Ship navigation risk identification and early warning method and system

Abstract

The invention discloses a ship navigation risk identification and early warning method and system, and relates to the technical field of ship navigation risk identification, and the method comprises the following steps: collecting navigation data under historical static data and dynamic data of a current ship, constructing a ship operation situation model, combining the static data and dynamic data of the current ship, and carrying out the early warning of the ship navigation risk. The operation situation of the current ship is predicted, predicted trajectory coordinates are obtained, the conditions of ships around the current ship are obtained through images shot by the unmanned aerial vehicle and mapped to a two-dimensional plane to construct a ship risk prediction map, and the predicted trajectory is projected to the ship risk prediction map in real time; arranging a sensor at the front part of the current ship, collecting water flow data around the current ship through the sensor, and predicting the distribution condition of other ships around the ship risk prediction map. According to the method, the static and dynamic data of the ship and the water flow fluctuation information are integrated, so that surrounding ship situation prediction and risk effective early warning are realized, and the safety of inland river shipping is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of ship navigation risk identification, in particular to a ship navigation risk identification and early warning method and system. BACKGROUND

[0002] As an important part of waterway transportation, inland river shipping plays a key role in regional economic development due to its low cost and large capacity. However, the inland river channel is narrow and the water flow is complex, and the number of ships is large and the types are various, collision accidents occur from time to time, which seriously threatens the safety of life and property and the order of navigation; The traditional ship navigation risk identification and early warning method has many limitations, which depends on the visual observation and experience judgment of the crew, is greatly affected by human factors, and is difficult to fully grasp the dynamic of surrounding ships, especially in low visibility or dense ship areas, it is easy to make mistakes in judgment. At the same time, in the process of inland river shipping, the existing early warning system based on AIS has the problem of incomplete information coverage in the case that most small ships on inland rivers are not equipped with the system, which cannot effectively monitor the ships without AIS signal, resulting in a blind area in collision risk identification, and the problems of low practicality and functionality. SUMMARY

[0003] In view of the problems in the related art, the present application provides a ship navigation risk identification and early warning method and system to overcome the above technical problems existing in the prior art.

[0004] To this end, the specific technical solutions adopted by the present application are as follows: A ship navigation risk identification and early warning method, the method comprising the following steps: S1, collect the navigation data of the current ship under the historical static data and dynamic data, construct a ship operation situation model, combine the static data and dynamic data of the current ship, predict the operation situation of the current ship, obtain the predicted trajectory coordinates, obtain the surrounding ship situation of the current ship through the image shooting of the unmanned aerial vehicle and map it to a two-dimensional plane to construct a ship risk prediction map, and project the predicted trajectory onto the ship risk prediction map in real time; S2, deploy a sensor in the front of the current ship, collect the water flow data around the current ship through the sensor, combine the distribution of other ships around the ship risk prediction map, construct a surrounding ship operation situation prediction model, and determine the risk ship based on the operation situation of the surrounding ships and the heading of the current ship; S3, based on the current ship operation situation model and the surrounding risk ships of the current ship, identify the risk of the current ship operation situation, and combine the operation situation prediction result of the surrounding risk ships to warn the risk of the current ship route.

[0005] As a preferred embodiment, the S1 comprises the following steps: S11, collect static data and dynamic data in the current ship history running process, build a ship running situation model based on a neural network, wherein the static data includes ship length, width, load, and the dynamic data includes ship speed, heading, acceleration, water flow speed, flow direction, water level, and ship position information at different times, collect the current ship history static data and dynamic data, identify and process abnormal values through the Z-score method, divide the data into a training set, a validation set, and a test set according to a 7:2:1 ratio, and train the ship running situation model through LSTM, wherein the ship navigation data and hydrological data are processed through multiple LSTM layers in turn, and finally a fully connected layer is connected to convert the output of the LSTM layer into a predicted ship position; S12, in combination with real-time static data and dynamic data in the current ship route process, predict the running situation of the current ship to obtain a predicted trajectory.

[0006] As a preferred embodiment, the S12 includes the following steps: S121, obtain the load of the current ship through the ship cargo loading and unloading record, collect the ship speed, heading, and acceleration of the current ship in real time through the instrument panel of the current ship, receive hydrological monitoring station information to obtain the hydrological data of the area where the current ship is located, including the water flow speed, flow direction, and water level, record the above data as the navigation static data and dynamic data of the current ship, input the ship running situation model of the current ship, and obtain the predicted trajectory coordinates; S122, take an overhead view of the navigation area by a drone, keep a circular area with a radius R centered at the current ship in the overhead view of the navigation area, identify the ship outline in the circular area through Canny edge detection, map the circular area to a two-dimensional plane in proportion, construct a ship risk prediction map, convert the latitude and longitude coordinates to two-dimensional plane coordinates using geographic information system technology, project the predicted trajectory coordinates onto the ship risk prediction map, and update the ship risk prediction map in real time according to a fixed shooting interval.

[0007] As a preferred embodiment, the S2 includes the following steps: S21, continuously collect the speed vector of the surrounding water flow at different times using the acoustic Doppler current profiler deployed around the current ship, calculate the relative motion through the known position of itself and the position of the surrounding ships, deduce the motion situation of the surrounding ships, and construct a surrounding ship running situation prediction model to predict the position of the surrounding ships; S22, mark the risk ships based on the heading of the current ship and the running situation prediction of the surrounding ships.

[0008] As a preferred embodiment, the S21 includes the following steps: S211. Use an acoustic Doppler current profiler to collect vector data of the water velocity around the ship. The water velocity at time t is... Including speed magnitude With direction The current position of the ship is obtained through the ship risk prediction map. and the position of the i-th surrounding ship. Further calculations are performed to obtain the positions of surrounding ships relative to the current ship. Calculate the relative displacement and estimate the relative velocity within adjacent time intervals, where adjacent time intervals are... Relative displacement within relative speed for ; S212. The relative velocity under the influence of water flow includes the ship's still water velocity and the water flow velocity. The still water velocity is calculated by eliminating the water flow velocity obtained from the acoustic Doppler current profiler. By analyzing the motion patterns of surrounding vessels in still water, the magnitude and direction of velocity changes are obtained, including the following steps: Still water speed Then acceleration ; Direction of still water velocity Then angular velocity ,in These represent the northward and eastward components of the still water velocity, respectively. S213. Use a polynomial to fit the motion trajectories of surrounding ships in still water, and extract the curvature of the trajectory based on the fitted trajectory. As features, the rate of change of speed magnitude, the rate of change of speed direction, and the trajectory curvature are used to construct a feature vector. The Naive Bayes classifier is used for classification and prediction. The prior probability and conditional probability are calculated based on historical data. The posterior probability is calculated for the newly obtained feature vector. The category with the highest posterior probability is selected as the predicted navigation intention direction. S214. Based on the predicted navigation intentions and speed information of surrounding ships at the current time, predict the position of the ship at time T. For the i-th surrounding ship, the position of the ship at future time T is... .

[0009] In a preferred embodiment, S22 includes the following steps: S221. Based on the current vessel's course and the predicted operational status of surrounding vessels, the navigation intention is obtained, and the angle between the current vessel's course and the predicted course intention of surrounding vessels is calculated, including the following steps: Based on the current ship heading vector ,in respectively represent the components of the current ship heading vector on the x-axis and y-axis, combined with the surrounding ship predicted route intended direction wherein respectively represent the components of the surrounding ship predicted route intended direction vector on the x-axis and y-axis, the included angle is calculated , the surrounding ship of is marked as a risk ship, wherein is the included angle threshold.

[0010] As a preferred embodiment, the S3 comprises the following steps: S31, for the determined risk ship, according to the navigation intention prediction result and speed information of the risk ship, the positions of the ship at consecutive N time points are predicted to obtain consecutive position coordinate points, and the consecutive position coordinate points are recorded in the ship risk prediction map; the intersection position is identified by combining the current ship predicted trajectory in the ship risk prediction map and the consecutive position coordinate points of the risk ship, to determine the risk and give an alarm.

[0011] As a preferred embodiment, the S31 comprises the following steps: S311, according to the navigation intention prediction result and speed information of the risk ship, the positions of the ship at consecutive N time points are predicted to obtain consecutive position coordinate points, and the calculated position coordinate at each time point is projected in the ship risk prediction map, and the current ship predicted trajectory coordinates at consecutive N time points are retained; S312, for each consecutive position coordinate point predicted for the risk ship , find the point closest to it in the predicted trajectory of the current ship , and calculate the distance between the two points , wherein when , it represents that the current ship and the risk ship will intersect at time A, which has a collision risk, wherein represents the distance threshold, an audible and visual alarm is given to remind the duty personnel, and the current risk ship is marked in the radar.

[0012] A ship navigation risk identification and early warning system, comprising a data acquisition module, a model construction module, a risk identification and early warning module; The data acquisition module comprises a UAV and an acoustic Doppler current profiler, the UAV is used for shooting an overhead view of a navigation area, the acoustic Doppler current profiler is used for continuously acquiring velocity vectors of water flow around a current ship at different times, the load of the current ship is obtained through ship cargo loading and unloading records, the speed, heading and acceleration of the current ship are acquired in real time through an instrument panel of the current ship, hydrological data of an area where the current ship is located, including water flow velocity, flow direction and water level, are obtained by receiving information of a hydrological monitoring station, static data and dynamic data in a historical operation process of the current ship are collected, the static data include the length, width and load of the ship, and the dynamic data include the speed, heading, acceleration of the ship, water flow velocity, flow direction, water level and position information of the ship at different times, and a ship operation situation model is constructed based on a neural network; The model construction module constructs a ship operation situation model based on the navigation data under the historical static data and dynamic data of the current ship collected by the data acquisition module, obtains the situation of other ships around the current ship through UAV shooting images and maps the situation to a two-dimensional plane to construct a ship risk prediction map, and constructs a surrounding ship operation situation prediction model in combination with the distribution of other surrounding ships on the ship risk prediction map based on the water flow data around the current ship collected by the sensor; The risk identification and early warning module predicts the operation situation of the current ship based on the ship operation situation model of the current ship, obtains predicted trajectory coordinates and projects the predicted trajectory onto the ship risk prediction map in real time, determines risk ships based on the operation situation of the surrounding ships and the heading of the current ship, identifies the motion situation of the current ship based on the ship motion situation model in combination with the risk ships around the current ship, and gives a warning for the risks in the navigation process of the current ship in combination with the operation situation prediction results of the surrounding risk ships.

[0013] The present application has the following advantages: 1. The present application collects the static data and dynamic data of the current ship in the navigation process, predicts the navigation route of the current ship based on the load and inertia of the current ship, draws a risk prediction map in combination with the situation of other ships around the current ship, determines the risk ships based on the water flow fluctuation data around the current ship, and identifies and gives a warning for the risks in combination with the predicted trajectory of the current ship, so that the collision situation with the current ship can be found in advance, a warning is given in time when the collision risk is high, the crew has enough time to take evasive action, and the probability of ship collision accidents is reduced; 2. The present application can realize effective risk identification and early warning in the navigation process of the ship based on the situation prediction of the surrounding ships based on the water flow fluctuation data around the ship, which is helpful for the crew to better plan the navigation path, reduce the collision risk caused by information loss, and enhance the functionality; 3. By predicting the situation of surrounding vessels, this invention allows crew members to plan their navigation routes and speeds more effectively, avoiding frequent evasive maneuvers and unnecessary deceleration, thus improving the navigation efficiency of inland waterway vessels. Predicting the dynamics of surrounding vessels based on water flow fluctuation data also enables better coordination of the navigation order between the current vessel and other vessels, enhancing its practicality. Attached Figure Description

[0014] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0015] Figure 1 This is a flowchart of a method for identifying and warning of ship navigation risks according to an embodiment of the present invention.

[0016] Figure 2 This is a block diagram of a ship navigation risk identification and early warning system according to an embodiment of the present invention. Detailed Implementation

[0017] To further illustrate the various embodiments, the present invention provides accompanying drawings, which are part of the disclosure of the present invention. These drawings are mainly used to illustrate the embodiments and can be used in conjunction with the relevant descriptions in the specification to explain the operating principles of the embodiments. With reference to these drawings, those skilled in the art should be able to understand other possible implementation methods and the advantages of the present invention. The components in the drawings are not drawn to scale, and similar component symbols are generally used to represent similar components.

[0018] According to embodiments of the present invention, a method and system for identifying and warning of ship navigation risks are provided.

[0019] The present invention will now be further described in conjunction with the accompanying drawings and specific embodiments: Example 1: As Figure 1 As shown, a method for identifying and warning of ship navigation risks according to an embodiment of the present invention includes the following steps: S1. Collect navigation data based on historical static and dynamic data of the current vessel, construct a vessel operation status model, combine the current vessel's static and dynamic data to predict the current vessel's operation status, obtain the predicted trajectory coordinates, obtain the surrounding vessel situation through images captured by drones and map them onto a two-dimensional plane to construct a vessel risk prediction map, and project the predicted trajectory onto the vessel risk prediction map in real time. S11, collect static data and dynamic data in the current ship historical running process, build a ship running situation model based on a neural network, wherein the static data includes ship length, width, load, and the dynamic data includes ship speed, heading, acceleration, water flow speed, flow direction, water level, and ship position information at different times, collect the current ship historical static data and dynamic data, identify and process outliers through the Z-score method, divide the data into a training set, a validation set, and a test set according to a 7:2:1 ratio, and train the ship running situation model through LSTM, wherein the ship navigation data and hydrological data are sequentially processed through multiple LSTM layers, and finally a fully connected layer is connected to convert the output of the LSTM layer into a predicted ship position; It should be noted that the ship navigation data includes ship speed, heading, acceleration, and the hydrological data includes water flow speed, flow direction, and water level. When training the LSTM, the mean square error is used as the loss function to measure the error between the predicted ship position of the model and the actual historical data. The Adam optimizer is used to adaptively adjust the learning rate of each parameter. The training set data is input into the model in batches to calculate the predicted value through forward propagation. The error between the predicted value and the actual value is calculated according to the loss function, and the gradient is calculated through back propagation. The optimizer is used to update the parameters of the model. The iteration is repeated until the preset training number is reached, and the ship running situation model of the current ship is obtained.

[0020] S12, predict the running situation of the current ship by combining real-time static data and dynamic data in the current ship route, and obtain a predicted trajectory; S121, obtain the load of the current ship through the ship cargo loading and unloading record, collect the ship speed, heading, and acceleration of the current ship in real time through the instrument panel of the current ship, receive hydrological monitoring station information to obtain the hydrological data of the area where the current ship is located, including water flow speed, flow direction, and water level, record the above data as the navigation static data and dynamic data of the current ship, input the ship running situation model of the current ship, and obtain the predicted trajectory coordinates; S122, take a top view of the navigation area by a drone, keep a circular area with a radius R centered at the current ship in the top view of the navigation area, identify the ship outline in the circular area through Canny edge detection, map the circular area to a two-dimensional plane in proportion, construct a ship risk prediction map, convert the latitude and longitude coordinates to two-dimensional plane coordinates using geographic information system technology, project the predicted trajectory coordinates onto the ship risk prediction map, and update the ship risk prediction map in real time according to a fixed shooting interval.

[0021] It should be noted that the radius R needs to be set empirically based on the parameters of the current drone camera and the size of the flight area. The shooting interval is usually set to 10 seconds, but it can also be adjusted according to the actual situation. The coordinate transformation is implemented using the pyproj library of Python. By creating a binary mask with the same size as the image, a circular area is drawn with the center coordinates of the current ship in the image as the center and the radius as R. The pixel values ​​in the circular area are set to 1, and the rest are set to 0. The image of the circular area with the current ship as the center and the radius as R is extracted. A blank two-dimensional image is created based on the range of the mapped two-dimensional circular area as the basis for the ship risk prediction map. The ship outline and predicted trajectory corresponding to the two-dimensional plane coordinates are drawn on the blank two-dimensional image to obtain the ship risk prediction map.

[0022] Example 2: S2. Deploy sensors at the front of the current vessel to collect water flow data around the current vessel. Combine this with the distribution of other vessels around the vessel on the vessel risk prediction map to construct a prediction model of the surrounding vessel's operating status. Based on the operating status of the surrounding vessels and the current vessel's course, determine the risky vessels. S21. Using an acoustic Doppler current profiler deployed around the current vessel, continuously collect the velocity vector of the surrounding water flow at different times. Calculate the relative motion using the known position of the vessel itself and the positions of the surrounding vessels, deduce the motion state of the surrounding vessels, and construct a prediction model of the surrounding vessel's operating state to predict the position of the surrounding vessels. S211. Use an acoustic Doppler current profiler to collect vector data of the water velocity around the ship. The water velocity at time t is... Including speed magnitude With direction The current position of the ship is obtained through the ship risk prediction map. and the position of the i-th surrounding ship. Further calculations are performed to obtain the positions of surrounding ships relative to the current ship. Calculate the relative displacement and estimate the relative velocity within adjacent time intervals, where adjacent time intervals are... Relative displacement within relative speed for ; S212. The relative velocity under the influence of water flow includes the ship's still water velocity and the water flow velocity. The still water velocity is calculated by eliminating the water flow velocity obtained from the acoustic Doppler current profiler. By analyzing the motion patterns of surrounding vessels in still water, the magnitude and direction of velocity changes are obtained, including the following steps: Still water speed Then acceleration ; Direction of still water velocity Then the angular velocity where and represent the north and east components of the static water velocity respectively; It should be noted that the static water velocity is calculated and analyzed, when continuously increases, it means that the ship is accelerating, continuously decreases, it means that the ship is decelerating, and the analysis changes over time, if the rate of change is large, it means that the ship is turning.

[0023] S213, using a polynomial to fit the motion trajectory of the surrounding ship in the static water, extracting the curvature of the trajectory according to the fitted trajectory as a feature, the rate of change of the speed, the rate of change of the direction of the speed, and the curvature of the trajectory are used as features to construct a feature vector, which is classified and predicted by a naive Bayes classifier, the prior probability and the conditional probability are calculated according to the historical data, the posterior probability is calculated for the newly obtained feature vector, and the class with the maximum posterior probability is selected as the predicted sailing intention direction; It should be noted that the position data of the ship is fitted using a quadratic polynomial, and the coefficients are solved by the least squares method, so that the sum of the squared errors of the fitted curve and the actual trajectory is minimized. When the naive Bayes classifier is used for classification and prediction, it is assumed that there are m categories of sailing intentions , the prior probability and the conditional probability are calculated according to the historical data, where is a feature vector constructed by taking the rate of change of the speed, the rate of change of the direction of the speed, and the curvature of the trajectory as features, for a new feature vector , the posterior probability is calculated according to Bayes' formula.

[0024] S214, according to the sailing intention prediction result and the speed information of the surrounding ships at the current time, the position of the ship at time T is predicted, and for the i-th surrounding ship, the position of the ship at future time T is .

[0025] S22, based on the current ship's heading and the running situation prediction of the surrounding ships, mark the risk ships; S221, based on the current ship's heading and the running situation prediction of the surrounding ships, obtain the sailing intention, calculate the included angle between the current ship's heading and the predicted sailing intention direction of the surrounding ships, including the following steps: According to the current ship's heading vector , where represent the components of the current ship's heading vector on the x-axis and y-axis respectively, and the predicted sailing intention direction of the surrounding ships , where respectively represent the components of the surrounding ship's predicted course intention direction vector on the x-axis and the y-axis, and the included angle is calculated , the surrounding ship is marked as a risk ship, wherein is the included angle threshold value.

[0026] It should be noted that is the included angle threshold value, which is usually set to 30°, and can also be adjusted according to actual needs, when the included angle is less than the included angle threshold value, it represents that the current ship and the surrounding ship have a collision risk.

[0027] S3, based on the current ship motion situation model, the risk of the current ship motion situation is identified in combination with the surrounding risk ships of the current ship, and the risk in the navigation process of the current ship is warned in combination with the running situation prediction result of the surrounding risk ships. S31, for the determined risk ship, the position of the ship at consecutive N time points is predicted according to the navigation intention prediction result and the speed information of the risk ship to obtain consecutive position coordinate points, and the consecutive position coordinate points are recorded in the ship risk prediction map, the intersection position is identified to determine the risk and give an alarm in combination with the current ship prediction track in the ship risk prediction map and the consecutive position coordinate points of the risk ship. S311, according to the navigation intention prediction result and the speed information of the risk ship, the position of the ship at consecutive N time points is predicted to obtain consecutive position coordinate points, and each time position coordinate calculated is projected in the ship risk prediction map, and the current ship prediction track coordinates at consecutive N time points are retained. S312, for each consecutive position coordinate point of the risk ship predicted , find the point closest to it in the prediction track of the current ship , and calculate the distance between the two points , wherein , when , it represents that the current ship and the risk ship will intersect at time A and have a collision risk, wherein represents the distance threshold value, an audible and visual alarm is given to remind the duty personnel, and the current risk ship is marked in the radar.

[0028] It should be noted that represents the distance threshold value, which needs to be set in combination with the current navigation area, and the distance threshold value can be set smaller in narrow channels and appropriately increased in open sea areas.

[0029] Embodiment 3: as shown in Figure 2 , a ship navigation risk identification and warning system, comprising a data acquisition module, a model construction module, a risk identification and warning module. The data acquisition module comprises a UAV and an acoustic Doppler current profiler, the UAV is used for shooting an overhead view of a navigation area, the acoustic Doppler current profiler is used for continuously collecting velocity vectors of water flow around a current ship at different times, the load of the current ship is obtained through ship cargo loading and unloading records, the speed, heading and acceleration of the current ship are collected in real time through an instrument panel of the current ship, hydrological data of an area where the current ship is located, including water flow velocity, flow direction and water level, are obtained by receiving information of a hydrological monitoring station, meanwhile, static data and dynamic data in a historical operation process of the current ship are collected, the static data comprises the length, width and load of the ship, and the dynamic data comprises the speed, heading, acceleration of the ship, water flow velocity, flow direction, water level and position information of the ship at different times, and a ship operation situation model is constructed based on a neural network; The model construction module is used for constructing a ship operation situation model through navigation data under the historical static data and dynamic data of the current ship based on the data collected by the data acquisition module, obtaining the situation of other ships around the current ship through images shot by the UAV and mapping the situation to a two-dimensional plane to construct a ship risk prediction map, and constructing a surrounding ship operation situation prediction model in combination with the distribution of other surrounding ships on the ship risk prediction map based on water flow data around the current ship collected by sensors; The risk identification and early warning module is used for predicting the operation situation of the current ship based on the ship operation situation model of the current ship, projecting a predicted trajectory to the ship risk prediction map in real time, determining a risk ship based on the operation situation of the surrounding ships and the heading of the current ship, identifying the motion situation of the current ship based on the ship motion situation model in combination with the risk ships around the current ship, and giving a warning to the risk of the current ship during navigation in combination with the prediction results of the operation situation of the surrounding risk ships In conclusion, the current ship static data and dynamic data during navigation are collected, the navigation route of the current ship is predicted based on the load and inertia of the current ship, the risk prediction map is drawn in combination with the situation of other ships around the current ship, the risk ship is determined based on the water flow fluctuation data around the current ship, and the risk identification and early warning are performed in combination with the predicted trajectory of the current ship, so that the collision with the current ship can be found in advance, a warning is given in time when the collision risk is high, the crew has enough time to take evasive action, and thus the probability of ship collision accidents is reduced; The situation of the surrounding ships is predicted based on the water flow fluctuation data around the ship, so that the effective risk identification and early warning during ship navigation can be realized in the case that the river is small and there are many small ships, and most of the ships are not equipped with AIS systems, which helps the crew to better plan the navigation path, reduces the collision risk caused by information loss, and enhances the functionality; Through the prediction of the surrounding ship situation, the crew can conveniently plan the sailing route and speed, avoid frequent avoidance operation and unnecessary deceleration, improve the navigation efficiency of the inland river ship, and better coordinate the navigation order of the current ship and other ships according to the flow fluctuation data to predict the dynamic of the surrounding ships, and enhance the practicability.

[0030] The above merely describes preferred embodiments of the present application and is not intended to limit the present application, and any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method for ship navigation risk identification and early warning, characterized in that, The method comprises the following steps: S1, collecting the sailing data of the current ship under the historical static data and dynamic data, constructing a ship running situation model, predicting the running situation of the current ship in combination with the static data and dynamic data of the current ship, obtaining the predicted trajectory coordinates, obtaining the surrounding ship situation of the current ship through the image shooting of the unmanned aerial vehicle and mapping it to a two-dimensional plane to construct a ship risk prediction map, and projecting the predicted trajectory onto the ship risk prediction map in real time; S2, deploying sensors in the front of the current ship, collecting the surrounding water flow data of the current ship through the sensors, constructing a surrounding ship running situation prediction model in combination with the distribution of other surrounding ships on the ship risk prediction map, and determining the risk ships based on the surrounding ship running situation and the heading of the current ship; S3, risk identification of the motion situation of the current ship based on the current ship motion situation model and the surrounding risk ships of the current ship, and early warning of the risks in the route of the current ship in combination with the running situation prediction results of the surrounding risk ships.

2. The method of claim 1, wherein, The S1 comprises the following steps: S11, collecting the static data and dynamic data of the current ship in the historical running process, constructing a ship running situation model based on a neural network, wherein the static data includes the length, width and load of the ship, and the dynamic data includes the ship speed, heading, acceleration, water flow speed, flow direction, water level and position information of the ship at different times, collecting the historical static data and dynamic data of the current ship, identifying and processing abnormal values through the Z-score method, dividing the data into a training set, a validation set and a test set according to a ratio of 7:2:1, and training the ship running situation model through LSTM, wherein the ship navigation data and hydrological data are processed through multiple LSTM layers in sequence, and finally a fully connected layer is connected to convert the output of the LSTM layer into the predicted ship position; S12, predicting the running situation of the current ship in combination with the real-time static data and dynamic data in the route of the current ship, and obtaining the predicted trajectory.

3. The method of claim 2, wherein, The S12 comprises the following steps: S121, obtaining the load of the current ship through the ship cargo loading and unloading record, collecting the ship speed, heading and acceleration of the current ship in real time through the instrument panel of the current ship, receiving the hydrological monitoring station information to obtain the hydrological data of the area where the current ship is located, including the water flow speed, flow direction and water level, recording the above data as the sailing static data and dynamic data of the current ship, inputting the ship running situation model of the current ship, and obtaining the predicted trajectory coordinates; S122, shooting the overhead view of the sailing area through the unmanned aerial vehicle, reserving a circular area with a radius R in the overhead view of the sailing area with the current ship as the center, identifying the ship contour in the circular area through Canny edge detection, mapping the circular area to a two-dimensional plane in proportion, constructing a ship risk prediction map, converting the latitude and longitude coordinates into two-dimensional plane coordinates using geographic information system technology, projecting the predicted trajectory coordinates onto the ship risk prediction map, and updating the ship risk prediction map in real time according to a fixed shooting interval.

4. The method of claim 1, wherein, The S2 comprises the following steps: S21, continuously collecting velocity vectors of surrounding water flow at different times using acoustic Doppler current profiler deployed around the current ship, calculating relative motion through known self-position and position of surrounding ships, deducing motion trend of surrounding ships, and constructing a surrounding ship motion trend prediction model to predict the position of surrounding ships; S22, marking a risk ship based on the heading of the current ship and the motion trend prediction of the surrounding ships.

5. The method of claim 4, wherein, The S21 includes the following steps: S211. Use an acoustic Doppler current profiler to collect vector data of the water velocity around the ship. The water velocity at time t is... Including speed magnitude With direction The current position of the ship is obtained through the ship risk prediction map. and the position of the i-th surrounding ship. Further calculations are performed to obtain the positions of surrounding ships relative to the current ship. Calculate the relative displacement and estimate the relative velocity within adjacent time intervals, where adjacent time intervals are... Relative displacement within relative speed for ; S212, the relative speed under the influence of the water flow contains the static water speed of the ship and the water flow speed, the water flow speed obtained by eliminating the acoustic Doppler current profiler is used to calculate the static water speed , the speed size change and the speed direction change are obtained by analyzing the surrounding ship movement mode through the static water speed, including the following steps: The magnitude of the static water velocity The acceleration ; direction of the current velocity then the angular velocity where and represent the north and east components of the current velocity, respectively; S213, using polynomial fitting the motion trajectory of the surrounding ships in still water, extracting the curvature of the trajectory according to the fitted trajectory As a feature, the rate of change of speed, the rate of change of speed direction, the trajectory curvature are taken as the features to construct the feature vector, the classification prediction is carried out through the naive Bayes classifier, the prior probability and the conditional probability are calculated according to the historical data, the posterior probability is calculated for the newly obtained feature vector, and the class with the maximum posterior probability is selected as the predicted sailing intention direction. S214, according to the sailing intention prediction result and speed information of the surrounding ships around the current time, the position of the ship at time T is predicted, and for the i th surrounding ship, the ship position at future time T is .

6. The method of claim 5, wherein, The S22 includes the following steps: S221, obtaining a sailing intention based on the heading of the current ship and the motion trend prediction of the surrounding ships, and calculating the included angle between the heading of the current ship and the predicted sailing intention of the surrounding ships, including the following steps: Based on the current ship heading vector ,in These represent the components of the current ship's heading vector on the x and y axes, respectively, which, combined with the surrounding ships, predict the intended course direction. ,in These represent the components of the predicted course direction vectors of surrounding vessels on the x and y axes, respectively. The included angle is then calculated. ,Will The surrounding vessels are marked as risk vessels, among which The included angle threshold.

7. The method of claim 6, wherein, The S3 includes the following steps: S31, for the determined risk ship, predicting the position of the ship at consecutive N times according to the sailing intention prediction result and speed information of the risk ship to obtain consecutive position coordinate points, recording the consecutive position coordinate points in the ship risk prediction map, and identifying the intersection position by combining the predicted trajectory of the current ship in the ship risk prediction map with the consecutive position coordinate points of the risk ship to determine the risk and issue an alarm.

8. The method of claim 1, wherein, The S31 includes the following steps: S311, predicting the position of the ship at consecutive N times according to the sailing intention prediction result and speed information of the risk ship to obtain consecutive position coordinate points, projecting the calculated position coordinate at each time on the ship risk prediction map, and retaining the predicted trajectory coordinate of the current ship at consecutive N times; S312, for each continuous position coordinate point obtained by the risk ship prediction find the point closest to the current ship in the predicted trajectory of the current ship and calculate the distance between the two points wherein when , it represents that the current ship and the risk ship will intersect at time A, which has a collision risk, wherein represents the distance threshold, an audible and light alarm is issued to remind the duty personnel, and the current risk ship is marked in the radar.

9. A ship navigation risk identification and early warning system, characterized in that, The system adopts a ship sailing risk identification and early warning method according to any one of claims 1-8, comprising a data acquisition module, a model construction module, and a risk identification and early warning module. The data acquisition module comprises a UAV and an acoustic Doppler current profiler, takes an overhead view of the sailing area by the UAV, continuously collects velocity vectors of surrounding water flow at different times of the current ship by the acoustic Doppler current profiler, obtains the load of the current ship through ship cargo loading and unloading records, collects the ship speed, heading, and acceleration of the current ship in real time through the instrument panel of the current ship, receives hydrological monitoring station information to obtain hydrological data of the area where the current ship is located, including water flow speed, flow direction, and water level, collects static data and dynamic data in the historical operation process of the current ship, constructs a ship motion trend model based on neural network, wherein the static data includes ship length, width, and load, and the dynamic data includes ship speed, heading, acceleration, water flow speed, flow direction, water level, and position information of the ship at different times; The model construction module constructs a ship motion trend model based on the data collected by the data acquisition module through sailing data under the historical static data and dynamic data of the current ship, obtains the situation of surrounding ships of the current ship through UAV image shooting and mapping to a two-dimensional plane to construct a ship risk prediction map, and constructs a surrounding ship motion trend prediction model through surrounding water flow data of the current ship collected by sensors in combination with the distribution of other surrounding ships on the ship risk prediction map. The risk identification and early warning module predicts the operation situation of the current ship through a current ship operation situation model, obtains predicted trajectory coordinates and projects the predicted trajectory onto a ship risk prediction map in real time, determines risk ships based on the operation situation of surrounding ships and the heading of the current ship, identifies the risk of the operation situation of the current ship based on the current ship operation situation model and the risk ships around the current ship, and early warns the risks in the navigation process of the current ship based on the operation situation prediction results of the surrounding risk ships.

Images