Ship collision early warning system based on multi-source data

Abstract

The invention discloses a ship collision early warning system based on multi-source data, and belongs to the technical field of collision early warning. The system comprises a detection system, a prediction system, an analysis system and an early warning system. The detection system is used for obtaining a collision risk area based on ship driving information; the prediction system is used for predicting a future path of an obstacle in the collision risk area; the analysis system is used for carrying out risk level calculation based on the future path and planning the future path of the ship to obtain an obstacle avoidance path; the early warning system is used for carrying out collision early warning based on the risk level. According to the invention, future path prediction and collision grade calculation are carried out on obstacles which may collide in advance, the obstacle avoidance path of the ship is planned by taking the future path as an obstacle avoidance point, and meanwhile, an existing obstacle avoidance path planning method is optimized and improved to obtain the obstacle avoidance path. Static or dynamic obstacles can be flexibly avoided to a greater extent, and accidents are avoided.

Description

Technical Field

[0001] The invention belongs to the technical field of collision warning, and in particular relates to a ship collision warning system based on multi-source data. Background Art

[0002] In recent years, with the prosperous development of the maritime transportation industry, the types, number and tonnage of ships have increased, the density of ships passing through the waterway has become increasingly large, and the navigation environment has become more complex and changeable, resulting in frequent maritime ship collision accidents. Ship collision accidents not only cause serious casualties and economic losses, but may also cause serious marine environmental pollution, becoming one of the major hidden dangers to ship navigation safety. Therefore, early warning of ship collision accidents, prevention and avoidance of maritime collision accidents have become the focus of research on maritime ship navigation risks.

[0003] At present, the International Maritime Organization has formulated international collision avoidance regulations for ships sailing at sea, aiming to standardize the navigation methods and rules of ships and reduce the probability of ship collision accidents. However, for emergency encounters of ships sailing at sea, it is difficult to avoid collisions by manual operation alone. Intelligent collision avoidance of ships aims to provide ships with real-time, reasonable and optimal collision avoidance solutions on the premise of predicting the risk of collision, so as to effectively solve the collision problem. At present, a large number of path planning studies use the method of randomly generating obstacles to simulate the real environment. This method is quite different from the terrain information of the real environment and it is easy to ignore the influence of the size of the ship itself, making it difficult to guarantee the performance of the model in the real scene. In addition, it is inevitable that ships will encounter unexpected situations such as ship encounters and unknown obstacles during navigation. Some studies usually do not deliberately distinguish such situations, resulting in the inability to ensure safety when dealing with such situations. At the same time, the traditional path planning algorithm based on reinforcement learning involves a lot of storage and calculation in the path planning process, resulting in long path planning time and high accident rate. The traditional heuristic search algorithm requires a lot of space storage during the search process, which makes it impossible to handle real-time collision avoidance. Summary of the invention

[0004] The present invention aims to solve the shortcomings of the prior art and proposes a ship collision warning system based on multi-source data. By collecting information on the ship and obstacles within the detection range, the collision risk area of ​​the two ships is determined, and the future path of obstacles that may collide with are predicted and the collision level is calculated in advance. By using the future path as an obstacle avoidance point, the obstacle avoidance path of the ship is planned, which can avoid accidents to a greater extent.

[0005] To achieve the above-mentioned object, the present invention provides the following scheme: a ship collision warning system based on multi-source data, comprising: a detection system, a prediction system, an analysis system and a warning system;

[0006] The detection system is used to obtain the collision risk area based on the ship's driving information;

[0007] The prediction system is used to predict the future path of the obstacle in the collision risk zone;

[0008] The analysis system is used to calculate the risk level based on the future path, and plan the future path of the ship to obtain an obstacle avoidance path;

[0009] The warning system is used for performing collision warning based on the risk level.

[0010] Further preferably, the detection system comprises: a first division module and a second division module;

[0011] The first division module is used to collect the ship travel information and divide the detection area based on the ship travel information;

[0012] The second division module is used to collect obstacle information in the detection area and divide the collision risk area based on the obstacle information and the ship driving information.

[0013] Further preferably, the method of dividing the collision risk area by the second dividing module includes:

[0014]

[0015] In the formula, P represents the side angle; DCPA represents the shortest distance encountered; D represents the distance between them; and K represents the ship speed ratio.

[0016] Further preferably, the prediction system comprises: a prediction data module, a data processing module and a path prediction module;

[0017] The prediction data module is used to obtain AIS trajectory data of obstacles;

[0018] The data processing module is used to pre-process the AIS trajectory data to obtain path prediction data;

[0019] The path prediction module is used to estimate and predict obstacles based on the path prediction data to obtain the future path.

[0020] Further preferably, the preprocessing method performed by the data processing module includes: trajectory data extraction, missing value completion and normalization processing.

[0021] Further preferably, the analysis system comprises: a first collision risk calculation module, a second collision risk calculation module and a judgment module;

[0022] The first collision risk calculation module is used to calculate the first probability of collision between the ship and the obstacle at the spatial level during the driving process;

[0023] The second collision risk calculation module is used to calculate the second probability of collision between the ship and the obstacle at the time level during the driving process;

[0024] The judgment module is used to judge whether a collision event between a ship and an obstacle will occur based on the first probability and the second probability;

[0025] The path planning module is used to plan the future path of the ship to obtain the obstacle avoidance path.

[0026] Further preferably, the method for calculating the first probability includes:

[0027]

[0028] Where P1 represents the first probability; d1 and d2 represent the safe encounter distance and safe passing distance respectively; DCPA represents the shortest encounter distance.

[0029] Further preferably, the method for calculating the second probability includes:

[0030]

[0031] Where P2 represents the second probability; t1 and t2 represent the latest steering time and the ship heading time when the relative distance is 12n mile respectively; TCPA represents the shortest encounter time.

[0032] Compared with the prior art, the present invention has the following beneficial effects:

[0033] The present invention collects information about the ship and obstacles within the detection range, determines the collision risk area of ​​the two ships, predicts the future path of obstacles that may collide and calculates the collision level in advance, and plans the obstacle avoidance path of the ship by using its future path as the obstacle avoidance point. At the same time, the existing obstacle avoidance path planning method is optimized and improved to obtain the obstacle avoidance path. It can flexibly avoid static or dynamic obstacles to a greater extent and avoid accidents. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the technical solution of the present invention, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0035] Figure 1 The diagram is a structural diagram of a ship collision warning system based on multi-source data according to an embodiment of the present invention. DETAILED DESCRIPTION

[0036] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0037] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0038] Embodiment 1:

[0039] like Figure 1 As shown, this embodiment provides a ship collision warning system based on multi-source data, including: a detection system, a prediction system, an analysis system and a warning system; the detection system is used to obtain a collision risk area based on ship driving information; the prediction system is used to predict the future path of obstacles in the collision risk area; the analysis system is used to calculate the risk level based on the future path, and plan the future path of the ship to obtain an obstacle avoidance path; the warning system is used to provide collision warnings based on the risk level.

[0040] In this embodiment, the detection system includes: a first division module and a second division module; the first division module is used to collect ship driving information and divide the detection area based on the ship driving information; the second division module is used to collect obstacle information in the detection area and divide the collision risk area based on the obstacle information and the ship driving information.

[0041] The detection area refers to the area where the ship and other marine navigation equipment (obstacles) are traveling at a relatively high speed (determined according to the actual sea area) in the propagation direction, and the collision can be completely avoided in time.

[0042] The method for dividing the collision risk area by the second dividing module includes:

[0043]

[0044] In the formula, P represents the side angle; DCPA represents the shortest distance encountered; D represents the distance between them; and K represents the ship speed ratio.

[0045] In this embodiment, the prediction system includes: a prediction data module, a data processing module and a path prediction module; the prediction data module is used to obtain the AIS trajectory data of the obstacle; the data processing module is used to pre-process the AIS trajectory data to obtain the path prediction data; the path prediction module is used to estimate and predict the obstacle based on the path prediction data to obtain the future path.

[0046] Among them, the preprocessing method of the data processing module includes: trajectory data extraction, missing value filling and normalization processing.

[0047] The methods for extracting trajectory data include:

[0048] For the data point at the time i of data acquisition, the time interval between it and the previous data point is calculated. When the time interval is greater than 6 minutes (according to ITU-R rules, AIS navigation information is updated every 6 minutes), it is determined whether the time interval is greater than 6 hours. If it is greater than 6 hours, the previous data point is taken as the focus, and the data point at time i is taken as the starting point to obtain the trajectory data set, and the next data is determined. If the time interval is less than 6 hours, the data between the data point at time i and the previous data point is supplemented with missing values, and all the supplemented trajectory points are added to the trajectory data set. In this embodiment, the missing value supplementation adopts the mean difference supplementation method.

[0049] When the time interval is less than 6 minutes, determine whether the current speed of the ship is less than 0.3 nautical miles per hour. If it is less than, take the data point at time i as the end point, obtain the trajectory data set, and make the next data judgment. If the speed is greater than 0.3 nautical miles per hour, the data at time i will be used for the next round of judgment.

[0050] The path prediction module uses a trajectory prediction model to process the path prediction data to obtain the future path. The path prediction model consists of two parts: an encoder and a decoder. In the encoder part, two parallel time convolutions are set: global time convolution and local time convolution. The global time convolution is mainly used to extract features from the path prediction data to obtain a feature vector that represents the continuity of the obstacle trajectory state; the local time convolution is mainly used to extract features from the path prediction data to obtain a feature vector that represents the rapid mutation of the obstacle trajectory. The two feature vectors are then concatenated and output. The decoder part uses an RNN-based neural network to generate a sequence of future paths using a fully connected layer for the concatenated feature vectors output by the encoder and the path prediction data.

[0051] In this embodiment, the analysis system includes: a first collision risk calculation module, a second collision risk calculation module and a judgment module; the first collision risk calculation module is used to calculate the first probability of collision between the ship and the obstacle at the spatial level during the driving process; the second collision risk calculation module is used to calculate the second probability of collision between the ship and the obstacle at the temporal level during the driving process; the judgment module is used to judge whether a collision event between the ship and the obstacle will occur based on the first probability and the second probability; the path planning module is used to plan the future path of the ship to obtain an obstacle avoidance path.

[0052] The calculation method of the first probability includes:

[0053]

[0054] Where P1 represents the first probability; d1 and d2 represent the safe encounter distance and safe passing distance respectively; DCPA represents the shortest encounter distance.

[0055] The calculation method of the second probability includes:

[0056]

[0057] Where, P2 represents the second probability; t1 and t2 represent the latest steering time and the ship heading time when the relative distance is 12 nmile respectively; TCPA represents the shortest encounter time.

[0058] Afterwards, the risk level is determined according to the minimum value of the first probability and the second probability, and whether the ships will collide is determined based on the risk level.

[0059] Regardless of whether the ship will collide, the path planning module plans the ship's obstacle avoidance path in real time based on the prediction of the future path of obstacles, so that when the staff decides to change the route, they can act quickly and reduce calculation time.

[0060] In this embodiment, the method for the path planning module to plan the obstacle avoidance path of the ship includes:

[0061] Based on the principle of A* algorithm, two lists are established, one of which is list a for collecting search points and list b for collecting points with minimum cost. Connect the points with minimum cost in list b to determine whether the target point has been reached. If it has, stop searching; if not, continue searching and collect the minimum cost node into list b. After the search is completed, an obstacle-free global path from the starting point to the end point will be generated. After reaching the target point, redundant points are removed, the path is smoothed, and the key point is captured. After the key point is captured, the node is initialized and the speed is sampled. The evaluation function is constructed based on the principle of DWA algorithm, and the parameters of the evaluation function are designed by designing a fuzzy controller. The evaluation function after parameter optimization is used to select the optimal path and determine whether the node is the target node. If not, it is updated and selected again until the target point is reached.

[0062] The embodiments described above are only descriptions of the preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Without departing from the design spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by ordinary technicians in this field should all fall within the protection scope determined by the claims of the present invention.

Claims

1. A ship collision warning system based on multi-source data, characterized in that: include: Detection systems, prediction systems, analysis systems, and early warning systems; The detection system is used to obtain the collision risk area based on the ship's driving information; The prediction system is used to predict the future path of the obstacle in the collision risk zone; The analysis system is used to calculate the risk level based on the future path, and plan the future path of the ship to obtain an obstacle avoidance path; The warning system is used for performing collision warning based on the risk level.

2. According to claim 1, a ship collision warning system based on multi-source data is characterized in that: The detection system comprises: a first division module and a second division module; The first division module is used to collect the ship travel information and divide the detection area based on the ship travel information; The second division module is used to collect obstacle information in the detection area and divide the collision risk area based on the obstacle information and the ship driving information.

3. According to claim 2, a ship collision warning system based on multi-source data is characterized in that: The method for dividing the collision risk area by the second dividing module includes: In the formula, P represents the side angle; DCPA represents the shortest distance encountered; D represents the distance between them; and K represents the ship speed ratio.

4. According to claim 1, a ship collision warning system based on multi-source data is characterized in that: The prediction system includes: a prediction data module, a data processing module and a path prediction module; The prediction data module is used to obtain AIS trajectory data of obstacles; The data processing module is used to pre-process the AIS trajectory data to obtain path prediction data; The path prediction module is used to estimate and predict obstacles based on the path prediction data to obtain the future path.

5. A ship collision warning system based on multi-source data according to claim 4, characterized in that: The method for preprocessing by the data processing module includes: trajectory data extraction, missing value filling and normalization processing.

6. The ship collision warning system based on multi-source data according to claim 1, characterized in that: The analysis system includes: a first collision risk calculation module, a second collision risk calculation module, a judgment module and a path planning module; The first collision risk calculation module is used to calculate the first probability of collision between the ship and the obstacle at the spatial level during the driving process; The second collision risk calculation module is used to calculate the second probability of collision between the ship and the obstacle at the time level during the driving process; The judgment module is used to judge whether a collision event between a ship and an obstacle will occur based on the first probability and the second probability; The path planning module is used to plan the future path of the ship to obtain the obstacle avoidance path.

7. A ship collision warning system based on multi-source data according to claim 6, characterized in that: The calculation method of the first probability includes: Where P1 represents the first probability; d1 and d2 represent the safe encounter distance and safe passing distance respectively; DCPA represents the shortest encounter distance.

8. The ship collision warning system based on multi-source data according to claim 6, characterized in that: The calculation method of the second probability includes: Where, P2 represents the second probability; t1 and t2 represent the latest steering time and the ship heading time when the relative distance is 12 nmile respectively; TCPA represents the shortest encounter time.

Images