A method and system for analyzing evolution of seabed erosion and deposition

Abstract

The present application relates to the technical field of evolution modeling, in particular to a seabed erosion and deposition evolution analysis method and system, in the present application, the multidimensional adjacency relationship between nodes is recognized through a graph attention network, three parameters of the slope direction included angle, the elevation trend and the flow direction consistency index are jointly judged and the influence degree is quantified, the expression ability of the influence factors of the complex boundary direction is improved, the direction is not judged according to the fixed rule, but based on the quantitative comparison of the flow direction and the slope component, the nodes of the dominant erosion path are screened and the direction trajectory is rearranged, the time sequence expression has the dynamic reconstruction ability, combined with the slot corner rate ratio calculation, the angle mutation section is clustered and screened by means of the greedy algorithm, the node set representing the mutation trend is formed, the extraction stability of the local extreme erosion behavior is enhanced, the connection logic expression of the physical state between nodes, the abnormal extraction efficiency of local evolution change and the expression continuity and overall recognition stability of the erosion and deposition trend result are effectively improved.

Description

Technical Field

[0001] This invention relates to the field of evolution modeling technology, and in particular to an evolution analysis method and system for seabed erosion and deposition. Background Technology

[0002] The field of evolution modeling technology mainly focuses on modeling changes in time and space, simulating and predicting dynamic processes, and emphasizing the use of mathematical or numerical models to simulate state variables in real-world systems over time, thereby understanding development patterns and evolutionary trends.

[0003] The purpose of this method for analyzing the evolution of seabed erosion and deposition is to achieve systematic modeling and computable prediction of seabed morphological changes, so as to support the scientific assessment of the impact of engineering activities on the seabed environment, provide a quantifiable data foundation and evolution trend judgment for marine engineering planning, waterway maintenance and ecological protection, realize the structured, modeled and visualized processing of complex natural erosion and deposition processes, and improve the ability to respond to dynamic changes.

[0004] Existing technologies exhibit strong temporal dependence when constructing dynamic process representations. By performing differential processing on continuous elevation snapshot data, it is difficult to construct a directional spatial logical structure in a single temporal phase, resulting in insufficient ability to identify abrupt changes and difficulty in accurately defining the boundaries of structural changes. In multi-node state judgment, a single parameter is usually used as the judgment benchmark, without forming a multi-dimensional joint judgment method, which limits the identification of key trend directions in complex changing environments. Furthermore, trajectory judgment is often constrained by preset evolution direction rules, lacking dynamic screening and direction rearrangement mechanisms, which can easily lead to discrepancies between the node path reconstruction results and the actual scour path. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of existing technologies by proposing a method and system for analyzing the evolution of seabed erosion and deposition.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a method for analyzing the evolution of seabed erosion and deposition, comprising the following steps:

[0007] S1: Acquire single-phase seabed grid elevation data through sensors, calculate elevation difference in the beach and trough area and extract slope vector, calculate the included angle of adjacent units and select edge pairs that meet the directional conditions, and establish a directional connection weighted network graph structure.

[0008] S2: Based on the directional connection weighted network graph structure, a graph attention network is used to extract the adjacency relationship of the slope boundary nodes, and to determine whether the slope direction, elevation increase / decrease trend, and flow direction consistency indicators all meet the threshold. By superimposing the influence of directional clusters, a group of directional dominant influence indicators is generated.

[0009] S3: Based on the aforementioned direction-dominant influence index group, select nodes in the control area grid where the flow direction component is greater than the slope component, extract the evolution path trajectory and rearrange the direction sequence to establish a temporal scour and sedimentation change trend array;

[0010] S4: Based on the time-series scour and sedimentation trend array, call the continuous angle change value of the inflection point of the channel and calculate the ratio of adjacent turning rate. Use a greedy algorithm to screen the abrupt change segment and identify the scour starting point. Generate the set of nodes of the dominant path of scour and sedimentation by clustering the extreme value sequence of the trajectory chain.

[0011] S5: Based on the set of nodes of the dominant scour and siltation paths, determine the trend of the angle change of the continuous path, extract monotonic segments and calculate the sequence of angle difference and length values, aggregate and reorganize the path structure according to the segments, and establish a partition map of the scour and siltation evolution characteristics.

[0012] As a further aspect of the present invention, the specific steps for generating the directed connection weighted network graph structure are as follows:

[0013] The elevation data of the seabed grid in a single time phase is acquired by the sensor. The elevation values ​​of each grid unit in the beach and trough area are extracted point by point and the index of the four neighboring units is extracted. The slope direction vector is generated by calculating the elevation difference between the current node and the four neighboring nodes and dividing it by the horizontal spacing. A local slope direction vector set is generated.

[0014] Based on the local slope direction vector set, the cosine angle between the direction vectors of each pair of connected units is calculated and it is determined whether it is less than the limited angle threshold. By filtering out grid unit pairs that meet the standard of similar directions and establishing effective adjacency relationships, a set of grid unit pairs with consistent directions is obtained.

[0015] Based on the set of grid cells with consistent orientation, a connection edge is constructed for each pair of nodes and a direction attribute is attached. The edge weight is generated by the reciprocal of the cosine angle and a connection matrix is ​​constructed in a unified manner to establish a directional connection weighted network graph structure.

[0016] As a further aspect of the present invention, the specific steps for generating the directional dominant influence index group are as follows:

[0017] Based on the directional connection weighted network graph structure, a graph attention network is used to filter the boundary nodes of the bank slope and extract the indexes of all adjacent nodes. The direction change values ​​of adjacent slopes, the elevation increase / decrease difference between nodes and the angle of flow direction are extracted respectively. By constructing a set of three parameters, the set of bank slope node association indicators is obtained.

[0018] Based on the set of associated indicators for the bank slope nodes, each of the three indicators for each group of nodes is judged item by item, and the node pairs that simultaneously meet all threshold conditions are marked. The node pairs that meet the requirements are numbered and classified in sequence, and the corresponding connection path number set is extracted to obtain the set of nodes with high consistency direction.

[0019] Based on the highly consistent directional node set, the reciprocal of the directional angle between each pair of nodes, the mean of the flow direction distance, and the normalized value of the slope difference are taken and then superimposed. Continuous path segments are aggregated and extracted by sequential numbering to generate a group of directional dominant influence indicators.

[0020] As a further aspect of the present invention, the specific steps for generating the time-series siltation change trend array are as follows:

[0021] Based on the dominant influence index group, the magnitude of the flow direction value and the slope direction value of each node is extracted. The two values ​​are compared by the difference operation and the judgment condition that the flow direction component is greater than the slope component is set. Nodes that meet the condition are screened to obtain the dominant flow direction node set.

[0022] Based on the set of dominant flow direction nodes, the nodes are connected by time sequence numbering, and the spatial distance and directional change angle of adjacent nodes are calculated according to the node index. The node index sequence is generated by rearranging the connection direction order, and a set of scour and siltation trajectory direction sequence is generated.

[0023] Based on the set of scour and sedimentation trajectory directions, the time difference sequence and direction vector change value of each node are merged, and the change angle sequence of each node in the continuous path is sorted and cumulatively rearranged to establish a temporal scour and sedimentation change trend array.

[0024] As a further aspect of the present invention, the specific steps for generating the set of nodes representing the dominant paths of scouring and sedimentation are as follows:

[0025] Based on the time-series scouring and silting trend array, continuous node direction vector groups are extracted and the angle change values ​​are calculated sequentially. The turning rate sequence is generated by dividing the difference by the time step and the amplitude mutation judgment rule is set to obtain the trajectory direction mutation ratio sequence.

[0026] Based on the trajectory direction mutation ratio sequence, a greedy algorithm is used to perform mutation segment division, mutation point indexing and location, and extraction of adjacent point sequences. By accumulating and summing continuous angle difference segments, segments with mutation amplitude greater than a set value are selected and their path position indices are recorded to obtain a set of scour mutation path segments.

[0027] Based on the set of scour mutation path segments, the maximum value of the set of flow direction angles of nodes within the segment is located and it is determined whether there is a reversal point of angle difference on both sides. The nodes where the reversal point is located are spatially reorganized and numbered to generate a set of nodes for the dominant scour and sedimentation paths.

[0028] As a further aspect of the present invention, the greedy algorithm is described by the formula:

[0029]

[0030] Wherein: T sψ represents the overall mutation intensity value of the mutated fragment. i Let |ψ| represent the tangential direction angle of the i-th trajectory point. i+1 -ψ i | represents the difference in direction angle between adjacent trajectory points, reflecting the degree of path turning, e i Let d represent the seabed elevation value of the i-th trajectory point. i,i+1 This represents the horizontal distance between trajectory points i and i+1. κ represents the rate of change of elevation per unit horizontal distance between adjacent trajectory points, reflecting abrupt changes in local slope. i |κ represents the local curvature value at the i-th trajectory point. i+1 -κ i | represents the magnitude of curvature change between adjacent points, s i This represents the unit path velocity at the i-th trajectory point. This represents the average flow velocity across all trajectory points within the current mutation fragment. The value represents the point velocity offset, and μ1, μ2, μ3, and μ4 represent the weighting coefficients of abrupt change in direction, elevation gradient, curvature fluctuation, and velocity offset factor, respectively.

[0031] As a further aspect of the present invention, the specific steps for generating the aforementioned scour-deposition evolution characteristic structural partition map are as follows:

[0032] Based on the set of nodes of the dominant scour and sedimentation path, the direction vectors of continuous nodes are extracted and the angle between adjacent nodes is calculated. The direction change pattern is identified by the positive and negative sign sequence of the angle difference, and the start and end positions of the monotonic continuous sequence are marked to generate a set of monotonic direction path segments.

[0033] Based on the monotonic direction path segment set, the direction angle difference and the distance value between adjacent nodes of each node pair are extracted in the order of the path segments. An angle sequence and a length sequence are constructed respectively. The path angle length parameter sequence set is obtained by grouping by node number and merging with the code.

[0034] Based on the path angle length parameter sequence set, the angle sequence fluctuation value and length spacing change value are extracted according to the node number within the segment. After classifying the value difference range within continuous segments, the structural segment numbers are rearranged, and paths of the same type are merged and aggregated to establish a structural partition map of scour and sedimentation evolution characteristics.

[0035] A system for analyzing the evolution of seabed erosion and deposition, the system being used to execute the aforementioned method for analyzing the evolution of seabed erosion and deposition, the system comprising:

[0036] Elevation analysis module: Based on single-phase seabed grid elevation data, extract the elevation values ​​of adjacent grid cells in the beach and channel area, calculate the elevation difference and slope vector direction for each pair, obtain the slope vector angle value and determine whether it falls within the set angle range, filter to form unidirectional edge pairs, and use the elevation difference and slope direction as edge weight values ​​to establish a directional edge weight graph structure.

[0037] Edge-based graphing module: Based on the aforementioned directional edge weight graph structure, a graph attention network is used to extract directional edge relationships from the set of adjacent nodes of each node. It determines whether the slope direction angle, elevation increase / decrease continuity indicator, and flow direction consistency ratio of each connecting edge meet the corresponding thresholds. All directional edge groups that meet the conditions are filtered out, and the cumulative connection frequency of directional clusters in the graph is counted. An index set is generated according to the frequency to form a directional dominant influence index group.

[0038] Dominant Direction Module: Based on the dominant influence index group, extract the flow direction component value and slope component value of each grid node in the control area, filter all node numbers where the flow direction component is greater than the slope component, read the flow direction path in multiple time periods for each numbered node, sort and reorganize the flow direction sequence, summarize the structural change trend of all sequences in the time dimension, and establish a direction evolution path sequence group.

[0039] Path evolution module: Based on the directional evolution path sequence group, extract the inflection point numbers of the grooves in each path, calculate the ratio of the angle difference between each pair of inflection points to the rate of change of continuous turning angles, determine whether the rate of change ratio exceeds the threshold, mark the first node of the path mutation segment as the scour starting point, read all extreme points to form a path chain, call the greedy algorithm to cluster the extreme nodes of the path, output the cluster center position, and form a scour and siltation distribution node set;

[0040] Structural partitioning module: Based on the scour and sedimentation distribution node set, determine whether the direction of the angle change between adjacent nodes in the path segment is consistent, filter out all monotonic angle path segments, read the angle difference sequence and path length value sequence in each path segment respectively, remap the path structure order according to the combination value of angle and length, aggregate all segment paths to form a path fragment set, and establish a scour and sedimentation evolution partition map.

[0041] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0042] In this invention, a graph attention network is used to identify multidimensional adjacency relationships between nodes. The three parameters of slope direction angle, elevation trend, and flow direction consistency index are jointly judged and the degree of influence is quantified, thereby improving the ability to express the influence factors of complex boundary directions.

[0043] In this invention, direction is not determined by fixed rules, but by quantitative comparison of flow direction and slope components, the dominant scour path nodes are selected and their directional trajectories are rearranged, so that the time series expression has dynamic reconstruction capability.

[0044] In this invention, the ratio of slot angle rate is calculated, and a greedy algorithm is used to cluster and screen angle change segments to form a set of nodes representing the change trend, thereby enhancing the stability of extracting local extreme scouring behavior.

[0045] In this invention, path reconstruction is achieved by combining the included angle sequence and the path length value, thereby constructing a structural map that can identify evolution patterns. This effectively improves the connection logic expression of the physical state between nodes, the efficiency of extracting anomalies in local evolution changes, and the continuity of the expression of scour and siltation trend results as well as the overall identification stability. Attached Figure Description

[0046] Figure 1 This is a schematic diagram of the workflow of the present invention;

[0047] Figure 2 This is a system flowchart of the present invention. Detailed Implementation

[0048] 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 and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0049] In the description of this invention, it should be understood that the terms "length," "width," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientation or positional relationships, are based on the orientation or positional relationships shown in the accompanying drawings and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, in the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0050] Example 1

[0051] Please see Figure 1 This invention provides a technical solution: a method for analyzing the evolution of seabed erosion and deposition, comprising the following steps:

[0052] S1: Acquire single-phase seabed grid elevation data through sensors, calculate elevation difference in the beach and trough area and extract slope vector, calculate the included angle of adjacent units and select edge pairs that meet the directional conditions, and establish a directional connection weighted network graph structure.

[0053] S2: Based on the directional connection weighted network graph structure, a graph attention network is used to extract the adjacency relationship of the boundary nodes of the bank slope, and to determine whether the slope direction, elevation increase and decrease trend, and flow direction consistency indicators all meet the threshold. By superimposing the influence of directional clusters, a group of directional dominant influence indicators is generated.

[0054] S3: Based on the direction-dominant influence index group, select nodes in the control area grid where the flow direction component is greater than the slope component, extract the evolution path trajectory and rearrange the direction sequence to establish a temporal scour and sedimentation change trend array;

[0055] S4: Based on the time-series scour and sedimentation trend array, call the continuous angle change value of the inflection point of the channel and calculate the ratio of adjacent turning rate. Use a greedy algorithm to screen the abrupt change segment and identify the scour starting point. Generate the set of nodes of the dominant path of scour and sedimentation by clustering the extreme value sequence of the trajectory chain.

[0056] S5: Based on the node set of the dominant path of scour and siltation, determine the trend of the angle change of the continuous path, extract the monotonic segment and calculate the angle difference and length value sequence, aggregate and reorganize the path structure according to the segment, and establish a partition map of the scour and siltation evolution characteristics.

[0057] The specific steps for generating a directed connection weighted network graph structure are as follows:

[0058] The elevation data of the seabed grid in a single time phase is acquired by the sensor. The elevation values ​​of each grid unit in the beach and trough area are extracted point by point and the index of the four neighboring units is extracted. The slope direction vector is generated by calculating the elevation difference between the current node and the four neighboring nodes and dividing it by the horizontal spacing. A local slope direction vector set is generated.

[0059] Based on the local slope direction vector set, the cosine angle between the direction vectors of each pair of connected units is calculated and it is determined whether it is less than the limited angle threshold. By filtering out grid unit pairs that meet the standard of similar directions and establishing effective adjacency relationships, a set of grid unit pairs with consistent directions is obtained.

[0060] Based on the set of grid cells with consistent orientation, a connection edge is constructed for each pair of nodes and a direction attribute is attached. The edge weight is generated by the reciprocal of the cosine angle and a connection matrix is ​​constructed in a unified manner to establish a directional connection weighted network graph structure.

[0061] Based on single-temporal seabed grid elevation data acquired by sensors, a two-dimensional index mapping method is used to assign row and column numbers to each grid cell in the beach and channel area. After extracting the number value of each cell, an index of four neighboring cells is constructed. The elevation values ​​of the central cell and the four neighboring cells are read, and the elevation difference is calculated sequentially. The obtained difference is divided item by item with a fixed horizontal spacing of 2 meters to obtain a set of slope component values. Four directional unit vectors are extracted according to the orientation setting. Each directional unit vector is multiplied dimension by dimension with the corresponding component value and then summed to obtain the directional vector coordinate components. These components are then uniformly normalized to generate a local slope direction vector set.

[0062] Based on the local slope direction vector set, the direction relationship between adjacent grid cells is calculated using the vector cosine calculation method. The three-dimensional coordinate components of each pair of direction vectors are read, and the multiplication and summation of the component values ​​of each dimension are performed in sequence. The square root operation is then performed on the sum of the squares of the components of each dimension of the two direction vectors. The obtained values ​​are combined and processed by ratio, and the inverse trigonometric function operation is performed to obtain the angle value. The obtained angle value is uniformly judged to be less than the set threshold of 20 degrees. All the number combinations that meet the conditions are extracted and written into the valid cell pair record table to obtain the set of grid cell pairs with consistent direction.

[0063] Based on the set of oriented grid cells, a connection edge is constructed for each pair of nodes and a direction attribute is assigned. The inverse weighting function method is used to read the direction angle value corresponding to each pair of nodes. The inverse angle value processing is performed sequentially to form an initial weight value array. All edge weight values ​​are normalized and standardized with a uniform length unit. The direction is set to point from the node with the smaller index value to the node with the larger index value. When establishing edge connection relationships, a direction label is attached and the original node number is retained. After constructing all directed edge record tables, the contents of the node set, edge set and weight set are merged and uniformly written into the connection matrix array to establish a directional connection weight network graph structure.

[0064] The specific steps for generating the direction-dominant indicator group are as follows:

[0065] Based on the directed connection weighted network graph structure, a graph attention network is used to filter the boundary nodes of the bank slope and extract the indexes of all adjacent nodes. The direction change values ​​of adjacent slopes, the elevation increase / decrease difference between nodes and the angle of flow direction are extracted respectively. By constructing a set of three parameters, the set of bank slope node association indicators is obtained.

[0066] Based on the set of association indicators for the riverbank nodes, each of the three indicators of each group of nodes is judged item by item and the node pairs that simultaneously meet all threshold conditions are marked. The node pairs that meet the requirements are numbered and classified in sequence, and the corresponding connection path number set is extracted to obtain the set of nodes with high consistency direction.

[0067] Based on a highly consistent set of directional nodes, the reciprocal of the directional angle between each pair of nodes, the mean of the flow distance, and the normalized value of the slope difference are taken and then superimposed. Continuous path segments are aggregated and extracted by sequential numbering to generate a group of directional dominant influence indicators.

[0068] Based on the directed connection weighted network graph structure, a graph attention network is used for boundary node selection. Specifically, a multi-head attention mechanism initialization operation is performed on all nodes in the graph, with 4 heads, 64 input dimensions, and 16 output dimensions. The feature value of each node is linearly transformed with the feature value corresponding to the index of its neighboring nodes. The LeakyReLU activation function is called to activate the transformation result, with a negative slope parameter of 0.2. The activation result is then subjected to softmax normalization to obtain the attention coefficient matrix. The matrix is ​​used to weight the slope direction change value, elevation difference value, and flow direction angle value corresponding to the connection edge between nodes, and written into a standardized parameter table to construct a three-item structured parameter set, generating a set of bank slope node association indicators.

[0069] Based on the set of correlation indicators for slope nodes, threshold judgment operations are performed on three parameters in each group of nodes. The threshold values ​​for slope direction change are set to 15 degrees, elevation difference is set to 0.5 meters, and flow direction angle is set to 30 degrees. Each value in the parameter set is read one by one and compared with the above three threshold values. If all three parameters meet the condition of being less than or equal to the threshold, the corresponding node pair is marked as 1; otherwise, it is marked as 0. The number mapping operation is performed on all node pairs marked as 1 and the results are written into the path label set. The nodes are written into the output list in the order of the path number combination of node pairs to obtain the high consistency direction node set.

[0070] Based on a highly consistent directional node set, the corresponding directional angle value is read for each pair of node numbers and the reciprocal is taken. The flow distance value between nodes is read and the average of two values ​​is performed. The slope value of the node is read and the difference between the two values ​​is calculated and then normalized using the MinMaxScaler method. The three values ​​are directly superimposed to obtain the combined influence value. For all node combinations, the set of continuous path numbers is extracted according to the numbering order. The numbering intervals are merged according to the segment aggregation method and written into the node sequence output table to generate the directional dominant influence index group.

[0071] The specific steps for generating a time-series scouring and sedimentation trend array are as follows:

[0072] Based on the dominant influence index group, the magnitude of the flow direction value and the slope direction value of each node is extracted. The two values ​​are compared by the difference operation and the judgment condition that the flow direction component is greater than the slope component is set. Nodes that meet the condition are screened to obtain the set of dominant flow direction nodes.

[0073] Based on the set of dominant flow direction nodes, the nodes are connected by time sequence numbering, and the spatial distance and directional change angle of adjacent nodes are calculated according to the node index. The node index sequence is generated by rearranging the connection direction order, and a set of scour and sedimentation trajectory direction sequence is generated.

[0074] Based on the scour and siltation trajectory direction sequence set, the time difference sequence and direction vector change value of each node are merged, and the change angle sequence of each node in the continuous path is sorted and cumulatively rearranged to establish a time-series scour and siltation change trend array.

[0075] Based on the dominant influence index group, the vector magnitude difference method is used to extract the flow direction value and slope direction value of each node. Specifically, the three-dimensional vector of the flow direction and the three-dimensional vector of the slope direction are read from each node, the sum of squares of the three-dimensional components is calculated and the square root is taken as the magnitude value. The magnitude of the flow direction is subtracted from the magnitude of the slope direction and stored in the difference array. All values ​​in the difference array are logically judged to be greater than zero and a Boolean index is returned. For nodes with a Boolean value of True, the record number is extracted according to the original index order to generate the dominant flow direction node set.

[0076] Based on the set of dominant flow direction nodes, the node index and timestamp number are read using the sequence interpolation connection method. All nodes are sorted in ascending order according to time. The two-dimensional coordinate information of adjacent nodes is extracted and Euclidean distance calculation is performed in sequence. The cosine value of the vector direction of two adjacent nodes is processed and the inverse trigonometric function is performed to obtain the angle value. An array of time number and angle order is established. Then, the original node index table is reconstructed through the ascending sequence and the direction chain is generated by combining them in sequence. All chains are aggregated and merged into a fragment path set according to the node number to generate a set of scour and siltation trajectory direction sequences.

[0077] Based on the scour and sedimentation trajectory direction sequence set, the time number of adjacent nodes in each path is extracted using the path sequence stacking method, and the pairwise subtraction operation is performed and written into the time difference array. The direction vector coordinate value set between the corresponding nodes is read and the difference processing is performed to obtain the direction change vector. Then, the square summation and square root operation of each dimension of the change vector is performed to generate the direction change value sequence. The time difference array and the direction change value sequence are stacked in the path dimension. The direction change angle of each path is extracted and sorted in ascending order according to the path order. The sorted sequence is read and the item-by-item addition operation from beginning to end is performed to generate the cumulative change value. Finally, all path numbers and change sequences are combined to establish a temporal scour and sedimentation change trend array.

[0078] The specific steps for generating the node set of the dominant scour and sedimentation paths are as follows:

[0079] Based on the time-series scouring and silting trend array, the direction vector group of continuous nodes is extracted and the angle change value is calculated sequentially. The turning rate sequence is generated by dividing the difference by the time step and the amplitude mutation judgment rule is set to obtain the trajectory direction mutation ratio sequence.

[0080] Based on the trajectory direction mutation ratio sequence, a greedy algorithm is used to perform mutation segment division, mutation point indexing and location, and extraction of adjacent point sequences. By accumulating and summing continuous angle difference segments, segments with mutation amplitude greater than a set value are selected and their path position indices are recorded to obtain the scour mutation path segment set.

[0081] Based on the set of scour mutation path segments, the maximum value of the set of flow direction angles of nodes within the segment is located and it is determined whether there is a reversal point of angle difference on both sides. The nodes where the reversal point is located are spatially reorganized and numbered to generate a set of nodes for the dominant scour and sedimentation paths.

[0082] Based on the temporal scouring and silting trend array, the direction vector group of continuous nodes in the path is extracted by the direction vector angle difference calculation method. The direction vector values ​​of each pair of adjacent nodes are read sequentially, and the three-dimensional coordinate components are used to perform a dimension-wise multiplication operation and then accumulated. The magnitude value of each vector is read and divided to obtain the cosine value of the included angle. Then, the inverse trigonometric function is performed to convert it into an angle value, generating a sequence of included angle change values. Subsequently, the time number corresponding to the node is read and the difference between adjacent time numbers is calculated as the time step. The included angle change value of each group is divided with the corresponding time step and written into the turning rate sequence array. The turning rate judgment threshold is set to 10. It is judged whether the difference between adjacent rates exceeds the threshold. The node number that meets the condition is output, and the trajectory direction change ratio sequence is obtained.

[0083] Based on the mutation ratio sequence of the trajectory direction, a greedy algorithm is used to perform mutation segment division. Each mutation ratio in the sequence is read and the corresponding node index position is recorded in the path order. The initial clustering starting point index is set as the number of the first mutation node in the path. Starting from the current position, the node index is moved backward by one node each time and the current cumulative angle difference value of the cluster is recorded. If the cumulative value exceeds the threshold of 30, the process ends and the starting and ending node indices are saved. Then, the last node is used as the new starting point to continue the above process. The mutation segment division and index extraction of all path segments are performed in sequence. All mutation segment number sets and node combination path index tables are recorded to obtain the set of scour mutation path fragments.

[0084] Based on the set of scour mutation path segments, the directional extreme value determination method is used to read the index values ​​of all nodes in each segment, extract the flow direction angle set of the corresponding node, perform maximum value search on each set of angles, extract the node number where the maximum angle value is located, and then read the angle values ​​of the three adjacent nodes to the left and right of the maximum angle point in sequence and construct the difference sequence. It is determined whether the signs of adjacent difference values ​​are opposite. If there is an alternation of positive and negative, it is marked as a reversal point. All reversal points are numbered and uniformly encoded and their relative path segment positions are recorded. According to the numbering order, the two-dimensional coordinate positions of adjacent reversal points are extracted and rearranged, written into the node sequence table, and a set of nodes for the dominant scour and sedimentation paths is generated.

[0085] Greedy algorithm, according to the formula:

[0086]

[0087] Wherein: T s ψ represents the overall mutation intensity value of the mutated fragment. i Let |ψ| represent the tangential direction angle of the i-th trajectory point. i+1 -ψ i | represents the difference in direction angle between adjacent trajectory points, reflecting the degree of path turning, e i Let d represent the seabed elevation value of the i-th trajectory point. i,i+1 This represents the horizontal distance between trajectory points i and i+1. κ represents the rate of change of elevation per unit horizontal distance between adjacent trajectory points, reflecting abrupt changes in local slope. i |κ represents the local curvature value at the i-th trajectory point. i+1 -κ i | represents the magnitude of curvature change between adjacent points, s i This represents the unit path velocity at the i-th trajectory point. This represents the average flow velocity across all trajectory points within the current mutation fragment. The value represents the point velocity offset, and μ1, μ2, μ3, and μ4 represent the weighting coefficients of abrupt change in direction, elevation gradient, curvature fluctuation, and velocity offset factor, respectively.

[0088] Execution process: First, potential mutation regions are preliminarily identified from the trajectory direction mutation ratio sequence, and the start and end indices p and q of each mutation segment are determined. Then, for each segment, the required parameter information is extracted point by point, and the tangential direction angle ψ of adjacent trajectory points is calculated. i This leads to the directional mutation value |ψ i+1 -ψ i | is used to characterize the degree of abrupt changes in path turning, and simultaneously extracts the seabed elevation value e of the trajectory points. i Calculate the elevation difference and the corresponding horizontal distance d. i,i+1 ratio As the local slope change rate of the segment, the fitted circle formed by three local points is extracted, and the local curvature value κ of the trajectory point is calculated. i Obtain the curvature mutation amount |κ i+1 -κ i | is used to measure the severity of path curvature, while simultaneously reading the unit path velocity s from hydrodynamic simulations or observations. i Calculate the average flow rate of the current segment. offset The improved abrupt change intensity value T is obtained by multiplying the above four features by the corresponding weighting coefficients μ1, μ2, μ3, and μ4 respectively, and then summing them across the entire fragment to reflect the areas of severe scouring that may be caused by local velocity anomalies. s The weight coefficients were determined using a parameter tuning method based on the optimal accuracy of the validation set. By combining grid search with cross-validation, the coefficient combination that performed best in the mutation fragment identification task was selected to enhance the model’s comprehensive response to direction, slope, structure and dynamic anomalies, thereby accurately extracting the set of scour mutation path fragments.

[0089] The specific steps for generating the scour-deposition evolution characteristic structural zoning map are as follows:

[0090] Based on the node set of the dominant path of scour and sedimentation, the direction vector of continuous nodes is extracted and the angle between adjacent nodes is calculated. The direction change pattern is identified by the positive and negative sign sequence of the angle difference, and the start and end positions of the monotonic continuous sequence are marked to generate a set of monotonic direction path segments.

[0091] Based on the set of monotonic direction path segments, the difference in the directional angle between each pair of nodes and the distance between adjacent nodes are extracted in the order of the path segments. An angle sequence and a length sequence are constructed respectively. The codes are merged and processed by grouping by node number to obtain the set of path angle and length parameter sequences.

[0092] Based on the path angle length parameter sequence set, the angle sequence fluctuation value and length spacing change value are extracted according to the node number within the segment. After classifying the value difference range within continuous segments, the structural segment numbering is rearranged, and paths of the same type are merged and aggregated to establish a structural partition map of scour and sedimentation evolution characteristics.

[0093] Based on the node set of the dominant path of scour and siltation, the direction vector of continuous nodes in the path is extracted by the symbol sequence recognition method. The three-dimensional direction vector coordinate values ​​of each node and the next node are read and the angle between adjacent direction vectors is calculated. The angle extraction operation is completed by combining the dot product of the direction vectors with the magnitude value. All angle differences are recorded in the order of the path segments. The sign of each angle difference is read and converted into ±1 mark to construct the direction change symbol sequence. The sequence is traversed item by item using a sliding window and the interval of continuous symbols is found. The starting node number and ending node number of each interval of symbols are marked. All marked segment number lists are merged to generate a monotonic direction path segment set.

[0094] Based on a set of monotonic direction path segments, a path parameter decomposition method is used to read the set of node number pairs for each path segment, extract the direction vector for each pair of nodes and calculate the angle difference, and simultaneously read the spatial coordinates of the corresponding nodes and perform Euclidean distance calculation to generate an angle value sequence and a distance value sequence. Each item in the two sequences is bound to the corresponding node number pair to form a path parameter item. A path segment dictionary structure is established based on the path segment number. For each group of path segments, a key-value mapping relationship between angle and distance is constructed and encoded into a unified structured table. The encoding result is output and written into the number item list to obtain the path angle length parameter sequence set.

[0095] Based on the path angle length parameter sequence set, the interval classification and aggregation method is used to extract all angle change values ​​and corresponding length values ​​of each path segment according to the node numbering order within the segment. The range of each group of angle change values ​​is calculated and the range level is recorded. Then, the length difference calculation results between all adjacent nodes are read and classified into five segment level intervals. A unique category identifier is constructed according to the combination of angle range level and length interval level. The category grouping operation is performed on all path segment numbers and the numbers are sorted from smallest to largest. The segment numbers of the classified path set are rearranged and the partition index is completed. The results are written into the structural map index table to establish a structural partition map of scour and sedimentation evolution characteristics.

[0096] Please see Figure 2 A seabed erosion and deposition evolution analysis system, used to perform the aforementioned seabed erosion and deposition evolution analysis method, the system comprising:

[0097] Elevation analysis module: Based on single-phase seabed grid elevation data, extract the elevation values ​​of adjacent grid cells in the beach and channel area, calculate the elevation difference and slope vector direction for each pair, obtain the slope vector angle value and determine whether it falls within the set angle range, filter to form unidirectional edge pairs, and use the elevation difference and slope direction as edge weight values ​​to establish a directional edge weight graph structure.

[0098] Edge-based graphing module: Based on the directional edge weight graph structure, a graph attention network is used to extract the directional edge relationships of the adjacent node set of each node. It judges whether the slope direction angle, elevation increase / decrease continuity indicator, and flow direction consistency ratio of each connecting edge meet the corresponding thresholds. It filters out all directional edge groups that meet the conditions, counts the cumulative connection frequency of directional clusters in the graph, and generates an index set according to the frequency to form the directional dominant influence index group.

[0099] Dominant Direction Module: Based on the dominant direction influence index group, extract the flow direction component value and slope component value of each grid node in the control area, filter all node numbers where the flow direction component is greater than the slope component, read the flow direction path in multiple time periods for each numbered node, sort and reorganize the flow direction sequence, summarize the structural change trend of all sequences in the time dimension, and establish a direction evolution path sequence group.

[0100] Path evolution module: Based on the direction evolution path sequence group, extract the inflection point number of the groove in each path, calculate the ratio of the angle difference between each pair of inflection points to the rate of change of continuous angle, determine whether the rate of change ratio exceeds the threshold, mark the first node of the path mutation segment as the scour starting point, read all extreme points to form a path chain, call the greedy algorithm to cluster the extreme nodes of the path, output the cluster center position, and form a scour and siltation distribution node set;

[0101] Structural partitioning module: Based on the distribution node set of scour and sedimentation, it determines whether the direction of the angle change between adjacent nodes in the path segment is consistent, filters out all monotonic angle path segments, reads the angle difference sequence and path length value sequence in each path segment, remaps the path structure order according to the combination value of angle and length, aggregates all path segments to form a path fragment set, and establishes a scour and sedimentation evolution partition map.

[0102] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A method for analyzing the evolution of seabed erosion and deposition, characterized in that, Includes the following steps: S1: Acquire single-phase seabed grid elevation data through sensors, calculate elevation difference in the beach and trough area and extract slope vector, calculate the included angle of adjacent units and select edge pairs that meet the directional conditions, and establish a directional connection weighted network graph structure. S2: Based on the directional connection weighted network graph structure, a graph attention network is used to extract the adjacency relationship of the slope boundary nodes, and to determine whether the slope direction, elevation increase / decrease trend, and flow direction consistency indicators all meet the threshold. By superimposing the influence of directional clusters, a group of directional dominant influence indicators is generated. S3: Based on the aforementioned direction-dominant influence index group, select nodes in the control area grid where the flow direction component is greater than the slope component, extract the evolution path trajectory and rearrange the direction sequence to establish a temporal scour and sedimentation change trend array; S4: Based on the time-series scour and sedimentation trend array, call the continuous angle change value of the inflection point of the channel and calculate the ratio of adjacent turning rate. Use a greedy algorithm to screen the abrupt change segment and identify the scour starting point. Generate the set of nodes of the dominant path of scour and sedimentation by clustering the extreme value sequence of the trajectory chain. S5: Based on the set of nodes of the dominant scour and siltation path, determine the trend of the angle change of the continuous path, extract monotonic segments and calculate the sequence of angle difference and length values, aggregate and reorganize the path structure according to the segmentation, and establish a partition map of the scour and siltation evolution characteristic structure. The specific steps for generating the directed connection weighted network graph structure are as follows: The elevation data of the seabed grid in a single time phase is acquired by the sensor. The elevation values ​​of each grid unit in the beach and trough area are extracted point by point and the index of the four neighboring units is extracted. The slope direction vector is generated by calculating the elevation difference between the current node and the four neighboring nodes and dividing it by the horizontal spacing. A local slope direction vector set is generated. Based on the local slope direction vector set, the cosine angle between the direction vectors of each pair of connected units is calculated and it is determined whether it is less than the limited angle threshold. By filtering out grid unit pairs that meet the standard of similar directions and establishing effective adjacency relationships, a set of grid unit pairs with consistent directions is obtained. Based on the set of grid cells with consistent orientation, a connection edge is constructed for each pair of nodes and a direction attribute is attached. The edge weight is generated by the reciprocal of the cosine angle and a connection matrix is ​​constructed in a unified manner to establish a directional connection weighted network graph structure. The specific steps for generating the aforementioned dominant influence index group are as follows: Based on the directional connection weighted network graph structure, a graph attention network is used to filter the boundary nodes of the bank slope and extract the indexes of all adjacent nodes. The direction change values ​​of adjacent slopes, the elevation increase / decrease difference between nodes and the angle of flow direction are extracted respectively. By constructing a set of three parameters, the set of bank slope node association indicators is obtained. Based on the set of associated indicators for the bank slope nodes, each of the three indicators for each group of nodes is judged item by item, and the node pairs that simultaneously meet all threshold conditions are marked. The node pairs that meet the requirements are numbered and classified in sequence, and the corresponding connection path number set is extracted to obtain the set of nodes with high consistency direction. Based on the highly consistent directional node set, the reciprocal of the directional angle between each pair of nodes, the mean of the flow distance, and the normalized value of the slope difference are taken and superimposed. Continuous path segments are aggregated and extracted by sequential numbering to generate a group of directional dominant influence indicators. The specific steps for generating the time-series siltation and sedimentation trend array are as follows: Based on the dominant influence index group, the magnitude of the flow direction value and the slope direction value of each node is extracted. The two values ​​are compared by the difference operation and the judgment condition that the flow direction component is greater than the slope component is set. Nodes that meet the condition are screened to obtain the dominant flow direction node set. Based on the set of dominant flow direction nodes, the nodes are connected by time sequence numbering, and the spatial distance and directional change angle of adjacent nodes are calculated according to the node index. The node index sequence is generated by rearranging the connection direction order, and a set of scour and siltation trajectory direction sequence is generated. Based on the set of scour and sedimentation trajectory directions, the time difference sequence and direction vector change value of each node are merged, and the change angle sequence of each node in the continuous path is sorted and cumulatively rearranged to establish a time-series scour and sedimentation change trend array. The specific steps for generating the set of nodes representing the dominant scour and sedimentation paths are as follows: Based on the time-series scouring and silting trend array, continuous node direction vector groups are extracted and the angle change values ​​are calculated sequentially. The turning rate sequence is generated by dividing the difference by the time step and the amplitude mutation judgment rule is set to obtain the trajectory direction mutation ratio sequence. Based on the trajectory direction mutation ratio sequence, a greedy algorithm is used to perform mutation segment division, mutation point indexing and location, and extraction of adjacent point sequences. By accumulating and summing continuous angle difference segments, segments with mutation amplitude greater than a set value are selected and their path position indices are recorded to obtain a set of scour mutation path segments. Based on the set of scour mutation path segments, the maximum value of the set of flow direction angles of nodes within the segment is located and it is determined whether there is a reversal point of angle difference on both sides. The nodes where the reversal point is located are spatially reorganized and numbered to generate a set of nodes for the dominant scour and sedimentation paths.

2. The method for analyzing the evolution of seabed erosion and deposition according to claim 1, characterized in that, The greedy algorithm is based on the formula: Wherein: T s ψ represents the overall mutation intensity value of the mutated fragment. i Let |ψ| represent the tangential direction angle of the i-th trajectory point. i+1 -ψ i | represents the difference in direction angle between adjacent trajectory points, reflecting the degree of path turning, e i Let d represent the seabed elevation value of the i-th trajectory point. i,i+1 This represents the horizontal distance between trajectory points i and i+1. κ represents the rate of change of elevation per unit horizontal distance between adjacent trajectory points, reflecting abrupt changes in local slope. i |κ represents the local curvature value at the i-th trajectory point. i+1 -κ i | represents the magnitude of curvature change between adjacent points, s i This represents the unit path velocity at the i-th trajectory point. This represents the average flow velocity across all trajectory points within the current mutation fragment. The value represents the point velocity offset, and μ1, μ2, μ3, and μ4 represent the weighting coefficients of abrupt change in direction, elevation gradient, curvature fluctuation, and velocity offset factor, respectively.

3. The method for analyzing the evolution of seabed erosion and deposition according to claim 1, characterized in that, The specific steps for generating the erosion and sedimentation evolution characteristic structural partition map are as follows: Based on the set of nodes of the dominant scour and sedimentation path, the direction vectors of continuous nodes are extracted and the angle between adjacent nodes is calculated. The direction change pattern is identified by the positive and negative sign sequence of the angle difference, and the start and end positions of the monotonic continuous sequence are marked to generate a set of monotonic direction path segments. Based on the monotonic direction path segment set, the direction angle difference and the distance value between adjacent nodes of each node pair are extracted in the order of the path segments. An angle sequence and a length sequence are constructed respectively. The path angle length parameter sequence set is obtained by grouping by node number and merging with the code. Based on the path angle length parameter sequence set, the angle sequence fluctuation value and length spacing change value are extracted according to the node number within the segment. After classifying the value difference range within continuous segments, the structural segment numbers are rearranged, and paths of the same type are merged and aggregated to establish a structural partition map of scour and sedimentation evolution characteristics.

4. A system for analyzing the evolution of seabed erosion and deposition, characterized in that, The method for analyzing the evolution of seabed erosion and deposition according to any one of claims 1-3, wherein the system comprises: Elevation analysis module: Based on single-phase seabed grid elevation data, extract the elevation values ​​of adjacent grid cells in the beach and channel area, calculate the elevation difference and slope vector direction for each pair, obtain the slope vector angle value and determine whether it falls within the set angle range, filter to form unidirectional edge pairs, and use the elevation difference and slope direction as edge weight values ​​to establish a directional edge weight graph structure. Edge-based graphing module: Based on the aforementioned directional edge weight graph structure, a graph attention network is used to extract directional edge relationships from the set of adjacent nodes of each node. It determines whether the slope direction angle, elevation increase / decrease continuity indicator, and flow direction consistency ratio of each connecting edge meet the corresponding thresholds. All directional edge groups that meet the conditions are filtered out, and the cumulative connection frequency of directional clusters in the graph is counted. An index set is generated according to the frequency to form a directional dominant influence index group. Dominant Direction Module: Based on the dominant influence index group, extract the flow direction component value and slope component value of each grid node in the control area, filter all node numbers where the flow direction component is greater than the slope component, read the flow direction path in multiple time periods for each numbered node, sort and reorganize the flow direction sequence, summarize the structural change trend of all sequences in the time dimension, and establish a direction evolution path sequence group. Path evolution module: Based on the directional evolution path sequence group, extract the inflection point numbers of the grooves in each path, calculate the ratio of the angle difference between each pair of inflection points to the rate of change of continuous turning angles, determine whether the rate of change ratio exceeds the threshold, mark the first node of the path mutation segment as the scour starting point, read all extreme points to form a path chain, call the greedy algorithm to cluster the extreme nodes of the path, output the cluster center position, and form a scour and siltation distribution node set; Structural partitioning module: Based on the scour and sedimentation distribution node set, determine whether the direction of the angle change between adjacent nodes in the path segment is consistent, filter out all monotonic angle path segments, read the angle difference sequence and path length value sequence in each path segment respectively, remap the path structure order according to the combination value of angle and length, aggregate all segment paths to form a path fragment set, and establish a scour and sedimentation evolution partition map.

Images