Pilot's risk warning method and system for getting on and off a ship in rough sea conditions

Abstract

The application discloses a pilot boarding and leaving a ship risk early warning method and system in severe sea conditions, and relates to the technical field of ship piloting. The method comprises the following steps: obtaining a historical pilot boarding and leaving a ship data set of a target pilot boat and a target piloted ship; identifying an overlapping influence area in different severe sea conditions and a corresponding severe sea condition characteristic set; analyzing real-time sea condition characteristics according to preset motion parameters and preset sea condition environment indexes; extracting a matched overlapping influence area after matching; collecting wave spectrum data in the area and iteratively analyzing the wave spectrum data; and issuing a boarding and leaving a ship risk early warning instruction if the iterative characteristics of the target area wave spectrum do not meet preset requirements. The application solves the technical problem that, in severe sea conditions, traditional boarding and leaving a ship risk monitoring means cannot early warn of risks in the process of pilots boarding and leaving a ship based on related data of an alarm device and a ship, achieves precise early warning of pilot boarding and leaving a ship risks in severe sea conditions, and improves the technical effect of monitoring reliability.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of ship piloting technology, in particular to a pilot boarding and leaving a ship in a severe sea state and a risk warning method and system thereof. BACKGROUND

[0002] In ship piloting operations, pilots need to board and leave a pilot ship in a severe sea state, and the safety of the operation is crucial. In the prior art, traditional boarding and leaving ship risk monitoring methods are mostly based on a single sea state parameter or ship motion data, and rely on simple alarm devices to achieve early warning. However, due to the complex relative motion of the pilot ship and the pilot ship in a severe sea state and the variable wave spectrum characteristics, the traditional method is difficult to comprehensively analyze the dynamic risk in the overlapping influence area, resulting in inaccurate risk warning during the pilot boarding and leaving the ship process, and failing to provide comprehensive and reliable data support for the alarm device, which is difficult to meet the safety requirements of ship piloting operations in severe sea states. SUMMARY

[0003] The present application provides a pilot boarding and leaving a ship risk warning method and system in a severe sea state, which is used to solve the technical problem that traditional boarding and leaving ship risk monitoring methods cannot warn the risk during the pilot boarding and leaving the ship process based on the related data of the alarm device and the ship in a severe sea state.

[0004] In a first aspect, the present application provides a pilot boarding and leaving a ship risk warning method in a severe sea state, the method comprising: obtaining a set of historical pilot boarding and leaving ship data of a target pilot ship and a target pilot ship; identifying the overlapping influence area of the target pilot ship and the target pilot ship in different severe sea states based on the set of historical pilot boarding and leaving ship data, determining a set of overlapping influence areas and a set of corresponding severe sea state characteristics; performing real-time sea state characteristic analysis on the target pilot ship and the target pilot ship according to a preset motion parameter and a preset sea state environment index, and obtaining real-time sea state characteristics; matching the real-time sea state characteristics with the set of severe sea state characteristics, and mapping and extracting the overlapping influence area according to the matching result, and obtaining a matching overlapping influence area; continuously collecting regional wave spectrum data according to the matching overlapping influence area, and performing iterative analysis according to the collection result, and determining a target regional wave spectrum iterative characteristic; determining whether the target regional wave spectrum iterative characteristic meets a preset requirement, and if not, obtaining a boarding and leaving ship risk warning instruction.

[0005] In a second aspect of the present application, a pilot boarding and leaving a ship risk early warning system in severe sea conditions is provided, and the system comprises: a boarding and leaving a ship data set acquisition module, configured to acquire a historical pilot boarding and leaving a ship data set of a target pilot boat and a target guided ship; an overlapping influence area identification module, configured to identify overlapping influence areas of the target pilot boat and the target guided ship in different severe sea conditions based on the historical pilot boarding and leaving a ship data set, determine a set of overlapping influence areas and a corresponding set of severe sea condition characteristics; a real-time sea condition characteristic acquisition module, configured to perform real-time sea condition characteristic analysis on the target pilot boat and the target guided ship according to preset motion parameters and preset sea condition environment indexes, and obtain real-time sea condition characteristics; a matching overlapping influence area acquisition module, configured to match the real-time sea condition characteristics with the set of severe sea condition characteristics, and map and extract the overlapping influence areas according to a matching result, and obtain matching overlapping influence areas; a wave spectrum iteration characteristic acquisition module, configured to continuously acquire regional wave spectrum data according to the matching overlapping influence areas, and perform iteration analysis according to an acquisition result, and determine target regional wave spectrum iteration characteristics; and a wave spectrum iteration characteristic judgment module, configured to judge whether the target regional wave spectrum iteration characteristics meet preset requirements, and if not, obtain a boarding and leaving a ship risk early warning instruction.

[0006] The one or more technical solutions provided in the present application have at least the following technical effects or advantages:

[0007] In the present application, the historical boarding and leaving a ship data of the pilot boat and the guided ship is acquired, the corresponding relationship between the sea conditions and the overlapping influence areas is established through aggregation, mapping and other processes, the real-time sea condition characteristics are matched, the wave spectrum multi-scale iteration analysis is performed, the risk is judged, and the early warning instruction is issued, so that the boarding and leaving a ship risk of the pilot in severe sea conditions is accurately identified, a reliable basis is provided for the alarm device, the boarding and leaving a ship operation safety is ensured, and the technical effect of accurately warning the boarding and leaving a ship risk of the pilot in severe sea conditions and improving the monitoring reliability is achieved. BRIEF DESCRIPTION OF DRAWINGS

[0008] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0009] Figure 1 is a flowchart of the pilot boarding and leaving a ship risk early warning method in severe sea conditions provided by the embodiments of the present application.

[0010] Figure 2 is a structural schematic diagram of the pilot boarding and leaving a ship risk early warning system in severe sea conditions provided by the embodiments of the present application.

[0011] Explanation of reference signs: the data set acquisition module 1 of the boarding and disembarking wheel, the overlapping influence area identification module 2, the real-time sea state feature acquisition module 3, the matching overlapping influence area acquisition module 4, the wave spectrum iterative feature acquisition module 5, and the wave spectrum iterative feature judgment module 6. DETAILED DESCRIPTION

[0012] The application provides a pilot boarding and disembarking wheel risk early warning method and system in severe sea conditions, which is used to solve the technical problem that the traditional boarding and disembarking wheel risk monitoring means cannot early warn the risk in the pilot boarding and disembarking wheel process based on the alarm device and the related data of the ship in severe sea conditions.

[0013] The technical solutions in the embodiments of the application will be clearly and completely described in combination with the accompanying drawings in the embodiments of the application. Obviously, the described embodiments are only part of the embodiments of the application, not all the embodiments. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the application.

[0014] It should be noted that the terms "first", "second", and the like in the specification and the above drawings of the application are used to distinguish similar objects, and do not necessarily indicate a specific order or sequence. It should be understood that the data thus used can be interchanged under appropriate circumstances, so that the embodiments of the application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion, for example, a process, method, system, product or server including a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but can include other steps or modules that are not clearly listed or inherent to these processes, methods, products or devices.

[0015] Embodiment one, as shown in the figure, a pilot boarding and disembarking wheel risk early warning method in severe sea conditions, wherein the method comprises: Figure 1

[0016] Step A100: acquiring a historical pilot boarding and disembarking wheel data set of a target pilot boat and a target guided ship.

[0017] In the embodiments of the application, the pilot boat is a small ship used to pick up and drop off pilots to the guided ship, and undertakes the transportation function of the pilot boarding and disembarking wheel in severe sea conditions. The guided ship is a target ship that accepts the pilot boarding and completes the navigation operation with the help of the pilot.

[0018] ​Specifically, first, the skilled person in the art batch collects original data of the target pilot boat and the target towed ship in the past boarding and leaving operation from the port pilot management system database, the ship automatic identification system (AIS) historical trajectory library and the archives of the maritime meteorological monitoring station. These data cover the pilotage operation records in the past 10 years, each record contains the motion parameters of the pilot boat speed, pilot boat heading, towed ship speed, towed ship heading, etc., and is associated with the sea state environmental indicators such as significant wave height, wave period, wind direction at the corresponding time, forming an original data set.

[0019] Subsequently, the original data is preprocessed in three layers: the first layer synchronizes the motion data of the pilot boat and the towed ship and the sea state indicators to the same time sequence through timestamp alignment technology, with an error of ±5 seconds; the second layer uses the density-based outlier detection algorithm (DBSCAN) to eliminate abnormal records with pilot boat speed greater than 25 knots or wave height data exceeding 1.5 times the historical extreme value of the sea area, and to clean up invalid data; the third layer uses linear interpolation method to fill in the missing wave direction, flow rate and other data, so that the proportion of complete data records is increased to 98%, and finally a structured historical pilot boarding and leaving ship data set is constructed.

[0020] Through multi-source data collection, multi-layer cleaning and preprocessing and classification labeling, a historical data set covering various severe sea state scenarios and containing complete motion parameters and sea state indicators is formed, which lays a precise data foundation for subsequent data-driven overlapping influence area recognition and risk warning model construction, and enables historical data to effectively support the mining and analysis of pilotage operation risk characteristics under different severe sea conditions.

[0021] Step A200: identifying the overlapping influence area of the target pilot boat and the target towed ship under different severe sea conditions based on the historical pilot boarding and leaving ship data set, determining the overlapping influence area set and the corresponding severe sea condition feature set.

[0022] In the embodiments of the present application, the overlapping influence area is the interaction area formed by the relative motion of the target pilot boat and the target towed ship under different severe sea conditions. The severe sea condition is a complex sea condition composed of sea state environmental indicators, as well as the motion parameters of the towed ship speed, heading and the pilot boat speed, heading, etc. When the combination of these parameters reaches the preset risk threshold, it constitutes a severe sea condition scenario that threatens the pilot boarding and leaving ship operation.

[0023] Optionally, based on the historical pilot boarding and leaving ship data set, the historical severe sea condition feature set is extracted with the severe sea condition feature as the index and aggregated according to the preset similarity threshold, the historical overlapping influence area set is extracted with the affected area as the index, and the overlapping influence area set and the corresponding severe sea condition feature set are determined through the mapping of aggregated features and historical areas and the mean value calculation. The specific steps are described in detail in A210-A250.

[0024] Step A300: Real-time sea state feature analysis of the target pilot boat and the target towed ship according to preset motion parameters and preset sea state environment indicators, to obtain real-time sea state features.

[0025] In an embodiment of the present application, first, the motion parameters and sea state environment data of the target pilot boat and the target towed ship are collected in real time by a ship sensor network (including GPS, Doppler current meter, weather station, etc.). Among them, the speed of the pilot boat and the towed ship is collected by the Doppler speed meter at a frequency of 10 Hz, with an accuracy of 0.1 knots; the heading is obtained by the fiber-optic gyroscope, with an accuracy of 1°; the significant wave height is measured by the radar altimeter, with a sampling interval of 5 seconds, and an accuracy of 0.1 meters; the wave period, wave direction and other environmental indicators are transmitted by the weather buoy at an interval of 30 seconds, with an error controlled within ±5%. For example, at a certain moment, the towed ship speed is collected as 12.5 knots, the heading is 240°, the pilot boat speed is 8.3 knots, the heading is 270°, and at the same time, the environmental indicators show that the significant wave height is 2.8 meters, the wave period is 7 seconds, the wind direction is 310°, and the surface flow speed is 1.2 knots.

[0026] The collected data enters the preprocessing step, first, the dynamic parameters such as speed and heading are denoised by Kalman filter to eliminate the instantaneous abnormal values caused by ship rolling. Taking the speed data as an example, the random noise variance is controlled within 0.05 knots² by establishing a state space model. Then, the DBSCAN algorithm is used to cluster the wave data, and the abnormal points with significant wave height exceeding 3 times the historical average are removed, such as the historical significant wave height average of a certain sea area is 1.5 meters, when the data greater than 4.5 meters is collected, it is automatically marked as a noise point. After preprocessing, the data integrity is improved, providing a reliable basis for feature analysis.

[0027] Then, the feature extraction stage is entered. For motion parameters, the relative speed (vector synthesis) and the heading angle of the pilot boat and the towed ship are calculated, for example, the relative speed is calculated by the difference between the two ship speed vectors, and the heading angle is calculated by the cosine theorem. For sea state indicators, the sliding window length of 10 minutes is used to calculate the significant wave height average, wave period median and other statistical quantities, and the wave height time series is converted to frequency domain features by Fourier transform to obtain the energy density distribution. At the same time, the angle between the flow direction and the ship heading, the offset of the surface flow speed to the actual ship track and other parameters are normalized to convert them into feature values within the range of 0-1.

[0028] Finally, the pre-processed motion parameters are fused with the sea state index features to construct a multi-dimensional real-time sea state feature vector. The vector contains 8 dimensions: relative speed, heading angle, significant wave height mean, wave period, wave direction and heading angle, wind direction and wind speed combined vector, surface current speed influence factor, and water depth correction coefficient. For example, when the significant wave height is 2.8 meters and the wave period is 7 seconds, combined with the relative speed of 4.2 knots and the heading angle of 30° between the two ships, the feature vector [4.2, 30, 2.8, 7, 50, 8.5, 0.8, 15.6] is generated, where the wind direction and wind speed combined vector is calculated by wind speed 8.5 m / s and wind direction 310°, the surface current speed influence factor is normalized according to the deviation of 1.2 knot flow speed on the track, and the water depth of 15.6 meters is the real-time measurement value of the operation sea area.

[0029] Through real-time data acquisition of multiple source sensors, joint denoising of Kalman filter and DBSCAN, statistical calculation of feature parameters, and multi-dimensional data fusion, a real-time feature vector containing motion state and sea state environment is constructed, which improves the accuracy of real-time sea state feature analysis, provides high-reliability input data for subsequent matching with historical severe sea state feature set, ensures the accuracy and timeliness of the matching overlap influence area, and lays a data foundation for the risk warning of the pilot boarding and leaving the ship.

[0030] Step A400: matching based on the real-time sea state features and the severe sea state feature set, and mapping and extracting the overlap influence area according to the matching result to obtain the matching overlap influence area.

[0031] Specifically, after the real-time sea state feature construction is completed, the feature matching and area mapping phase is entered. First, an improved K nearest neighbor (KNN) algorithm is used for feature matching, which compares the real-time feature vector with the records in the historical severe sea state feature set. The K value is set to 15, and the similarity is calculated by Euclidean distance, for example, the distance between the real-time feature vector [4.2, 30, 2.8, 7, 50, 8.5, 0.8, 15.6] and a record [4.5, 28, 3.0, 6.8, 47, 9.0, 0.75, 16.0] in the historical record is 0.43. After calculation, multiple historical records with the smallest distance are selected. To improve the matching accuracy, different parameters are assigned weights: the weights of significant wave height and wave period are 0.3, the weights of relative speed and heading angle are 0.2, and the weights of the remaining parameters are 0.1, so that the matching accuracy of the severe sea state feature is improved.

[0032] After the matching is completed, enter the overlapping influence area mapping phase. For the screened multiple historical records, the corresponding overlapping area coordinate sequence is extracted, each sequence containing the relative position change of the pilot boat and the guided ship during the boarding and leaving process, with a sampling interval of 10 seconds. The spatiotemporal consistency verification algorithm is used to align the historical area coordinate sequence with the current ship position. For example, the current relative position is calculated in real time through the ship GPS coordinates, and if the transverse distance between the two ships at a certain time in the historical record is 5 meters, and the current measured value is 4.8 meters, it is considered that the region has spatiotemporal consistency. After verification, the records that do not conform to the current operation scene are removed, and the remaining valid mapping areas are retained.

[0033] Subsequently, the area fusion and optimization are carried out. The remaining valid areas are fused by using the weighted average method, and the weight is determined according to the feature matching similarity. For example, the weight of the record with the highest similarity is 0.15, and the weight of the record with the lowest similarity is 0.05. During the fusion process, Gaussian smoothing processing is performed on the area boundary to eliminate the boundary sawtooth caused by measurement error, so that the area contour is more consistent with the actual fluid mechanics characteristics. At the same time, a water depth correction factor is introduced, such as a current water depth of 15.6 meters corresponding to a correction coefficient of 0.92, which is used to shrink and adjust the overlapping area in shallow water area to avoid misjudgment caused by wave refraction effect.

[0034] Finally, the system generates a matched overlapping influence area. The area is presented in the form of a polygon coordinate sequence, containing multiple key points, each with an accuracy of 0.1 meters. The wave energy density distribution (unit: kW / m²) is labeled in the area, for example, the energy density in the central area is 12 kW / m², and the energy density in the edge area is 5 kW / m², which is visually displayed by color gradient to show the risk level. At the same time, the duration and area change rate of the area are calculated to provide quantitative basis for subsequent risk assessment.

[0035] Through the KNN weighted matching algorithm, accurate comparison between real-time features and historical data is realized. The spatiotemporal consistency verification and Gaussian smoothing fusion technology are used to complete the mapping and extraction of the overlapping area. Combined with the water depth correction and energy density analysis, the matched overlapping influence area with spatiotemporal continuity and physical significance is generated, which improves the accuracy of area extraction and provides a high-precision spatial analysis basis for dynamic assessment of the risk of pilots boarding and leaving the ship in adverse sea conditions.

[0036] Step A500: According to the matched overlapping influence area, the wave spectrum data is continuously collected, and the iterative analysis is carried out according to the collection result to determine the iterative characteristics of the target area wave spectrum.

[0037] In the embodiments of the present application, the wave spectrum is a mathematical function used to describe the distribution characteristics of sea wave energy at different frequencies or wavelengths, i.e. after the random wave of sea wave is decomposed into a series of superimposed different frequency harmonic waves, the distribution graph of the energy density of each frequency component is characterized.

[0038] Specifically, by continuously collecting data on the matching overlapping influence area according to preset wave spectrum indicators such as frequency, energy density, wave height and wave period, an indicator sequence is formed, and then a feature set is obtained through multi-scale identification, and finally the iterative characteristics of the target area wave spectrum are determined through iterative analysis, thereby realizing dynamic monitoring and accurate characterization of wave characteristics in severe sea conditions. The specific steps are described in detail in A510-A530.

[0039] Step A600: Determine whether the target area wave spectrum iterative characteristics meet the preset requirements. If not, obtain the boarding and disembarking risk warning instruction.

[0040] Specifically, first, input the target area wave spectrum iterative characteristics into the preset risk judgment model. The model is trained based on historical accident data, and the preset requirements are determined through machine learning (random forest algorithm, accuracy 90%) of boarding and disembarking risk data in the past ten years, including safety thresholds for each feature: frequency drift rate ≤ ± 0.015 Hz / min, energy density gradient ≤ ± 0.25 kJ / m² / (Hz·min), wave height coefficient of variation ≤ 0.3, etc. The threshold is set as the 95th percentile of historical safe operation data, ensuring that 95% of safe scenarios are covered.

[0041] Subsequently, threshold comparison is performed using the parameter-by-parameter weighted judgment method. Taking the frequency drift rate as an example, if the measured value + 0.02 Hz / min exceeds the safety threshold + 0.015 Hz / min, a single-item warning is triggered; the energy density gradient + 0.3 kJ / m² / (Hz·min) exceeds + 0.25 kJ / m² / (Hz·min), triggering another single-item warning. Each warning item is assigned a weight according to the historical risk contribution, with the frequency drift rate weight being 0.3 and the energy density gradient weight being 0.25. When the weighted risk value exceeds the preset warning line 0.5, comprehensive risk judgment is started.

[0042] If a single or multiple characteristics do not meet the preset requirements, the system calls the warning rule engine to generate different levels of boarding and disembarking risk warning instructions according to the characteristic deviation. For example, if the frequency drift rate exceeds the threshold by 10% and the energy density gradient exceeds the threshold by 20%, it is determined to be a medium risk, and a suggestion to postpone boarding and disembarking is generated. If the wave height coefficient of variation exceeds 0.5 at the same time, it is upgraded to a high risk, and an immediate stop work instruction is generated. The instruction includes risk parameter details, such as frequency drift rate exceeding + 30%, impact analysis, such as short-period wave intensifying ship roll, and operation suggestions, such as adjusting the pilot boat heading to a wave direction of 60° angle, and pushing the instruction to the bridge through the shipboard terminal in the form of audio and visual signals and text pop-up windows.

[0043] The multi-dimensional safety threshold is determined by training historical data, and the feature compliance verification is realized by combining the weighted judgment with the rule engine, thereby effectively solving the problems of high missed judgment rate and insufficient guidance of the traditional single parameter early warning, and providing scientific and timely risk control basis for the pilot boarding and leaving the ship operation in severe sea conditions.

[0044] Further, the step A200 in the method provided by the embodiment of the application comprises:

[0045] A210: traversing the historical pilot boarding and leaving the ship data set to extract data by taking the severe sea condition feature as an index, to obtain a historical severe sea condition feature set.

[0046] A220: performing same-class aggregation on the historical severe sea condition feature set according to a preset feature similarity threshold, to obtain a plurality of aggregated historical severe sea condition feature sets.

[0047] A230: traversing the historical pilot boarding and leaving the ship data set to extract data by taking the affected area as an index, to obtain a historical overlapping affected area set.

[0048] A240: mapping the historical overlapping affected area set based on the plurality of aggregated historical severe sea condition feature sets, and performing mean value calculation on the mapping result, to obtain an overlapping affected area set.

[0049] A250: screening the plurality of aggregated historical severe sea condition feature sets to determine a severe sea condition feature set, wherein each severe sea condition feature corresponds to an aggregated historical severe sea condition feature set.

[0050] Specifically, first, from the historical pilot boarding and leaving the ship data set, taking the severe sea condition features, including the sea condition environment indexes such as significant wave height, wave period, and wind direction, and the motion parameters such as the speed of the guided ship and the heading of the pilot boat as the retrieval keywords, the data time series are synchronized by the time stamp alignment technology with an error of ±5 seconds, the DBSCAN algorithm (density-based spatial clustering application with noise algorithm) is used to eliminate abnormal records such as the pilot boat speed greater than 25 knots, the linear interpolation method is used to fill the missing wave direction data, and finally the historical data containing complete parameters are extracted to form a historical severe sea condition feature set. Different sea condition scenes in the set are classified and labeled, such as strong wind scene (wind speed ≥ 10 levels), huge wave scene (significant wave height ≥ 4 meters), etc., and each record contains the relative position coordinate sequence (sampling interval 10 seconds) of the pilot boat and the guided ship and the duration of the overlapping area and other key information.

[0051] Subsequently, the cosine similarity algorithm is used to calculate the feature vector distance of each data in the historical severe sea state feature set, and a preset feature similarity threshold of 0.8 is set, that is, a 15% parameter fluctuation tolerance interval is allowed, and data with a similarity higher than the threshold is divided into the same class. For example, data with a significant wave height of 3.0±0.5 meters and a wave period of 8±1 seconds are aggregated into the short-period category of a huge wave, forming multiple aggregated historical severe sea state feature sets. This process is automatically processed by the DBSCAN clustering algorithm, and the clustering error rate is controlled within 7%.

[0052] At the same time, the relative position coordinate sequence of the pilot boat and the guided ship during the boarding and disembarking operation is extracted from the historical pilot boarding and disembarking data set with the affected area as the index. Each data record contains a 10-second interval UTC timestamp, pilot boat latitude and longitude with an accuracy of ±0.0001°, and guided ship latitude and longitude. By calculating the convex polygon of the real-time positions of the two ships, an overlapping area polygon coordinate data containing 8-15 vertices is generated.

[0053] Since the original data may come from different coordinate systems, such as some data using UTM projection and some using local Mercator projection, spatial coordinate conversion technology is needed to unify them to the WGS84 geographic coordinate system. The specific steps are as follows: First, identify the original data coordinate system using the coordinate conversion parameter file (which contains the reference ellipsoid parameters and projection parameters of each coordinate system), and then convert the plane coordinates (units: meters) to latitude and longitude coordinates (units: degrees) using the seven-parameter conversion method, with a conversion error of ±0.0002° or less, corresponding to a real distance of ±20 meters. After conversion, Gaussian filtering is performed on all polygon vertex coordinates to remove abnormal points that deviate from the group mean by more than 2 standard deviations, such as jump coordinates caused by sensor failure, to improve the proportion of effective coordinate points.

[0054] Finally, the processed polygon coordinate data is stored in chronological order to form a historical overlapping area set. This set contains overlapping areas recorded during the past decade, each area is labeled with the start time, duration, and corresponding sea state feature label such as "heavy wind and wave" and "rapid current", and all coordinates are expressed in latitude and longitude of the WGS84 coordinate system, ensuring that overlapping areas at different times and in different sea areas are spatially comparable. For example, the overlapping area in a huge wave scenario in January 2024 is a 10-vertex polygon composed of latitude and longitude sequences, with an area of 215 square meters, which is accurately associated with the "significant wave height of 3.5 meters, wave period of 6 seconds" label in the same period sea state feature set.

[0055] Next, the obtained multiple severe sea condition feature sets (such as the aggregated set containing the categories of "huge wave short period", "strong wind long period", etc.) are taken as mapping indexes to associate corresponding data in the historical overlapping influence area set one by one. Taking the category of "huge wave short period" as an example, the set contains thousands of historical records, and each record corresponds to an overlapping area polygon composed of 10-15 latitude and longitude vertices, which has been projected to the WGS84 coordinate system with an accuracy of ±0.0001°.

[0056] For each aggregated sea condition category, the overlapping area polygons of all associated records are extracted, and the area of each polygon is first calculated. The Gauss area formula is used to convert the latitude and longitude coordinates into plane distances for calculation, with an error of less than 1%. For example, the area distribution of the thousands of areas in the "huge wave short period" category is 180-260 square meters. After removing the records with an area less than 150 square meters (abnormal values), the area of the remaining effective areas is arithmetically averaged to obtain the mean area of the overlapping influence area corresponding to the category, which is 220 square meters. At the same time, the mean value of the polygon vertex coordinates is calculated to obtain the geometric center coordinates of the overlapping area of the category, such as longitude 121.5° and latitude 31.2°, which are taken as representative values of the area location.

[0057] All aggregated sea condition categories are processed according to the above steps to finally form an overlapping influence area set containing multiple entries. Each entry contains the name of the corresponding sea condition category, the mean area of the overlapping area (in square meters), the geometric center coordinates (WGS84 latitude and longitude), and the number of associated historical records, etc.

[0058] Finally, the mean shift algorithm is used to extract the center values of the multiple aggregated historical severe sea condition feature sets to filter out the core feature parameters of each set (the specific steps are described in detail in A251), thereby forming a severe sea condition feature set containing multiple typical severe sea condition features. Each feature corresponds to an aggregated historical data set, and each feature allows a parameter tolerance interval of ±5%.

[0059] Through multi-dimensional feature index extraction, bandwidth tolerance similarity aggregation algorithm, spatial mapping and mean calculation, the structured classification of severe sea condition features and overlapping influence areas in historical data is realized, forming a feature set with typicality and fault tolerance, which makes the division of severe sea condition scenarios more consistent with the actual sea condition fluctuation characteristics, and provides a data model support with accuracy and adaptability for subsequent real-time sea condition feature matching and risk area identification, effectively solving the problem of rigid feature division in traditional methods that cannot adapt to sea condition parameter fluctuations.

[0060] Further, the step A250 in the method provided by the embodiment of the application comprises:

[0061] A251: filtering the plurality of aggregated historical severe sea state feature sets using mean shift, determining a center of each set, and aggregating to obtain the severe sea state feature set.

[0062] In the embodiments of the present application, mean shift is a non-parametric clustering and pattern recognition method based on density gradient rise. It is achieved by iteratively calculating the mean vector within the neighborhood of the data point, making the data point drift to the area with higher density, and finally converging to the local density peak.

[0063] Optionally, first, the plurality of aggregated historical severe sea state feature sets obtained by similarity aggregation are taken as input, for example, a data set containing a plurality of aggregated sets, each set storing multi-dimensional feature parameters such as significant wave height, wave period, wind direction, etc.

[0064] Subsequently, the mean shift algorithm is applied to each aggregated historical severe sea state feature set for iterative calculation. Taking three-dimensional features (significant wave height, wave period, wind speed) as an example, the bandwidth parameter h is set to 1.5 times the standard deviation of the significant wave height, such as 0.4 meters; 1.2 times the standard deviation of the wave period, such as 0.8 seconds; and 1.0 times the standard deviation of the wind speed, such as 1.5 knots. For each data point in the set, the vector mean of all points in the neighborhood is calculated, and the data point is iteratively drifted in the direction with the highest density until it converges to the local density peak. For example, in the giant wave short period set, after 15 iterations, all data points converge to a feature center point with a significant wave height of 3.2 meters, a wave period of 7.8 seconds, and a wind speed of 25 knots.

[0065] After completing the mean shift calculation of all aggregated sets, the converged density peak of each set is extracted as the center feature of the set. A corresponding number of center features are obtained by converging a plurality of aggregated sets, and each dimension parameter of each center feature is at the data distribution center position of the corresponding aggregated set. Taking the wave height parameter as an example, the deviation of each center value from the mean value of the corresponding set is less than 5%, ensuring the representativeness of the center feature.

[0066] Finally, these center features are aggregated to form a severe sea state feature set. Each severe sea state feature in the set corresponds to an aggregated historical severe sea state feature set, such as giant wave short period and strong wind long period, and each parameter dimension of each feature is accompanied by a fluctuation tolerance interval based on the mean shift bandwidth, so that the feature can reflect typical sea conditions and adapt to actual fluctuations.

[0067] The density peak value of the aggregated historical sea state features is extracted by the mean shift algorithm, and the optimal representative point of each aggregated set is determined in the multi-dimensional feature space. The originally dispersed massive historical data is converted into a plurality of typical sea state features with a clear center value and a tolerant interval. The data dimension reduction and typical feature extraction are used, and the efficiency and accuracy of subsequent real-time sea state feature matching are significantly improved. A simplified and statistically representative benchmark feature library is provided for the risk warning model.

[0068] Further, the step A300 in the method provided by the embodiment of the application includes:

[0069] The preset motion parameters at least include the towed ship speed, the towed ship heading, the pilot boat speed, and the pilot boat heading.

[0070] Specifically, the towed ship speed refers to the sailing speed (unit: knots) of the towed ship, which is a key parameter for measuring the speed of the ship on the sea. Its role is to affect the relative motion speed of the towed ship and the pilot boat, change the dynamic change frequency of the overlapping area of the two ships, and then affect the stability of the wave spectrum characteristics. For example, when the towed ship speed increases, the water flow disturbance around the ship body intensifies, which may cause the wave energy density in the overlapping area to increase, increasing the risk of boarding and disembarking.

[0071] The towed ship heading refers to the sailing direction (unit: degrees) of the towed ship, which determines the relative angle of the ship and the wave. Its role is that different headings will make the ship body bear different direction wave forces, thereby affecting the shape and area of the overlapping area. For example, when the included angle between the towed ship heading and the wave direction is less than 30°, the ship body is significantly affected by the cross sea, and the fluctuation variance of the wave height of the wave spectrum in the overlapping area may increase, which needs to be monitored.

[0072] The pilot boat speed refers to the speed (unit: knots) of the pilot boat when approaching or moving away from the towed ship, which directly affects the relative stability of the two ships when they are docked. Its role is that too high speed will make it difficult for the pilot boat to control in severe sea conditions, the relative motion trajectory of the overlapping area becomes complex, and the multi-scale identification of the wave spectrum characteristics increases in difficulty; too low speed may prolong the boarding and disembarking time, and increase the time exposed to high-risk sea conditions.

[0073] The pilot boat heading refers to the sailing direction (unit: degrees) of the pilot boat, which determines the path of the pilot boat approaching the towed ship. Its role is that a reasonable heading can reduce the impact of cross sea on the pilot boat and reduce the irregular fluctuation range of the overlapping area. For example, when the pilot boat heading and the wave direction maintain an included angle of 45°, the transverse force of the wave on the ship body can be reduced, the wave period characteristics of the wave spectrum in the overlapping area are more stable, and the risk judgment is facilitated.

[0074] By changing the relative motion state of the two ships in real time, the dynamic change of the overlapping influence area is affected, thereby becoming the core input of real-time sea state feature analysis, and providing kinematic basis for wave spectrum data acquisition and risk warning instruction generation.

[0075] Further, the step A300 in the method provided by the embodiment of the application comprises:

[0076] A320: The preset sea state environment index at least includes significant wave height, wave period, wave direction, wind direction, flow direction, surface flow velocity, and water depth.

[0077] Specifically, the significant wave height refers to the effective wave height, which is the average value of the wave height of the first 1 / 3 highest waves in the statistical period (unit: meter). Its role is to directly reflect the energy intensity of the sea wave. When the significant wave height exceeds 3 meters, the wave impact force in the overlapping influence area of the pilot boat and the guided ship is significantly enhanced, which may cause the relative displacement of the ship during boarding and disembarking to exceed the safety threshold, and is a core parameter for judging the severity of the sea state.

[0078] Wave period: refers to the time interval (unit: second) of adjacent waves passing through a fixed point. Its role is to affect the continuity of the wave and the frequency of the ship's heaving. Short period (such as <8 seconds) waves will make the wave spectrum energy density distribution in the overlapping area more concentrated, and the relative motion amplitude of the pilot boat and the guided ship will be intensified, increasing the difficulty of boarding and disembarking operation.

[0079] Wave direction: refers to the direction of wave propagation (unit: degree). Its role is to determine the angle of the wave acting on the ship. When the angle between the wave direction and the heading of the pilot boat is less than 45°, the cross-sea effect in the overlapping area is significant, which may cause the lateral swing amplitude of the pilot boat to exceed the safety range, and the wave spectrum characteristics in this direction need to be analyzed by the multi-scale identifier.

[0080] Wind direction: refers to the direction of the wind (unit: degree), which affects wave generation together with wind speed. Its role is that strong wind (such as ≥10 levels) will intensify the sea surface fluctuation and change the wave shape in the overlapping area. For example, under the condition of headwind, the forward resistance of the pilot boat increases, and the wave height fluctuation variance of the wave spectrum in the overlapping area may increase by more than 20%, which needs to be iteratively analyzed in combination with the wind speed data.

[0081] Flow direction: refers to the direction of water flow (unit: degree), which affects the actual track of the ship together with the surface flow velocity. Its role is that the angle between the flow direction and the heading of the ship will cause the relative position of the overlapping area to deviate. For example, when the angle between the flow direction and the heading of the pilot boat exceeds 90°, the surface flow velocity will cause the pilot boat to deviate from the predetermined docking area, and the mapping extraction of the overlapping influence area needs to be adjusted through real-time sea state feature matching.

[0082] Surface current speed: refers to the speed of the sea surface current (unit: knots), which affects the maneuverability of the ship and the relative motion trajectory. Its role is that high current speed (such as > 2 knots) will increase the difficulty of the pilot boat approaching the guided ship, causing the duration of the overlapping area to be shortened or lengthened, thereby affecting the integrity of the wave spectrum data collection, and the receptive field parameters need to be dynamically adjusted in multi-scale identification.

[0083] Water depth: refers to the depth of the operation sea area (unit: meters), which affects the wave shape and the safety of the ship's draft. Its role is that shallow water area (such as < 2 times the ship's draft depth) will cause wave refraction and shoaling effect (a physical phenomenon that the wave shape, energy distribution, etc. change when the wave propagates from deep water to shallow water due to the gradual change of water depth), which changes the wave spectrum frequency characteristics of the overlapping area, and the wave height may abnormally increase, and the safety threshold of wave spectrum iterative analysis needs to be corrected through water depth data.

[0084] Through the multi-dimensional coupling effect, the wave spectrum characteristics of the overlapping area are affected, the environmental basic data for real-time sea state feature analysis are provided, and then the dynamic judgment of the wave spectrum iterative characteristics and the generation of the risk warning instruction are supported, forming a three-dimensional monitoring system for severe sea conditions.

[0085] Further, the method provided in the embodiment of the application comprises the following steps:

[0086] A510: continuously collecting data in the matching overlapping influence area according to a preset wave spectrum index to obtain a sequence of regional wave spectrum indexes, wherein the preset wave spectrum at least includes frequency, energy density, wave height and wave period.

[0087] A520: performing multi-scale identification on the sequence of regional wave spectrum indexes to obtain a set of regional wave spectrum characteristics.

[0088] A530: performing iterative analysis on the set of regional wave spectrum characteristics to determine the target regional wave spectrum iterative characteristics.

[0089] Specifically, firstly, wave data is continuously collected at a sampling frequency of 1 Hz using wave height meters, pressure sensors, and ADCP (Acoustic Doppler Current Profiler) deployed in the overlapping influence area. Each collection session lasts for 10 minutes, forming raw time-series data containing frequency, energy density, wave height, and wave period. Frequency describes the periodicity of the wave, typically expressed as the number of vibrations per second, measured in Hertz (Hz). Wave frequency is closely related to wavelength (wave period). Energy density represents the amount of wave energy per unit frequency range; the wave spectrum shows the energy distribution of the wave across different frequencies or periods. Wave height refers to the vertical height of the wave, typically used to describe its intensity. Wave period refers to the time from one wave crest to the next, usually a key characteristic of the wave. For example, one collection yielded 600 data sets, with an average wave height of 2.8 meters, a wave period of 7 seconds, and an energy density peaking at 3.2 kJ / m² / Hz at 0.3 Hz, thus forming the initial regional wave spectrum index sequence.

[0090] After collection, the data enters the multi-scale identification stage. The fluctuation variance of the frequency, energy density, wave height, and wave period of the regional wave spectrum index sequence is calculated and normalized. The identification scale set is determined based on the normalization value. The multi-scale recognizer is constructed using this scale as the receptive field. Then, the index sequence is identified on multiple scales to obtain the regional wave spectrum feature set. The specific steps are explained in detail in A521-A525.

[0091] Subsequently, iterative analysis is performed on the feature set. The regional wave spectrum feature set is serialized according to the identification scale. The initial iterative features are determined by iterating the first and second regional wave spectrum features. Then, the remaining features are iterated based on the initial features to finally obtain the target region wave spectrum iterative features. The specific steps are explained in detail in A531-A533.

[0092] By constructing a multi-dimensional wave spectrum index sequence through high-frequency sensor data acquisition, wavelet transform is used to separate wave features at different time scales, and a recursive algorithm is combined to dynamically track feature parameters, forming an iterative feature vector containing time-frequency domain information and changing trends. This improves the dynamic capture accuracy of wave spectrum features and effectively solves the problem that traditional single-point detection cannot reflect the trend of sea state evolution. It provides predictive wave dynamic characteristics for real-time assessment of boarding and disembarking risks.

[0093] Furthermore, step A520 in the method provided in this application embodiment includes:

[0094] A521: Traverse the regional wave spectrum index sequence to identify the fluctuation variance of each type of regional wave spectrum index, and obtain the frequency fluctuation variance, energy density fluctuation variance, wave height fluctuation variance, and wave period fluctuation variance.

[0095] A522: normalizing the frequency fluctuation variance, the energy density fluctuation variance, the wave height fluctuation variance and the wave period fluctuation variance to obtain a frequency fluctuation variance normalized value, an energy density fluctuation variance normalized value, a wave height fluctuation variance normalized value and a wave period fluctuation variance normalized value.

[0096] A523: determining a recognition scale set according to the size of the frequency fluctuation variance normalized value, the energy density fluctuation variance normalized value, the wave height fluctuation variance normalized value and the wave period fluctuation variance normalized value.

[0097] A524: constructing a multi-scale recognizer by taking the recognition scale in the recognition scale set as the receptive field of a network layer.

[0098] A525: performing multi-scale recognition on the regional wave spectrum index sequence by using the multi-scale recognizer to obtain a regional wave spectrum feature set.

[0099] In the embodiment of the application, the multi-scale recognizer is an intelligent algorithm architecture capable of capturing features of data at different time or space scales simultaneously. The principle is to design processing units with different receptive fields to extract multi-granularity features in data in parallel or in series, thereby more comprehensively describing the dynamic characteristics of a complex system.

[0100] In one embodiment, first, the regional wave spectrum index sequence, for example, 600 groups of sampling data within 10 minutes, is subjected to index-by-index fluctuation variance calculation. Taking the frequency index as an example, the frequency fluctuation variance is obtained by calculating the square mean difference value of adjacent sampling points, which is 0.015 Hz²; after 20 times of window length 10-second sliding window calculation, the energy density fluctuation variance is 0.8 kJ² / m 4 / Hz²; the wave height sequence fluctuation variance is 0.25 m²; and the wave period fluctuation variance is 0.6 s², forming an initial variance set.

[0101] Subsequently, the Min-Max normalization method is used to map each variance to the [0, 1] interval. The frequency fluctuation variance 0.015 Hz² is normalized to 0.75, corresponding to the preset interval 0-0.02 Hz²; the energy density variance 0.8 kJ² / m 4 / Hz² is normalized to 0.4, corresponding to the interval 0-2 kJ² / m 4 / Hz²; the wave height variance 0.25 m² is normalized to 0.5, corresponding to the interval 0-0.5 m²; and the wave period variance 0.6 s² is normalized to 0.6, corresponding to the interval 0-1 s². The normalized values reflect the relative intensity of the fluctuation of each index, for example, the frequency fluctuation normalized value 0.75 indicates that its fluctuation degree is higher than that of other indexes.

[0102] Next, a neural network model is constructed to determine the recognition scale set: the input layer is designed with 4 nodes corresponding to the frequency, energy density, wave height, and wave period fluctuation variance normalized values with a value range of [0, 1], and each node receives a single parameter value after Min-Max normalization processing, such as a frequency fluctuation variance normalized value of 0.75, an energy density normalized value of 0.4, etc. The hidden layer is set to 8 nodes, a ReLU activation function is used to introduce non-linear mapping, and the input data is transformed through a weight matrix W1 (dimension 4x8) and a bias vector b1 to calculate where X is a 4-dimensional input vector. The output layer has 3 nodes corresponding to the large, medium, and small scale categories, and uses a softmax activation function to output a normalized probability distribution, and generates the final output where each element represents the probability of belonging to the corresponding scale category, such as [0.9, 0.08, 0.02] representing a large scale probability of 90%.

[0103] The model training is based on samples extracted from historical severe sea state data in the past ten years, each sample contains 4 normalized variance values and manually labeled scale labels, with large scale accounting for 30%, medium scale for 50%, and small scale for 20%, and the labels use one-hot encoding, such as large scale [1, 0, 0]. During training, the cross-entropy loss function where is the true label, and the Adam optimizer is used with a learning rate of 0.001, with 32 samples as a batch, and iteratively trained for 50 cycles. To prevent overfitting, a dropout layer is added after the hidden layer with a retention rate of 0.8, and the training set and validation set are divided in a ratio of 8:2, and the validation set loss is monitored in real time. After training, the model achieves a classification accuracy of over 90% on an independent test set (2000 samples), for example, input frequency normalized value 0.75, energy density 0.4, wave height 0.5, wave period 0.6, output [0.92, 0.07, 0.01], accurate mapping to large scale encoding [1, 0, 0].

[0104] Next, a multi-scale recognizer is constructed based on the recognized scale, using a convolutional neural network architecture, with specific configurations as shown in Table 1. A large-scale convolution kernel is applied to the frequency sequence to extract long-period fluctuation features, such as low-frequency components below 0.1 Hz; a medium-scale convolution kernel is applied to the energy density sequence to capture medium-frequency energy changes of 0.2-0.5 Hz. For example, in the wave height sequence, the medium-scale recognizer detects that the wave height of 5 consecutive sampling points exceeds 3 meters, generating a high-energy wave peak feature.

[0105] Finally, the multi-scale identifier extracts a set of features from the 600 sets of sampled data (regional wave spectrum indicator sequences): large-scale frequency features, medium-scale energy density features, medium-scale wave height features, medium-scale wave period features, etc., each feature accompanied by a corresponding scale label and confidence level, forming a structured regional wave spectrum feature set.

[0106] By quantifying the fluctuation intensity through the fluctuation variance, achieving cross-parameter comparison through normalization processing, and dynamically matching the recognition scale through the mapping mechanism, combined with the CNN multi-scale convolution architecture, a feature extraction model adaptive to the wave fluctuation characteristics of the sea state is constructed, which improves the multi-scale recognition accuracy of the wave spectrum features, effectively captures the wave change patterns of different time scales, provides feature input with timeliness and integrity for subsequent iterative analysis, and solves the problem that traditional single-scale analysis cannot adapt to the multi-frequency fluctuation of the sea state.

[0107] Table 1: Multi-scale identifier configuration table

[0108] Scale of recognition Receptive field (number of sampling points) Convolution kernel type / size Feature extraction target Activation function Output feature dimension Large scale 20 3x3 convolution kernel (padding 1) Long-period wave energy distribution (≤0.2 Hz) ReLU Low-frequency energy proportion Medium scale 10 2x2 convolution kernel (no padding) Medium-frequency wind wave feature (0.2-0.5 Hz) ReLU Energy density gradient Small scale 5 1x1 convolution kernel High-frequency breaking wave transient change (≥0.5 Hz) Sigmoid Wave height anomaly probability

[0109] Further, step A530 in the method provided by the embodiments of the present application comprises:

[0110] A531: According to the recognition scale size corresponding to each regional wave spectrum feature in the regional wave spectrum feature set, the regional wave spectrum feature set is sequentially processed to obtain a regional wave spectrum feature sequence.

[0111] A532: Iterating the regional wave spectrum feature sequence from the first regional wave spectrum feature and the second regional wave spectrum feature, a first initial regional wave spectrum iterative feature is determined.

[0112] A533: According to the first initial regional wave spectrum iterative feature, the remaining regional wave spectrum features in the regional wave spectrum feature set are iterated to obtain a target regional wave spectrum iterative feature.

[0113] Optionally, first, the regional wave spectrum feature set obtained by multi-scale recognition is sequentially processed. According to the recognition scale size corresponding to each feature, the large scale is preset to be 20, the medium scale is preset to be 10, and the small scale is preset to be 5. The features are sorted by scale from large to small to form a regional wave spectrum feature sequence. For example, the frequency fluctuation feature in a certain set corresponds to a large scale of 20, the energy density and wave height features correspond to a medium scale of 10, and the wave period feature corresponds to a small scale of 5. After sorting, the first two features in the sequence are large-scale frequency features and medium-scale energy density features, followed by medium-scale wave height features, small-scale wave period features, etc.

[0114] After sorting, the first two features (the first and second regional wave spectrum features) are extracted from the sequence for iteration, and the specific steps are described in detail in A532-1.

[0115] Subsequently, the remaining 10 features are iterated one by one based on the first initial iteration feature. The recursive update model is used, and each iteration associates the current iteration feature with the next feature in time sequence, and the weight is adjusted using the exponential smoothing method (smoothing coefficient a = 0.3). For example, the third feature is the mesoscale wave height feature, and the probability of continuous wave peak exceeding 3 meters is 25%. When iterating, the new feature value = a x the previous iteration value 0.272 + (1-a) x 25% = 0.266 is calculated, and the modification of the smoothing coefficient by the feature scale change (mesoscale to mesoscale) is recorded. If the scale change exceeds one level, a is adjusted to 0.4. After multiple iterations, the target area wave spectrum iteration features including the frequency-energy coupling degree, wave height-period correlation trend and other core parameters are finally obtained.

[0116] By sequentially analyzing the features by scale, a time series analysis framework is constructed, the dynamic correlation between features is realized by using weighted fusion and recursive smoothing algorithm, an iteration analysis model with scale dependence is constructed, the time series correlation recognition accuracy of the wave spectrum features is improved, the evolution law of different scale features with time is effectively captured, a predictive feature vector containing multi-scale interaction information is provided for dynamic assessment of the risk of boarding and leaving the ship, and the problem that the traditional independent feature analysis cannot reflect the evolution trend of the sea state is solved.

[0117] Further, step A532 in the method provided in the embodiments of the present application comprises:

[0118] A532-1: performing inner product mapping on the first area wave spectrum feature and the second area wave spectrum feature, performing normalization processing on the inner product mapping result to construct a mapping matrix, and using the mapping matrix to iterate the second area wave spectrum feature to obtain a first initial area wave spectrum iteration feature.

[0119] In one embodiment, first, the aforementioned sequenced first area wave spectrum feature, such as the large-scale frequency feature, is expressed as a vector [frequency fluctuation variance normalized value 0.75, low-frequency energy proportion 32%], and the second area wave spectrum feature, such as the mesoscale energy density feature, is expressed as a vector [energy density fluctuation variance normalized value 0.4, 0.3 Hz energy gradient + 0.2 kJ / m² / Hz·s], which are converted into 2-dimensional feature vectors. The feature correlation degree of the two is calculated by inner product operation, and the formula is , which reflects the synergy of the two features in the fluctuation variance and energy change dimensions.

[0120] Subsequently, all possible feature inner product results are normalized. The inner product value domain is preset to be [-1, 1], and the Min-Max normalization formula , 0.364 is mapped to 0.85 (assuming min = -0.5, max = 0.5), and the normalized result is filled into the initial empty matrix of 2x2, with the row and column being the first and second feature indexes, to construct a mapping matrix , the diagonal of the matrix is the feature self-association value 1, and the non-diagonal is the correlation degree between features.

[0121] Based on the mapping matrix, a graph convolution network (GCN) is constructed, the two features are regarded as nodes in the graph structure, the mapping matrix is regarded as the adjacency matrix, and the node feature vector is the initial input. The network architecture includes 2 convolution layers, the first layer uses the ReLU activation function, the output dimension of the second layer is 1 (the target iteration feature dimension), and the dropout rate is set to 0.5 to prevent overfitting. Taking the input data at a certain moment as an example, after the first layer convolution calculation , wherein is the adjacency matrix with self-loop, is the degree matrix, is the node feature matrix, is the randomly initialized weight matrix. After two convolution operations, the output iteration result is the frequency-energy coupling factor, which comprehensively reflects the synergistic effect strength of large-scale frequency fluctuation and medium-scale energy change.

[0122] By mapping and quantifying the correlation strength through the inner product of the feature vector, a structured mapping matrix is constructed by using normalization processing, and a graph convolution network is combined to realize nonlinear relationship modeling between features, to generate an initial iteration feature containing interaction information in a 2-dimensional feature space, so that the correlation recognition accuracy of multi-scale features is improved to more than 90%, effectively capturing the feature coupling effect that is difficult to represent by traditional weighted fusion methods, providing an initial input containing structural correlation information for subsequent full-feature iteration analysis, and solving the problem that linear fusion cannot model complex dependent relationships between features.

[0123] In summary, the pilot boarding and leaving the ship risk warning method in severe sea conditions provided by the embodiments of the present application has the following technical effects:

[0124] The present application establishes the corresponding relationship between sea conditions and overlapping influence areas by acquiring the historical boarding and leaving the ship data of the pilot boat and the guided ship, processing by aggregation, mapping and the like, combining real-time sea condition feature matching and wave spectrum multi-scale iteration analysis, judging the risk and issuing a warning instruction, thereby accurately identifying the boarding and leaving the ship risk of the pilot in severe sea conditions, making the warning result of the alarm device more reliable, and achieving the technical effects of accurate warning of the boarding and leaving the ship risk of the pilot in severe sea conditions and improving monitoring reliability.

[0125] Embodiment two, as shown in Figure 2 based on the same inventive concept as the foregoing embodiment one, the pilot boarding and leaving the ship risk warning system in severe sea conditions is provided, which comprises:

[0126] A landing and leaving wheel data set acquisition module 1 is configured to acquire a historical pilot landing and leaving wheel data set of a target pilot boat and a target towed ship.

[0127] An overlapping influence area identification module 2 is configured to identify overlapping influence areas of the target pilot boat and the target towed ship under different severe sea conditions based on the historical pilot landing and leaving wheel data set, determine a set of overlapping influence areas and a set of corresponding severe sea condition features.

[0128] A real-time sea condition feature acquisition module 3 is configured to perform real-time sea condition feature analysis on the target pilot boat and the target towed ship according to preset motion parameters and preset sea condition environment indicators, and obtain real-time sea condition features.

[0129] A matching overlapping influence area acquisition module 4 is configured to match the real-time sea condition features with the set of severe sea condition features, and map and extract the overlapping influence areas according to the matching result, and obtain matching overlapping influence areas.

[0130] A wave spectrum iterative feature acquisition module 5 is configured to continuously collect regional wave spectrum data according to the matching overlapping influence areas, and perform iterative analysis according to the collection result, and determine target regional wave spectrum iterative features.

[0131] A wave spectrum iterative feature judgment module 6 is configured to judge whether the target regional wave spectrum iterative features meet preset requirements, and if not, obtain a landing and leaving wheel risk early warning instruction.

[0132] Further, the overlapping influence area identification module 2 is configured to perform the following steps:

[0133] Indexing with severe sea condition features, traversing the historical pilot landing and leaving wheel data set to extract data, and obtaining a set of historical severe sea condition features; performing same-class aggregation on the set of historical severe sea condition features according to a preset feature similarity threshold, and obtaining a plurality of aggregated sets of historical severe sea condition features; indexing with affected areas, traversing the historical pilot landing and leaving wheel data set to extract data, and obtaining a set of historical overlapping influence areas; mapping the set of historical overlapping influence areas based on the plurality of aggregated sets of historical severe sea condition features, and performing mean value calculation on the mapping result, and obtaining a set of overlapping influence areas; filtering the plurality of aggregated sets of historical severe sea condition features, and determining a set of severe sea condition features, wherein each severe sea condition feature corresponds to an aggregated set of historical severe sea condition features.

[0134] Further, the overlapping influence area recognition module 2 is used to perform the following steps:

[0135] The plurality of aggregated historical severe sea state feature sets are screened by using mean shift to determine the center of each set, and the severe sea state feature set is obtained by summarizing.

[0136] Further, the real-time sea state feature acquisition module 3 is used to perform the following steps:

[0137] The preset motion parameters at least include the towed ship speed, the towed ship heading, the pilot boat speed and the pilot boat heading.

[0138] Further, the real-time sea state feature acquisition module 3 is used to perform the following steps:

[0139] The preset sea state environment indicators at least include the significant wave height, the wave period, the wave direction, the wind direction, the flow direction, the surface flow speed and the water depth.

[0140] Further, the wave spectrum iterative feature acquisition module 5 is used to perform the following steps:

[0141] According to the preset wave spectrum indicators, the matching overlapping influence area is continuously collected to obtain a regional wave spectrum indicator sequence, wherein the preset wave spectrum at least includes frequency, energy density, wave height and wave period; the regional wave spectrum indicator sequence is subjected to multi-scale recognition to obtain a regional wave spectrum feature set; and the regional wave spectrum feature set is subjected to iterative analysis to determine the target regional wave spectrum iterative feature.

[0142] Further, the wave spectrum iterative feature acquisition module 5 is used to perform the following steps:

[0143] The regional wave spectrum indicator sequence is traversed to identify the fluctuation variance of each type of regional wave spectrum indicator to obtain the frequency fluctuation variance, the energy density fluctuation variance, the wave height fluctuation variance and the wave period fluctuation variance; the frequency fluctuation variance, the energy density fluctuation variance, the wave height fluctuation variance and the wave period fluctuation variance are subjected to normalization processing to obtain the frequency fluctuation variance normalized value, the energy density fluctuation variance normalized value, the wave height fluctuation variance normalized value and the wave period fluctuation variance normalized value; according to the size of the frequency fluctuation variance normalized value, the energy density fluctuation variance normalized value, the wave height fluctuation variance normalized value and the wave period fluctuation variance normalized value, a recognition scale set is determined; the recognition scale in the recognition scale set is respectively taken as the receptive field of the network layer to construct a multi-scale recognizer; and the multi-scale recognizer is used to perform multi-scale recognition on the regional wave spectrum indicator sequence to obtain a regional wave spectrum feature set.

[0144] Further, the wave spectrum iterative feature acquisition module 5 is used to perform the following steps:

[0145] According to the identification scale size corresponding to each regional wave spectrum feature in the set of regional wave spectrum features, the set of regional wave spectrum features is serialized to obtain a sequence of regional wave spectrum features; the sequence of regional wave spectrum features is iterated from the first regional wave spectrum feature and the second regional wave spectrum feature to determine a first initial regional wave spectrum iterative feature; and the remaining regional wave spectrum features in the set of regional wave spectrum features are iterated according to the first initial regional wave spectrum iterative feature to obtain a target regional wave spectrum iterative feature.

[0146] Further, the wave spectrum iterative feature acquisition module 5 is configured to perform the following steps:

[0147] The first regional wave spectrum feature and the second regional wave spectrum feature are subjected to inner product mapping, and the inner product mapping result is normalized to construct a mapping matrix; and the second regional wave spectrum feature is iterated using the mapping matrix to obtain a first initial regional wave spectrum iterative feature.

[0148] The pilot boarding and leaving ship risk early warning system in severe sea conditions provided by the embodiment of the application can execute the pilot boarding and leaving ship risk early warning method provided by any embodiment of the application, has the corresponding function modules and beneficial effects of the execution method.

[0149] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, however, any number of different modules can be used and run on the user terminal and / or server, and each unit and module included is only divided according to the functional logic, but is not limited to the above division, as long as the corresponding functions can be realized; in addition, the specific name of each functional unit is only for easy mutual differentiation, and does not limit the protection scope of the present application.

[0150] The above specific embodiments do not constitute a limitation on the protection scope of the present application. Those skilled in the art should understand that various modifications, combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent replacement and improvement made within the spirit and principles of the present application should be included in the protection scope of the present application. In some cases, the actions or steps described in the present application can be executed in an order different from that in the embodiments and still achieve the desired results. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multi-task processing and parallel processing are possible or can be advantageous.

Claims

1. A method for early warning of pilot boarding and disembarking risks under adverse sea conditions, characterized in that, The method includes: Acquire the historical data set of pilot boarding and disembarking personnel for the target pilot boat and the target piloted vessel; Based on the historical pilot boarding and disembarking data set, the overlapping influence areas of the target pilot boat and the target guided vessel under different severe sea conditions are identified, and the set of overlapping influence areas and the corresponding set of severe sea condition characteristics are determined. Real-time sea state characteristics of the target pilot boat and the target guided vessel are analyzed according to preset motion parameters and preset sea state environment indicators to obtain real-time sea state characteristics. The real-time sea condition features are matched with the severe sea condition feature set, and the overlapping influence area is mapped and extracted according to the matching results to obtain the matching overlapping influence area. Based on the matching overlapping influence area, regional wave spectrum data is continuously collected, and iterative analysis is performed based on the collection results to determine the wave spectrum iterative characteristics of the target area. Determine whether the wave spectrum iteration characteristics of the target area meet the preset requirements. If not, obtain a risk warning instruction for boarding / disembarking from the ship.

2. The pilot boarding / disembarking risk warning method under adverse sea conditions as described in claim 1, characterized in that, Based on the historical pilot boarding and disembarking data set, the overlapping influence areas of the target pilot boat and the target guided vessel under different severe sea conditions are identified, and the set of overlapping influence areas and the corresponding set of severe sea condition characteristics are determined, including: Using the characteristics of severe sea conditions as an index, the historical data set of pilot boarding and disembarking is traversed to extract data and obtain the set of historical severe sea condition characteristics. According to a preset feature similarity threshold, the historical severe sea condition feature set is aggregated into similar categories to obtain multiple aggregated historical severe sea condition feature sets. Using the affected area as an index, the historical pilot boarding and disembarking data set is traversed to extract data and obtain the set of historical overlapping affected areas. The set of historical overlapping influence areas is mapped based on the multiple aggregated historical severe sea state feature sets, and the mean of the mapping results is calculated to obtain the set of overlapping influence areas. The multiple aggregated historical severe sea condition feature sets are filtered to determine the severe sea condition feature set, wherein each severe sea condition feature corresponds to an aggregated historical severe sea condition feature set.

3. The pilot boarding / disembarking risk warning method under adverse sea conditions as described in claim 2, characterized in that, The mean shift method is used to filter the multiple aggregated historical severe sea state feature sets, determine the center of each set, and summarize to obtain the severe sea state feature set.

4. The pilot boarding / disembarking risk warning method under adverse sea conditions as described in claim 1, characterized in that, The preset motion parameters include at least the speed of the guided vessel, the course of the guided vessel, the speed of the pilot boat, and the course of the pilot boat.

5. The pilot boarding and disembarking risk warning method under adverse sea conditions as described in claim 1, characterized in that, The preset sea state environmental indicators include at least the significant wave height, wave period, wave direction, wind direction, current direction, surface current velocity, and water depth.

6. The pilot boarding / disembarking risk warning method under adverse sea conditions as described in claim 1, characterized in that, Based on the matching and overlapping influence area, regional wave spectrum data is continuously collected, and iterative analysis is performed based on the collection results to determine the wave spectrum iterative characteristics of the target area, including: Data is continuously collected from the matched overlapping influence area according to the preset wave spectrum index to obtain a regional wave spectrum index sequence, wherein the preset wave spectrum includes at least frequency, energy density, wave height and wave period; Multi-scale identification is performed on the regional wave spectrum index sequence to obtain a set of regional wave spectrum features; Iterative analysis is performed on the wave spectrum feature set of the region to determine the wave spectrum iterative features of the target region.

7. The pilot boarding / disembarking risk warning method under adverse sea conditions as described in claim 6, characterized in that, Multi-scale identification is performed on the regional wave spectrum index sequence to obtain a set of regional wave spectrum features, including: The fluctuation variance of each type of regional wave spectrum index is identified by traversing the regional wave spectrum index sequence to obtain the frequency fluctuation variance, energy density fluctuation variance, wave height fluctuation variance, and wave period fluctuation variance. The frequency fluctuation variance, energy density fluctuation variance, wave height fluctuation variance, and wave period fluctuation variance are normalized to obtain normalized values ​​for frequency fluctuation variance, energy density fluctuation variance, wave height fluctuation variance, and wave period fluctuation variance. The set of identification scales is determined based on the normalized values ​​of frequency fluctuation variance, energy density fluctuation variance, wave height fluctuation variance, and wave period fluctuation variance. A multi-scale recognizer is constructed by using the recognition scales in the set of recognition scales as the receptive fields of the network layer. The multi-scale recognizer is used to perform multi-scale recognition on the regional wave spectrum index sequence to obtain a set of regional wave spectrum features.

8. The pilot boarding / disembarking risk warning method under adverse sea conditions as described in claim 7, characterized in that, include: According to the recognition scale size corresponding to each regional wave spectrum feature in the regional wave spectrum feature set, the regional wave spectrum feature set is serialized to obtain the regional wave spectrum feature sequence. The regional wave spectrum feature sequence is iterated from the first regional wave spectrum feature and the second regional wave spectrum feature to determine the first initial regional wave spectrum iterative feature; Based on the first initial regional wave spectrum iterative features, the remaining regional wave spectrum features in the regional wave spectrum feature set are iterated to obtain the target regional wave spectrum iterative features.

9. The pilot boarding / disembarking risk warning method under adverse sea conditions as described in claim 8, characterized in that, The wave spectrum features of the first region and the wave spectrum features of the second region are subjected to inner product mapping, and the inner product mapping results are normalized to construct a mapping matrix. The wave spectrum features of the second region are iterated using the mapping matrix to obtain the wave spectrum iterative features of the first initial region.

10. A pilot boarding / disembarking risk warning system under adverse sea conditions, characterized in that: The system is used to implement the pilot boarding and disembarking risk warning method under adverse sea conditions as described in any one of claims 1-9, the system comprising: The boarding and disembarking data acquisition module is used to acquire historical pilot boarding and disembarking data sets for the target pilot boat and the target piloted vessel. The overlapping influence area identification module identifies the overlapping influence areas of the target pilot boat and the target guided vessel under different severe sea conditions based on the historical pilot boarding and disembarking data set, and determines the set of overlapping influence areas and the corresponding set of severe sea condition features. The real-time sea state feature acquisition module is used to perform real-time sea state feature analysis on the target pilot boat and the target guided vessel according to preset motion parameters and preset sea state environment indicators, and obtain real-time sea state features. The overlapping influence area acquisition module matches the real-time sea state features with the severe sea state feature set, and maps and extracts the overlapping influence area based on the matching results to obtain the matching overlapping influence area. The wave spectrum iterative feature acquisition module is used to continuously collect regional wave spectrum data based on the matching overlapping influence area, and perform iterative analysis based on the collection results to determine the wave spectrum iterative features of the target area. The wave spectrum iteration feature judgment module is used to determine whether the wave spectrum iteration feature of the target area meets the preset requirements. If not, a boarding and disembarking risk warning instruction is obtained.

Images