Method for automatically planning shortest-time route of ship based on multi-neighborhood rule grid

Abstract

The application discloses a kind of based on the shortest time route of ship automatic planning method of multi-neighborhood rule grid, belong to the field of ocean surveying and mapping technology.Firstly, read the navigation-obstructing area dataset based on the fusion of multiple electronic charts;Regular grid is constructed to divide ship navigation space, and the connectivity of space division is improved by multi-neighborhood link;Remove the invalid link intersecting with the same navigation-obstructing area;Read the historical track data of ship, and construct the influence model of hydro-meteorological elements on ship navigation state in combination with hydro-meteorological data;Read real-time hydro-meteorological data, and realize the dynamic iteration of route by using the improved A* algorithm with route window constraint;Iterative solution route from starting point until end point.The application introduces ocean environmental information quantitatively into the process of ship route planning, overcomes the shortcomings of existing shortest time route planning method, such as single search direction and dependence on empirical formula;Can intelligently avoid bad sea conditions, and on the basis of ensuring the safety and reliability of the planned route, improve the applicability of the planned route.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of marine surveying and mapping, and relates to a ship shortest time route automatic planning method based on a multi-neighborhood regular grid. BACKGROUND

[0002] Since this century, the pace of human development of the ocean and construction of the ocean has been constantly advancing, and the advent of the big data era has directly guided the value realization of marine spatio-temporal big data. As an essential element of ship activities at sea, the route is generally obtained based on the comprehensive analysis of electronic charts, hydrology and meteorology, navigation experience and other data, and is a data-driven product. At present, many scholars have explored the automatic route planning method from different angles and have made a lot of work and achieved fruitful research results.

[0003] Chang et al. proposed a shortest distance route generation algorithm based on improved maze. Wang Zhu et al. proposed a route automatic generation method based on route binary tree in view of the weakness of greediness in route automatic generation. Zhu Qing selected the visibility graph method under free space, and equivalent the ship route automatic planning problem to a graph theory problem. Zhang Lihua et al. improved the efficiency of the method and expanded the practicability of the shortest distance route automatic planning method by effectively dividing and optimizing the results of the adaptive triangular net, on the basis of ensuring the theoretical rigor. However, these methods are still essentially limited to relatively static marine spatial information, and do not take into account the influence of dynamically changing hydro-meteorological elements (mainly surface wind, ocean current and sea wave) on ship navigation. The shortest distance route cannot be simply equated to the shortest time route. Vlachos et al. combined wind and wave forecasts and carried out research on dynamic optimization of ship route. Kosmas improved the optimization degree of the ship route in the dynamic environment on the basis of the theory of Vlachos through the Lagrange algorithm and the simulated annealing algorithm. Wu Gongxing et al. introduced the ship speed loss function and preliminarily realized the shortest time route planning under local meteorological conditions. However, the above-mentioned methods still have the following two shortcomings: first, when searching the grid space, the existing methods only consider the grids geometrically connected with the current grid and simply expand in four / eight directions, without considering the direct influence of the ship heading on the speed in the dynamic environment. The single search direction limits the subsequent route planning to a certain extent; second, the existing methods mostly rely on empirical formulas when calculating the ship speed under the influence of dynamic hydro-meteorological elements, and the fitting process is not only complex, but also limited to certain or several ship types, which is not strong in universality. SUMMARY

[0004] In order to overcome the shortcomings of the existing ship shortest distance route planning method in search direction and speed calculation, the application provides a ship shortest time route automatic planning method based on a multi-neighborhood regular grid.

[0005] The technical solution adopted by the present invention to achieve the above objectives is as follows:

[0006] An automatic shortest-time route planning method for ships based on a multi-neighborhood rule grid includes the following steps:

[0007] a. Based on the electronic chart data standard, analyze the electronic chart data of the navigation area, integrate multiple chart data, construct and read the navigation obstruction area dataset fused from multiple chart data;

[0008] b. Divide the ship's navigation space into grids and construct a multi-neighborhood link network between grids using user-specified neighborhood templates to improve the connectivity of the spatial division;

[0009] c. Based on the navigation obstruction area dataset fused from multiple nautical charts, remove invalid links that intersect with the fused navigation obstruction area;

[0010] d. Read the ship's historical navigation information and combine it with historical hydrological and meteorological data to construct a model of the impact of hydrological and meteorological elements on the ship's navigation status;

[0011] e. Read real-time hydrological and meteorological data, and based on the multi-neighborhood rule grid, use the updated timeliness of hydrological and meteorological data as the step size, and adopt the improved A* algorithm with route window constraints to realize the dynamic iteration of the route.

[0012] f. Iterate through the route from the starting point until the predetermined destination is reached, completing the route planning process and outputting the results in coordinate form.

[0013] The basic idea behind the multi-neighborhood link network described in step b is:

[0014] A neighborhood template of user-specified size is constructed. Arbitrary grids within the defined ship navigation space are connected to other grids within the template's range via lines, establishing adjacency relationships including adjacency angles and distances. This expands the number of grids linked to the current grid and increases the grid expansion directions, thus forming a multi-neighborhood linked network. The neighborhood template size refers to the total number of grids in the horizontal / vertical directions when the multi-neighborhood grid expands outwards. When the neighborhood template size is n×n, the central grid will be linked to n surrounding grids. 2 -1 grid is used to establish adjacency relationships, including adjacency angle and adjacency distance.

[0015] by Figure 2 Taking the 5×5 adjacent template construction shown in (a) as an example, without loss of generality, the center point (x0, y0) of the template center regular grid can be regarded as the local coordinate origin to construct a template coordinate system. In the first quadrant of this coordinate system, the grid (x0, y0) will connect the six grid centers indicated by the arrows to establish a connection including adjacent angles (such as...). Figure 2 (a) θ (1,2)The adjacent relationship is the adjacent relationship between the adjacent angle and the adjacent distance of the grid (x0, y0) and the grid (x1, y2). According to the symmetry of the coordinate system, the grid adjacent relationship of the remaining three quadrants can be easily constructed, and then the link network diagram is formed.

[0016] In the Euclidean plane geometry space, the adjacent distance of the constructed link network diagram is represented in (b) in the figure, and in the real marine geographic information data environment, the distance weight in the link network diagram will be replaced by the real ellipsoid distance L between grids. Figure 2

[0017] In step c, the specific method for removing invalid links intersecting with the fusion navigation-obstructing area is:

[0018] Using the discrete characteristics of the regular grid, based on the Bresenham algorithm and the adjacent relationship between the grids, the grid numbers passed through by the link are quickly obtained, and it is judged whether these grids contain navigation-obstructing areas or are covered by navigation-obstructing areas. If yes, it is considered that the link intersects with the navigation-obstructing area and is removed; if not, it is considered that the link is navigable and is retained.

[0019] In step d, the method for constructing the influence model of hydro-meteorological elements on the ship navigation state is:

[0020] (1) Read the ship historical track data recorded in the Automatic Identification System (AIS), and preprocess the AIS data according to ITU-R M.585-7, including parsing, cleaning, and integrity verification, to obtain ship trajectory information (including ship position, ground speed, and ground heading) and static ship parameter information (including ship length, width, and draft) that meet the requirements of the file;

[0021] (2) Read the static hydro-meteorological historical data to obtain wind direction, wind speed, wave height, wave period, flow direction, and flow speed;

[0022] (3) In order to obtain the corresponding relationship between the AIS data and the hydro-meteorological data, the input AIS data and hydro-meteorological data are normalized and spatially interpolated to obtain hydro-meteorological information at any ship position, and the data are randomly divided into training samples and test samples in proportion.

[0023] ​(4) As preferred, the training sample and the test sample are respectively input into a multi-layer extreme learning machine with regularization (MR-ELM) based on regular optimization, the output weight matrix of each hidden layer is obtained, a nonlinear model of the influence of hydro-meteorological elements on the ship speed is constructed, and the maximum speed of the ship under different hydro-meteorological conditions is calculated based on the model.

[0024] In the MR-ELM model, the relationship between the output of the i-th hidden layer and the output of the (i-1)-th hidden layer is expressed as follows:

[0025]

[0026] wherein H i represents the output of the i-th hidden layer, H i-1 represents the output of the (i-1)-th hidden layer, if i = 1, then H i-1 represents the training sample input into the MR-ELM; g(x) represents an activation function in the hidden layer; β i represents the output weight matrix calculated by regular optimization, β i is specifically calculated as follows:

[0027]

[0028] wherein C is a regularization control parameter, N is the number of normalized training samples, n is the number of hidden layer neurons, I is a unit matrix, and Y is a target matrix in the training sample.

[0029] In the step e, the calculation method of the heuristic function F(n) in the improved A* algorithm is as follows when searching the node:

[0030]

[0031] wherein T(n) is the cumulative sailing time from the starting point to the current node, L(n) is the ellipsoidal distance of the shortest distance route from the current node to the terminal, V0 is the ship speed in still water given by the user, p is the spatial feature of the hydro-meteorological data used for calculating the ship speed, t is the time feature of the hydro-meteorological data, θ 地 is the ship heading to the ground calculated according to the grid adjacency angle relationship, L(p, θ 地 ) is the ellipsoidal distance between nodes calculated according to the grid adjacency distance relationship, and V(p, θ 地 , t) is the ship speed calculated according to the influence model of the hydro-meteorological elements on the ship sailing state.

[0032] In the step e, the concept of the route window constraint is as follows:

[0033] The shortest distance route between the starting point and the ending point of the route is established, and the buffer zone formed by expanding the connecting line by a certain range is called a route window, and the linking network nodes contained in the route window are the search nodes of the path search algorithm. Without loss of generality, before the node search, the shortest distance route between the starting point and the ending point of the route is established by using an existing efficient algorithm (such as: an adaptive triangular network ship route automatic planning method with controllable error and considering efficiency), and a buffer zone Buffer of the shortest distance route is constructed by taking a user preset threshold as a buffer radius, the Buffer is regarded as a route window constraint, and the range is regarded as a node search range. When the searched node exceeds the constraint boundary, the node is skipped. The route window constraint can reduce the number of network nodes included in the calculation, and improve the route planning efficiency.

[0034] The hydro-meteorological information is quantitatively introduced into the ship route planning process, and the deficiencies of the existing shortest distance ship route planning method in the search direction and the calculation of the speed are overcome. The existing experiments show that the method can intelligently avoid the adverse sea conditions under the support of the marine space-time data, and on the basis of ensuring the safety and reliability of the planned route, the applicability of the planned route is improved. BRIEF DESCRIPTION OF DRAWINGS

[0035] Figure 1 It is a main flow block diagram of a ship shortest time route planning method based on a multi-neighborhood rule grid proposed by the application.

[0036] Figure 2 It is a neighborhood template schematic diagram constructed by the application, wherein (a) is a neighborhood relationship construction schematic diagram of a 5*5 neighborhood template, and (b) is a neighborhood distance of a linking network graph constructed by a 5*5 neighborhood template in a Euclidean plane geometric space.

[0037] Figure 3 It is a flow block diagram of a nonlinear model for constructing the influence of hydro-meteorological elements on the ship speed by using MR-ELM proposed by the application. DETAILED DESCRIPTION

[0038] The application will be further described below in combination with the drawings and specific embodiments. The embodiments are implemented on the premise of the technical scheme of the application, and specific implementation schemes and specific operation processes are given, but the protection scope of the application is not limited to the following embodiments.

[0039] The shortest time route automatic planning method for the ship provided by the application, the implementation process is to realize the automatic planning of the ship route based on the real marine space-time data by using the computer, assuming that the route between the starting point S and the ending point E in a certain sea area needs to be planned, the ship parameters used in the embodiment are shown in Table 1, the starting time is set to Beijing time August 1, 2017 0900, the efficient automatic planning of the route is carried out by using the method described in the application, including the following steps, as shown in Figure 1 .

[0040] Table 1 Ship parameters used in the embodiment

[0041]

[0042] Step a, according to the electronic chart data standard, parse the electronic chart data of the sailing sea area, fuse multiple chart data, construct and read the navigation area data set O={o1, o2, o3…o n} of multiple chart data fusion.

[0043] Step b, divide the ship sailing space in a grid manner to form a grid data set G={g1, g2, g3…g m}; construct a neighborhood template with a size of 5*5, establish the adjacency relationship including the adjacency angle and the adjacency distance between any grid after the rule division of the ship sailing space and other grids in the neighborhood template range in a line manner, and construct the multi-neighborhood link network between the grids.

[0044] Step c, according to the navigation area data set O={o1, o2, o3…o n}, based on the Bresenham algorithm and the adjacency relationship between the grids, quickly obtain the grid numbers passed by the grid link, if the grid contains the navigation area or is covered by the navigation area, remove the link.

[0045] Step d, read the historical track information of the ship, combine the historical hydro-meteorological data to construct the influence model of the hydro-meteorological elements on the ship sailing state; the application preferably uses MR-ELM to construct the model, and the process is as shown in Figure 3 .

[0046] Firstly, the ship history track data recorded in the Automatic Identification System (AIS) is read, and the AIS data is analyzed, cleaned and verified for integrity, etc. The ship trajectory information (including ship position, ground speed and ground heading) and static ship parameter information (including ship length, width and draft) meeting the requirements of ITU-R M.585-7 document are obtained; secondly, the static hydro-meteorological historical data is read to obtain hydro-meteorological information such as wind direction, wind speed, wave height, wave period, flow direction and flow rate; thirdly, the input AIS data and hydro-meteorological data are normalized and preprocessed, and the AIS trajectory data and hydro-meteorological data are spatio-temporally interpolated to obtain hydro-meteorological information at any ship position, and the data is randomly divided into training samples and test samples in the proportions of 80% and 20%; finally, the training samples and test samples are input into the MR-ELM to obtain the output weight matrix of each hidden layer, and a nonlinear model of the influence of hydro-meteorological elements on ship speed is constructed, and the maximum speed that the ship can reach under different hydro-meteorological conditions is calculated based on the model. In this embodiment, the number of layers of the hidden layer of the MR-ELM model used is 5, the number of neurons in each hidden layer is 300, and the regularization control parameter is 10 -3 .

[0047] Step e, read real-time hydro-meteorological data, on the basis of the multi-neighborhood regular grid, take the time-to-life of the hydro-meteorological data as the step (in this embodiment, the step is set to 72 hours), use the improved A* algorithm with route window constraint to realize the dynamic iteration of the route, specifically:

[0048] Before node searching, the shortest distance route Route between the starting point and the ending point of the route planning is established by using the adaptive triangular net ship route automatic planning method (patent 202110510128.2) with controllable error and considering efficiency D , and a buffer zone Buffer of the shortest distance route is constructed with the user preset threshold R as the buffer radius, the Buffer is regarded as the route window constraint, and the range of the Buffer is regarded as the node search range. When the searched node exceeds the constraint boundary, the node is skipped.

[0049] In the node search, the calculation method of the heuristic function F(n) in the improved A* algorithm is:

[0050]

[0051] Wherein, T(n) is the accumulated sailing time from the starting point to the current node, L(n) is the ellipsoidal distance of the shortest distance route from the current node to the ending point, V0 is the static water speed of the ship given by the user, p is the spatial feature of the hydro-meteorological data used in the calculation of the speed, t is the time feature of the hydro-meteorological data, and θ地 L(p, θ) is the shortest distance route calculated according to the grid adjacency angle relationship, V(p, θ 地 ) is the ellipsoid distance between nodes calculated according to the grid adjacency distance relationship, V(p, θ 地 , t) is the ship speed calculated according to the influence model of hydro-meteorological elements on the ship navigation state.

[0052] Step f: iteratively solving the route from the starting point S until reaching the predetermined end point E, completing the route planning process, obtaining all the waypoints between the start and end points, and finally obtaining a complete shortest time route Route T , and outputting in coordinate form.

[0053] The shortest distance route between the start and end points, the route window constraint in the route planning process, and the shortest time route between the start and end points are plotted together under the electronic chart platform. In order to more intuitively compare the differences between the shortest distance route and the shortest time route, the distances of the two are calculated, the sailing time under static water speed is calculated, and the actual sailing time under the influence of hydro-meteorological elements is calculated based on formula (6). The specific numerical comparison is shown in Table 2.

[0054] Table 2: Comparison of route distance and sailing time of the example route

[0055]

[0056] As can be clearly seen from the table, although the distance of the shortest distance route is shorter than that of the shortest time route, and the sailing time under static water speed is naturally shorter, after considering the hydro-meteorological information, the time spent along the shortest distance route is longer than that along the shortest time route. Moreover, the shortest distance route cannot be dynamically adjusted with time, and may face adverse sea conditions, which has certain safety risks. This fully embodies the effectiveness of the ship shortest time route planning method based on the multi-neighborhood rule grid proposed in the present application.

[0057] The present application is described through examples, and those skilled in the art know that various changes or equivalent replacements can be made to these features and examples without departing from the spirit and scope of the present application. In addition, under the guidance of the present application, these features and examples can be modified to adapt to specific conditions and materials without departing from the spirit and scope of the present application. Therefore, the present application is not limited by the specific examples disclosed herein, and all examples falling within the scope of the claims of the present application are within the protection scope of the present application.

Claims

1. A method for automatic shortest-time route planning for ships based on a multi-neighborhood rule grid, characterized in that, The method includes the following steps: a. Based on the electronic chart data standard, analyze the electronic chart data of the navigation area, integrate multiple chart data, construct and read the navigation obstruction area dataset fused from multiple chart data; b. Divide the ship's navigation space into grids and construct a multi-neighborhood link network between grids using user-specified neighborhood templates to improve the connectivity of the spatial division; The multi-neighborhood link network is constructed as follows: A neighborhood template of user-specified size is constructed. Any grid within the neighborhood template, after the ship's navigation space is divided according to rules, is connected to other grids within the template to establish adjacency relationships including adjacency angles and distances. This expands the number of grids linked to the current grid, thereby increasing the expansion directions. The neighborhood template size refers to the number of grids in the horizontal / vertical direction when multiple neighborhood grids are linked outwards. When the neighborhood template size is n×n, the central grid will connect with n surrounding grids. 2 -1 grid is used to establish adjacency relationships, including adjacency angle and adjacency distance; c. Based on the navigation obstruction area dataset fused from multiple nautical charts, remove invalid links that intersect with the fused navigation obstruction area; d. Read the ship's historical navigation information and combine it with historical hydrological and meteorological data to construct a model of the impact of hydrological and meteorological elements on the ship's navigation status; The specific method for constructing the model of the impact of hydrological and meteorological elements on the navigation status of ships is as follows: (1) Read the ship's historical track data recorded in the Automatic Identification System (AIS), and parse, clean and preprocess the AIS data in accordance with ITU-RM.585-7 to obtain ship trajectory information including ship position, ground speed and ground heading, and static ship parameter information including ship length, width and draft, which meet the requirements of the document. (2) Read static hydrological and meteorological historical data to obtain hydrological and meteorological information including wind direction, wind speed, wave height, wave period, flow direction, and flow velocity; (3) Normalize the spatiotemporal interpolation of the input AIS data and hydrological and meteorological data to obtain hydrological and meteorological information at any ship position, and randomly divide the data into training samples and test samples according to the proportion. (4) Input the training samples and test samples into the multilayer extreme learning machine MR-ELM based on regularization optimization respectively, obtain the output weight matrix of each hidden layer, construct a nonlinear model of the influence of hydrological and meteorological elements on ship speed, and calculate the maximum speed that ships can reach under different hydrological and meteorological conditions based on this model. The regularization optimization method in MR-ELM is as follows: In the MR-ELM model, the relationship between the output of the i-th hidden layer and the output of the (i-1)-th hidden layer is expressed by the following equation: Among them, H i H represents the output of the i-th hidden layer. i-1 This represents the output of the (i-1)th hidden layer. If i = 1, then H is... i-1 This represents the training samples input to MR-ELM; g(x) represents the activation function in the hidden layer; β i Let β represent the output weight matrix calculated through regularization optimization. i The specific calculation method is expressed as follows: Where C is the regularization control parameter, N is the number of normalized training samples, n is the number of hidden layer neurons, I is the identity matrix, and Y is the target matrix in the training samples; e. Read real-time hydrological and meteorological data, and based on the multi-neighborhood rule grid, use the updated timeliness of hydrological and meteorological data as the step size, and adopt the improved A* algorithm with route window constraints to realize the dynamic iteration of the route. During node search, the improved A* algorithm calculates the heuristic function F(n) as follows: Where T(n) is the accumulated travel time from the starting point to the current node, L(n) is the ellipsoidal distance of the shortest route from the current node to the destination, V0 is the ship's still-water speed given by the user, p is the spatial characteristic of the hydro-meteorological data used in calculating the speed, t is the temporal characteristic of the hydro-meteorological data, and θ 地 L(p,θ) is the ship's heading relative to the ground, calculated based on the grid's adjacency angle relationship. 地 V(p,θ) represents the ellipsoidal distance between nodes calculated based on the grid adjacency distance relationship. 地 ,t) is the ship speed calculated based on the model of the influence of hydrological and meteorological elements on the ship's navigation status; f. Iterate through the route from the starting point until the predetermined destination is reached, completing the route planning process and outputting the results in coordinate form.

2. The automatic shortest time route planning method for ships based on a multi-neighborhood rule grid as described in claim 1, characterized in that, The specific method for step c is as follows: Based on the navigation obstruction zone dataset, the grid numbers traversed by the link are quickly obtained using the Bresenham algorithm and the adjacency relationship between grids. It is then determined whether these grids contain or are covered by navigation obstruction zones. If so, the link intersects with the navigation obstruction zone and is removed; otherwise, the link is navigable and is retained.

3. The automatic shortest time route planning method for ships based on a multi-neighborhood rule grid as described in claim 1, characterized in that, In step e, before node search, the method for constructing the route window constraints is as follows: Establish routes to plan the shortest distance between the origin and destination. D The system constructs a buffer for the shortest route using a user-preset threshold R as the buffer radius. This buffer is considered a route window constraint, and the range of the buffer is considered the node search range. When a searched node goes beyond the constraint boundary, the node is skipped.

Images