes the entry time, departure time, sea area, water depth, tidal current velocity, and information version; the sea state forecast record includes the start time, end time, sea area, water depth, tidal current velocity, and information version of the tidal current window; the tidal current window is a continuous period of time in which the tidal current velocity is lower than the operational safety threshold; the risk verification conclusion description record includes the start time, end time, sea area, water depth, tidal current velocity, and information version.

3. The multi-agent cooperative optimization method for a large-scale vertical model in the marine domain according to claim 1, characterized in that, The step of selecting the target sea state forecast record for each local segment from all sea state forecast records based on the navigation environment forecast records includes: Any local segment is designated as a marked local segment, and the corresponding navigation environment forecast record is designated as a marked navigation environment forecast record. Any sea state forecast record is designated as a marked sea state forecast record. If the entry time of the marked passage environment forecast record is greater than or equal to the start time of the marked sea state forecast record, and the departure time of the marked passage environment forecast record is less than or equal to the end time of the marked sea state forecast record, and the sea area to which the passage included in the marked passage environment forecast record belongs is the same as the sea area included in the marked sea state forecast record, and the water depth to which the passage included in the marked passage environment forecast record belongs is the same as the water depth included in the marked sea state forecast record, and the information version included in the marked passage environment forecast record is the same as the information version included in the marked sea state forecast record, then the marked sea state forecast record is determined to be the target sea state forecast record corresponding to the marked local passage.

4. The multi-agent cooperative optimization method for a large-scale vertical model in the marine domain according to claim 1, characterized in that, The sea state forecast records that the risk verification agent selects to use when performing risk verification for each local segment constitute a candidate sea state forecast record set for each local segment, including: Any local flight segment is designated as a marked local flight segment, and the traffic environment forecast record corresponding to the marked local flight segment is designated as a marked traffic environment forecast record; The sea area to which the navigation segment belongs and its adjacent sea areas included in the marked navigation environment forecast record constitute the reference sea area set corresponding to the marked local navigation segment; The water depth to which the navigation segment belongs and its adjacent water depths are included in the marked navigation environment forecast record, which constitutes the reference water depth set corresponding to the marked local navigation segment; The information versions included in the marked passage environment forecast record and their adjacent information versions constitute the reference information version set corresponding to the marked local segment; From all sea state forecast records, select sea state forecast records whose sea area belongs to the reference sea area set, whose water depth belongs to the reference water depth set, and whose information version belongs to the reference information version set, and use them as candidate sea state forecast records corresponding to the marked local segment; All candidate sea state forecast records corresponding to the marked local segment are used to form the candidate sea state forecast record set corresponding to the marked local segment.

5. The multi-agent cooperative optimization method for a large-scale vertical model in the marine domain according to claim 1, characterized in that, The step of selecting risk verification matching records for each local segment from the candidate sea state forecast record set corresponding to each local segment, based on the risk verification conclusion description record corresponding to each local segment, includes: Any local segment is designated as a marked local segment, and the risk verification conclusion description record corresponding to the marked local segment is designated as a marked risk verification conclusion description record. Any candidate sea state forecast record in the candidate sea state forecast record set corresponding to the marked local segment is designated as a pending sea state forecast record. Based on the start and end times included in the marked risk verification conclusion description record and the start and end times included in the pending sea state forecast record, the time difference between the marked risk verification conclusion description record and the pending sea state forecast record is determined. Based on whether the target marine equipment needs to perform operations on the marked local section, determine the operation characterization time corresponding to the marked local section; Based on the end time included in the marked risk verification conclusion description record and the pending sea state forecast record, as well as the operation characterization time, the degree of operation difference between the marked risk verification conclusion description record and the pending sea state forecast record is determined; The absolute value of the difference between the tidal current velocity included in the marked risk verification conclusion description record and the tidal current velocity included in the undetermined sea state forecast record is determined as the tidal current velocity difference degree between the marked risk verification conclusion description record and the undetermined sea state forecast record. Based on the time difference, the operational difference, and the tidal current velocity difference, determine whether the pending sea state forecast record is the risk verification matching record corresponding to the marked local segment.

6. The multi-agent cooperative optimization method for a large-scale vertical model in the marine domain according to claim 5, characterized in that, The step of determining the time difference between the marked risk verification conclusion description record and the pending sea state forecast record based on the start and end times included in the marked risk verification conclusion description record and the pending sea state forecast record includes: The absolute value of the difference between the start time included in the marked risk verification conclusion description record and the start time included in the pending sea state forecast record is determined as the start difference degree between the marked risk verification conclusion description record and the pending sea state forecast record. The absolute value of the difference between the end time included in the marked risk verification conclusion description record and the end time included in the pending sea state forecast record is determined as the end difference degree between the marked risk verification conclusion description record and the pending sea state forecast record. The sum of the initial difference and the final difference is determined as the time difference between the marked risk verification conclusion description record and the pending sea state forecast record.

7. A multi-agent cooperative optimization method for a large-scale vertical model in the marine domain, as described in claim 5, is characterized in that... The step of determining the operation characterization time corresponding to the marked local segment based on whether the target marine equipment needs to perform operations on the marked local segment includes: If the target marine equipment needs to operate on the marked local section of the waterway, then the start time of the operation of the target marine equipment on the marked local section of the waterway shall be determined as the operation characterization time corresponding to the marked local section of the waterway. If the target marine equipment does not need to operate on the marked local segment, then half of the sum of the entry time and departure time of the navigation environment forecast record corresponding to the marked local segment shall be determined as the operation characterization time corresponding to the marked local segment.

8. A multi-agent cooperative optimization method for a large-scale vertical model in the marine domain, as described in claim 5, is characterized in that... The step of determining the operational discrepancy between the marked risk verification conclusion description record and the pending sea state forecast record based on the end time included in the marked risk verification conclusion description record and the pending sea state forecast record, and the operational characterization time, includes: The difference between the end time included in the risk verification conclusion description record and the operation characterization time corresponding to the marked local flight segment is determined as the remaining operable time corresponding to the risk verification conclusion description record. The difference between the end time included in the pending sea state forecast record and the operation characterization time corresponding to the marked local segment is determined as the remaining operable time corresponding to the pending sea state forecast record. The absolute value of the difference between the remaining workable time corresponding to the marked risk verification conclusion description record and the remaining workable time corresponding to the pending sea state forecast record is determined as the workability difference between the marked risk verification conclusion description record and the pending sea state forecast record.

9. A multi-agent cooperative optimization method for a large-scale vertical model in the marine domain according to claim 5, characterized in that, The step of determining whether the pending sea state forecast record is a risk verification matching record corresponding to the marked local navigation segment based on the time difference, the operational difference, and the tidal current velocity difference includes: If the time difference is less than or equal to a preset time difference threshold, and the operation difference is less than or equal to a preset operation difference threshold, and the tidal current velocity difference is less than or equal to a preset tidal current velocity difference threshold, then the pending sea state forecast record is determined to be the risk verification matching record corresponding to the marked local segment; otherwise, the pending sea state forecast record is determined not to be the risk verification matching record corresponding to the marked local segment.

10. A multi-agent cooperative optimization method for a large-scale vertical model in the marine domain according to claim 1, characterized in that, The process of verifying the risk verification and matching records and the target sea state forecast records corresponding to each local segment yields the verification results for each local segment, including: Any local segment is designated as a marked local segment. If the risk verification matching record corresponding to the marked local segment is the same as the target sea state forecast record, it is determined that the risk verification agent references the information of the marked local segment normally, and the record verification result with normal information reference is obtained. If the risk verification matching record and the target sea state forecast record corresponding to the marked local segment are different, it is determined that the risk verification agent has an error in referencing the information of the marked local segment, and the record verification result with the error in information reference is obtained.

Images