A ship inter-port route planning method fusing large model knowledge

Abstract

The application discloses a kind of ship interport route planning methods of fusing big model knowledge, including S1, route planning analysis based on big model: using big model analysis the question about route planning input by user, obtain question information, determine the route planning service that needs to be called using big model to question information is analyzed using tool, using big model to route planning service is analyzed, determine its required corresponding parameters;S2, route planning service call: call corresponding route planning service by inputting corresponding parameters, obtain route planning result;S3, route planning result display based on big model: route planning result is analyzed using big model, to realize line visualization.The advantage is: it can be used for the path planning between two ports that have not been navigated at all, the route is based on real trajectory, will not cross land and reef and other obstacles, can more accurately avoid strait channel.

Description

Technical Field

[0001] This invention relates to the field of route planning technology, and in particular to a method for planning ship routes between ports that integrates large-scale model knowledge. Background Technology

[0002] Route planning has always been a key research area in the maritime industry. Currently, there are several methods for route planning: (1) Experienced routes, which obtain experienced routes between two ports based on ship berthing records. This method is not suitable for situations where there are no historical routes between two ports and a new route needs to be opened; (2) Route planning only through grids, which divides the global latitude and longitude into small grids and assumes that two adjacent grids can be connected. Path planning is performed through shortest path or reinforcement learning algorithms. This method can plan new routes, but because it does not strictly determine the trajectory connectivity of the grids, it is easy to produce routes that cross land.

[0003] Moreover, traditional route planning requires users to know the latitude and longitude information of various ports around the world as input conditions, which requires a high level of expertise. Summary of the Invention

[0004] The purpose of this invention is to provide a method for planning ship routes between ports that integrates knowledge from large models, thereby solving the aforementioned problems existing in the prior art.

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

[0006] A method for planning ship routes between ports that integrates knowledge from a large model includes the following steps:

[0007] S1. Route planning analysis based on a large model: The large model is used to analyze the user's input questions about route planning, obtain the question information, use the large model to perform tool analysis on the question information to determine the route planning service that needs to be called, and use the large model to perform parameter analysis on the route planning service to determine the corresponding parameters required.

[0008] S2. Route planning service call: Input the corresponding parameters to call the corresponding route planning service and obtain the route planning results;

[0009] S3. Display of route planning results based on large model: Utilize large model to analyze route planning results to achieve route visualization.

[0010] Preferably, the route planning service includes the following:

[0011] S21. Divide the map from longitude -180° to 180° and latitude -90° to 90° into rectangular grids with units of 1°, resulting in a set of rectangular grids, denoted as follows:

[0012] Grid={(WKT,id,left,top,rigth,bottom)}

[0013] Wherein, Grid is a set of rectangular grids; WKT is the WKT format region representation of the grid; id is the unique number of the grid; left, right, top, and bottom are the left, right, top, and bottom edges of the grid, respectively;

[0014] S22. For the two ports of origin and destination, define a weighted directed graph G.

[0015] G = (V, E, W)

[0016] V = {v} (1.plon,plat) ,v (2.plon,plat) ,…,v (n.plon,plat)}

[0017] E={e ij}

[0018] W = {w ij}

[0019] Where V is the set of grids between the origin port and the destination port; n is the grid ID; plon and plat are the longitude and latitude of the port, respectively; E is the set of edges; e ij For the i-th grid v (i.plon,plat) and the j-th grid v (j.plon,plat) The edges between; W is the set of edge weights; w ij For grid v (i.plon,plat) and v (j.plon,plat) The boundary weight between;

[0020] For a directed graph G with edge set E, and a ship AIS data set S and a rectangular grid set Grid as input, when ship k (t y ,lon y ,lat y In Grid u In China, ship k (t y+1 ,lon y+1 ,lat y+1 In Grid v In the middle, and Grid u and Grid v If there is at least one identical edge, then (u,v,1) is added to the grid connected set L. Traverse all ships in the ship AIS data set S and output the grid connected set L(source,target,value).

[0021]

[0022] Among them, ship k For the ship with ship number k, k = 1, 2, ..., K; k [(t y ,lon y ,lat y )] for ship k At the y-th timestamp t y Longitude of the location y and latitude y ;source is the source node; target is the target node; value is the connection value, each connection is counted as 1; Grid u and Grid v These are the u-th and v-th grids in the rectangular grid set Grid, respectively.

[0023] For a set of edge weights W in a directed graph G, the distance impedance and connectivity impedance between adjacent nodes are calculated based on the length of the mesh connected set. The total impedance obtained by superimposing the distance impedance and connectivity impedance is used as the edge weight between adjacent nodes.

[0024] S23. Sort the nodes within the search radius of the nodes to be avoided according to betweenness centrality. For any node within the search radius, delete it from the directed graph G. Based on the edge weights between nodes involved in all paths from the source node to the target node, redetermine the shortest path from the source node to the target node and add it to the avoidance node list. Then, by calculating the similarity between the baseline route and the shortest path from the source node to the target node, determine whether it is necessary to update the search radius to obtain the avoidance node list again. Finally, obtain the route planning result.

[0025] Preferably, the calculation method for the total impedance in step S22 is as follows:

[0026] S221, Assume node i For the i-th grid, [S1(lon,lat),…,S N [(lon,lat)] is a node i The node contains N ship position information. i The central longitude and central latitude are respectively,

[0027]

[0028] Among them, node i_center_lon and node i_center_lat They are nodes iThe center longitude and center latitude; M is the length of the grid connected set L;

[0029] S222, Calculate adjacent nodes i and node j The total impedance between the two adjacent nodes is used as the edge weight between them.

[0030] ω i,j =10000*(θ) i,j *σ i,j )

[0031]

[0032] Where, ω i,j θ i,j σ i,j Each node i and node j The total impedance, connectivity impedance, and distance impedance between them.

[0033] Preferably, step S23 specifically includes the following:

[0034] S231, Given the node to be avoided (lon) suze ,lat suze The route from the source node to the target node is selected as the baseline route. base And set the search radius R;

[0035] S232, Calculate the avoidance node (lon) suze ,lat suze The search area within a radius R is used to obtain the set of search grid cells. list (node1,…,node k );

[0036] S233, Calculate the search grid set cell list (node1,…,node k The betweenness centrality of all nodes within the search grid is calculated, and all nodes within the search grid are sorted in descending order of betweenness centrality to obtain the betweenness centrality sorted set, cell. list_betweenness (node1,…,node k );

[0037]

[0038] Among them, c betweenness (node i ) is a node i Betweenness centrality; σst σ represents the number of shortest paths between the source node (source) and the target node (target). st (node i ) represents the nodes visited. i The number of shortest paths;

[0039] S234, For betweenness centrality ordered set cell list_betweenness node i Remove the node from the directed graph G. Based on the edge weights between nodes involved in all paths from the source node to the target node, redetermine the shortest path (route) from the source node to the target node. tmp Add the node to the avoidance node list AL; if the shortest path route tmp With the baseline route base If the similarity between nodes is less than the similarity threshold, output the avoidance node list AL and obtain the route planning result; otherwise, update the search radius by a preset distance and return to step S232.

[0040] Preferably, the path with the smallest edge weight among all paths between the source node and the target node is the shortest path between these two nodes.

[0041] The beneficial effects of this invention are: 1. The method of this invention utilizes large-scale model knowledge to obtain port latitude and longitude information, reducing additional retrieval costs; it also utilizes the coding capabilities of the large-scale model to generate HTML display effects for the path. 2. The method of this invention can be used for path planning between two ports that have never been in transit before; the route is based on the actual trajectory and will not cross obstacles such as land or islands. 3. The method of this invention can more accurately avoid strait passages. Attached Figure Description

[0042] Figure 1 This is a flowchart of the method in an embodiment of the present invention;

[0043] Figure 2 This is a schematic diagram illustrating the large model calling route planning in an embodiment of the present invention;

[0044] Figure 3 This is a sample image of global data divided into rectangular grids with 1 degree as the unit, as described in this embodiment of the invention.

[0045] Figure 4 This is a schematic diagram of adjacent rectangular grid data in an embodiment of the present invention;

[0046] Figure 5 This is a schematic diagram of selecting the Antwerp to Shanghai route as the baseline route in an embodiment of the present invention;

[0047] Figure 6 This is a schematic diagram illustrating the determination of the shortest path based on edge weights in an embodiment of the present invention;

[0048] Figure 7 This is a schematic diagram illustrating parameter and tool analysis of a large model in an embodiment of the present invention;

[0049] Figure 8 This is a schematic diagram of the default route planning for the New York to Shanghai route in an embodiment of the present invention;

[0050] Figure 9 This is a schematic diagram of the route from New York to Shanghai that avoids the Panama Canal in an embodiment of the present invention;

[0051] Figure 10 This is a schematic diagram of the route from New York to Shanghai in an embodiment of the present invention, after avoiding the Panama Canal and the Cape of Good Hope;

[0052] Figure 11 This is a schematic diagram of the default route planning for the Antwerp-Shanghai route in an embodiment of the present invention;

[0053] Figure 12 This is a schematic diagram of the Antwerp-Shanghai route planning in an embodiment of the present invention, which avoids the Suez Canal. Detailed Implementation

[0054] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0055] like Figure 1 As shown, this embodiment provides a method for ship-port route planning that integrates large-scale model knowledge, AIS data, and geographic raster data. By utilizing large-scale model knowledge and external tool parsing capabilities, user queries are analyzed. The parsed parameters and tools are then input into the route planning service. The route planning results are then analyzed by the large-scale model to generate an HTML rendering file of the trajectory. This method integrates large-scale model knowledge with ship-port route planning, enriching the application scenarios of route planning and achieving better interactive and visual effects.

[0056] The method specifically includes the following:

[0057] I. Route Planning Analysis Based on Large Model

[0058] The system uses a large model to analyze user input regarding flight route planning questions, obtains the question information, uses the large model to perform tool analysis on the question information to determine the flight route planning service that needs to be invoked, and uses the large model to perform parameter analysis on the flight route planning service to determine the corresponding parameters required.

[0059] Take a user's question about route planning as an example. The user asks, "Plan a route from Shanghai Port to New York Port." The large model parses the question text to obtain the information "route planning." Since the general large model knowledge base does not have a "route planning" function, it then performs "tool parsing"; according to... Figure 2 The tool description in the code determines which tool (i.e., the route planning service) to invoke. Simultaneously, the large model parses the parameters based on the tool description, discovering that the tool requires the latitude and longitude of two ports as parameters. Applying the large model's knowledge, it returns a list of the latitude and longitude of Shanghai and New York ports requested by the user, which is then input into the tool.

[0060] II. Route Planning Service Invocation

[0061] Enter the corresponding parameters to call the relevant route planning service and obtain the route planning results.

[0062] 2.1 To calculate the paths between ports, the map with longitudes from -180° to 180° and latitudes from -90° to 90° is divided into rectangular grids in 1° increments, resulting in a set of rectangular grids, denoted as follows:

[0063] Grid={(WKT,id,left,top,right,bottom)}

[0064] Wherein, Grid represents a set of rectangular grids; WKT represents the WKT format region representation of the grid; id is the unique identifier of the grid; left, right, top, and bottom are the left, right, top, and bottom edges of the grid, respectively. The grid data after partitioning is shown in Table 1. For grid partitioning and local sampling methods, please refer to the appendix. Figure 3 .

[0065] Table 1. Examples of global data divided into rectangular grids using 1 degree as the unit.

[0066]

[0067] 2.2 This invention is applicable to route planning between two ports. Therefore, for the originating port and the destination port, a weighted directed graph G needs to be defined.

[0068] G = (V, E, W)

[0069] V = {v} (1.plon,plat) ,v (2.plon,plat) ,…,v (n.plon,plat)}

[0070] E={e ij}

[0071] W = {w ij}

[0072] Where V is the set of grids between the origin port and the destination port; n is the grid ID; plon and plat are the longitude and latitude of the port, respectively; E is the set of edges; e ij For the i-th grid v (i.plon,plat) and the j-th grid v (j.plon,plat) The edges between; W is the set of edge weights; w ij For grid v (i.plon,plat) and v (j.plon,plat) The boundary weights between them.

[0073] 2.3. Connectivity between two grid cells can be calculated using ship AIS data and grid data. Current research methods determine connectivity by whether a ship's position falls within a grid cell and whether the grid cells are adjacent. However, this method is prone to generating incorrect connections, such as... Figure 4 In this case, the data in grid 5 and the data in grid 7 do not come from the same ship, so grid 5 is not connected to grid 7.

[0074] To address the problem of accurately determining the true connectivity between two grids, this invention designs a rectangular grid connectivity determination algorithm based on AIS data: For a directed graph G with an edge set E, and using a ship AIS data set S and a rectangular grid set Grid as input, when ship... k (t y ,lon y ,lat y In Grid u In China, ship k (t y+1 ,lon y+1 ,lat y+1 In Grid v In the middle, and Grid u and Grid v If there is at least one identical edge, then (u,v,1) is added to the grid connected set L. Traverse all ships in the ship AIS data set S and output the grid connected set L(source,target,value).

[0075]

[0076] Among them, ship k For the ship with ship number k, k = 1, 2, ..., K; k [(t y,lon y ,lat y )] for ship k At the y-th timestamp t y Longitude of the location y and latitude y ;source is the source node; target is the target node; value is the connection value, each connection is counted as 1; Grid u and Grid v Let u and v be the u-th and v-th grids of the rectangular grid set Grid, respectively. The specific algorithm flow is shown in Table 2.

[0077] Table 2. Mesh Connectivity Algorithm

[0078]

[0079] Step 2.3 is the specific implementation of the edge set E in the directed graph G in step 2.2.

[0080] 2.4 When calculating the weights of adjacent grids, the distance and connection probability between the two grids need to be considered to ensure that the planned route selects the most experienced route possible. Route planning aims for the shortest distance; therefore, the higher the connection probability between two grids, the lower the impedance during calculation. When calculating the distance between two grids, the distance between the grid's center point cannot be directly used as the distance between the two grids. Firstly, the points on the trajectory are unevenly distributed within the grid; secondly, using the center point may result in crossing land. Considering these factors, this invention designs a method for calculating the edge weights between nodes: For the edge weight set W in a directed graph G, based on the length of the connected grid set, the distance impedance and connectivity impedance between adjacent nodes are calculated. The total impedance obtained by superimposing the distance impedance and connectivity impedance is used as the edge weight between adjacent nodes. The specific process is as follows:

[0081] (1) Assume node i For the i-th grid, [S1(lon,lat),…,S N [(lon,lat)] is a node i The node contains N ship position information. i The central longitude and central latitude are respectively,

[0082]

[0083] Among them, node i_center_lon and node i_center_lat They are nodes i The center longitude and center latitude; M is the length of the grid connected set L.

[0084] (2) Calculate adjacent nodesi and node j The total impedance between the two adjacent nodes is used as the edge weight between them.

[0085] ω i,j =10000*(θ) i,j *σ i,j )

[0086]

[0087] Where, ω i,j θ i,j σ i,j Each node i and node j The total impedance, connectivity impedance, and distance impedance between them.

[0088] Step 2.4 is the specific calculation method for the set of edge weights W in the directed graph G in step 2.2.

[0089] 2.5 Shipping route planning often requires avoiding nodes such as the Panama Canal, the Cape of Good Hope, and the Suez Canal. Given latitude and longitude (which may not necessarily represent the most important nodes), it's necessary to determine the importance of surrounding nodes, remove important nodes from the network, and re-plan the route. To address this, this invention designs a core node and avoidance algorithm: Nodes within the search radius of the nodes to be avoided are sorted according to betweenness centrality. For any node within the search radius, it is deleted from the directed graph G. Then, based on the edge weights between nodes involved in all paths from the source node to the target node, the shortest path from the source node to the target node is re-determined and added to the avoidance node list. Next, by calculating the similarity between the baseline route and the shortest path from the source node to the target node, it is determined whether the search radius needs to be updated to re-obtain the avoidance node list; finally, the route planning result is obtained. The specific process is as follows.

[0090] Given the latitude and longitude of a node that needs to be avoided, such as the Suez Canal (lon suze ,lat suze Since the Suez Canal is the most crucial node connecting Europe to East Asia, the route from Antwerp to Shanghai was chosen as the baseline route. base route base like Figure 5 As shown.

[0091] (1) Given the node to be avoided (lon) suze ,lat suzeThe route from the source node to the target node is selected as the baseline route. base And set the search radius R;

[0092] (2) Calculate the avoidance node (lon) suze ,lat suze The search area within a radius R is used to obtain the set of search grid cells. list (node1,…,node k );

[0093] (3) Calculate the search grid set cell list (node1,…,node k The betweenness centrality of all nodes within the search grid is calculated, and all nodes within the search grid are sorted in descending order of betweenness centrality to obtain the betweenness centrality sorted set, cell. list_betweenness (node1,…,node k );

[0094]

[0095] Among them, c betweenness (node i ) is a node i Betweenness centrality; σ st σ represents the number of shortest paths between the source node (source) and the target node (target). st (node i ) represents the nodes visited. i The number of shortest paths;

[0096] (4) For the betweenness centrality ordered set cell list_betweenness node i Remove the node from the directed graph G. Based on the edge weights between nodes involved in all paths from the source node to the target node, redetermine the shortest path (route) from the source node to the target node. tmp Add the node to the avoidance node list AL; if the shortest path route tmp With the baseline route base If the similarity between nodes is less than the similarity threshold, the avoidance node list AL is output, and the route planning result is obtained; otherwise, the search radius is increased by a preset distance and the process returns to step S52. See Table 3 for details.

[0097] Table 3 Core Nodes and Avoidance Algorithms

[0098]

[0099]

[0100] Step 2.5 is to implement the avoidance function, which is to change the connection method of the edge set E in the directed graph G in step 2.2.

[0101] In this embodiment, the path with the smallest edge weight among all paths between the source node (source) and the target node (target) is the shortest path between these two nodes. See Appendix. Figure 6 In the diagram, 10 represents the edge weight between nodes A and B. Taking the path from A to B as an example, there are two paths: A->B with an edge weight of 10, and A->F->G->B with an edge weight of 11 + 17 + 16 = 44. Therefore, the shortest path is A->B. In traffic problems, edge weight is called impedance, and the shortest path is the path with the lowest impedance (i.e., the path with the smallest edge weight). When a yielding is required, such as yielding to point F along the way, the other path between A and B, A->F->G->B, will not be included in the calculation. The edge set E and edge weight set W of the directed graph G in step 2.2 must be recalculated to form new edge weights.

[0102] III. Display of Route Planning Results Based on Large Model

[0103] The route planning results are analyzed using a large-scale model to achieve route visualization. See the appendix for details on the tool analysis and result interpretation of the large-scale model. Figure 7 .

[0104] In this embodiment, the method of the present invention can achieve the following effects:

[0105] (1) Avoid the center point of the flight path crossing land.

[0106] The conventional route planning method yields the following results for the route planning from Tianjin Port (38.97075, 117.8143) to Bintulu Port (3.171322, 113.0419069): Figure 8 As shown, the center points of both the originating port and the destination port are on land. The method of this invention is as follows... Figure 8 As shown, both the planned starting and destination ports are located at sea, which is more in line with user habits.

[0107] (2) Route planning and node avoidance effect

[0108] (a) North America to China route planning and avoidance effect demonstration

[0109] Using the shipping route from New York Port to Shanghai Port as an example, we will plan shipping routes from North America to China (since the route crosses a 180° meridian, a scatter plot is used to display the route). The default generated route is as follows: Figure 9As shown, the default route passes through the Panama Canal and crosses the Pacific Ocean to China.

[0110] Adding avoidance points to the route planning that bypasses the Panama Canal, such as Figure 10 As shown, after choosing to avoid the Panama Canal, the route planning results show that the route passes through the Cape of Good Hope and the Strait of Malacca.

[0111] Adding avoidance points to route planning that bypasses the Panama Canal and the Cape of Good Hope, such as... Figure 11 As shown, after choosing to avoid the Panama Canal and the Cape of Good Hope, the route planning results show that the route passes through the Suez Canal and the Strait of Malacca.

[0112] (b) Demonstration of route planning and avoidance effects from Europe to China

[0113] Using the Antwerp-Shanghai route as an example, this section demonstrates route planning from Europe to China. The default generated route is as follows: Figure 12 As shown, the default route passes through the Suez Canal and the Strait of Malacca to China.

[0114] By incorporating avoidance nodes to circumvent the Suez Canal, the route planning results show that the route passes through the Cape of Good Hope and the Strait of Malacca after choosing to avoid the Suez Canal.

[0115] By adopting the above-disclosed technical solution of this invention, the following beneficial effects are obtained:

[0116] This invention provides a method for planning ship routes between ports that integrates large-scale model knowledge. This method utilizes the knowledge of large-scale models to obtain port latitude and longitude information, reducing additional retrieval costs; it also leverages the coding capabilities of large-scale models to generate HTML display effects for the routes. This method can be used for route planning between two ports that have never been connected by sea, and the routes are based on real trajectories, avoiding obstacles such as land and islands. This method also enables more accurate avoidance of strait passages.

[0117] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for planning a ship inter-port route by fusing large model knowledge, characterized in that: Comprising the following steps, S1, route planning analysis based on large model: using a large model to analyze the user's input question about route planning, obtaining the question information, using the large model to analyze the question information to determine the route planning service to be called, using the large model to analyze the parameters of the route planning service to determine the corresponding parameters required; S2, route planning service calling: input the corresponding parameters to call the corresponding route planning service to obtain the route planning result; S3, route planning result display based on large model: using a large model to analyze the route planning result to realize line visualization; The route planning service includes the following contents, S21, the map from longitude-180° to 180° and latitude-90° to 90° is divided into a rectangular grid with 1° as a unit to obtain a rectangular grid set, denoted as, ； wherein, is a rectangular grid set; is a unique number of the grid; is a format area representation; is a unique number of the grid; , , , are left, right, top and bottom edges of the grid, respectively; S22, for the two ports of the starting port and the destination port, define a weighted directed graph , ； ； ； ； wherein, is a set of grids between the start port and the destination port; is a grid number; and are the longitude and the latitude of the port, respectively; is a set of edges; is the grid and the grid between them; is a set of edge weights; is the edge weight between the grid and . For a directed graph with edge set , with ship AIS data set and rectangular grid set as input, when in , in , and and have at least one same edge, then is added to grid connectivity set , traversing all ships in ship AIS data set , outputting grid connectivity set ; ； wherein, is a ship number for a ship, ; is the longitude and latitude at which the ; is a source node; is a target node; is a connection value, each connection is recorded as 1; and are the first and second grids of a rectangular grid set ;​​​​​​ For directed graphs The set of edge weights in Based on the length of the connected set of the mesh, the distance impedance and connectivity impedance between adjacent nodes are calculated, and the total impedance obtained by superimposing the distance impedance and connectivity impedance is used as the edge weight between adjacent nodes. S23, the nodes in the search radius range of the node needing to avoid are sorted according to the betweenness centrality, for any node in the search radius range, the node is deleted from the directed graph , the shortest path from the source node to the target node is re-determined according to the edge weight between the nodes involved in all paths from the source node to the target node , and the node is added to the avoidance node list, and then whether the search radius needs to be updated to reacquire the avoidance node list is determined by calculating the similarity between the reference route and the shortest path from the source node to the target node ; finally, a flight route planning result is acquired; The calculation method of the total impedance in step S22 is specifically, S221, assuming the node is the first grid, is the Nth grid, is the Nth grid, is the Nth grid, the central longitude and the central latitude of the Nth grid, respectively, ； ； wherein, and are respectively the central longitude and the central latitude; is the length of the grid-connected set of the grid-connected set S222, compute the total impedance between adjacent nodes and and use it as the edge weight between the two adjacent nodes. ； ； ； wherein, , , are the total impedance, the connectivity impedance, the distance impedance between the nodes and , respectively. Step S23 specifically includes the following contents, S231, Given the nodes that need to be avoided ( Select source node To the target node The route is used as the baseline route. and set the search radius. ; S232、computing an avoidance node search radius range a grid within the search radius range, obtaining a search grid set ; S233、calculate the betweenness centrality of each node in the search grid set The betweenness centrality of each node in the search grid set is calculated, and the nodes in the search grid set are sorted in descending order of betweenness centrality to obtain a betweenness centrality sorted set ; ； wherein, is the betweenness centrality of the node ; is the number of shortest paths between the source node and the target node ; is the number of shortest paths passing through the node ; S234、for the betweenness centrality ranking set nodes in , delete the node from the directed graph , according to the edge weight between the nodes involved in all paths from the source node to the target node , re-determine the shortest path from the source node to the target node , add the node to the avoidance node list ; if the similarity between the shortest path and the reference route is less than the similarity threshold, output the avoidance node list , and obtain the flight route planning result; otherwise, increase the search radius by a preset distance, and return to step S232.

2. The method of claim 1, wherein: Source node To target node The path with the minimum edge weight among all paths between the source node and the target node is the shortest path between the two nodes.

Images