Beach restoration decision-making method and system based on multi-source data

Abstract

The invention relates to the technical field of coast engineering, and discloses a multi-source data-based beach restoration decision-making method and system, and the method comprises the following steps: obtaining multi-source data of beach monitoring, carrying out the data fusion processing, and generating a feature tensor with a unified temporal-spatial resolution; performing symbol regression analysis based on the feature tensor, and automatically discovering a control equation of beach evolution; and inputting the evolution equation into a decision optimization framework to generate a beach restoration decision scheme. According to the method, through organic combination of multi-source data fusion, symbolic regression analysis and a decision optimization framework, accurate description of a beach evolution process and intelligent generation of a repair decision are realized; according to the scheme, information of various monitoring data can be fully utilized, a control equation of beach evolution and time-varying characteristics of the control equation can be automatically found, and a physical rule is directly converted into a specific engineering decision suggestion, so that prediction precision and decision efficiency of beach repair are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of coastal engineering, more particularly, it relates to a beach restoration decision-making method and system based on multi-source data. BACKGROUND

[0002] In the field of coastal engineering, beach restoration is an important measure to maintain the stability of the coastline and ecological function. Traditional beach restoration decision-making mainly relies on expert experience and numerical simulation models, such as using the STWAVE model for wave field simulation, the GENESIS model for coastline evolution prediction, and the XBeach model for profile response analysis. Although these models have been widely used in engineering practice, they still face the following technical problems: Firstly, beach evolution involves multi-source heterogeneous data such as remote sensing images, acoustic sounding data, GPS coastline measurement, wave buoy monitoring, and tide station records. These data have significant differences in temporal and spatial resolution, data format, and physical meaning. Traditional numerical models cannot directly utilize all data sources, and usually can only selectively use part of the data, resulting in insufficient information utilization.

[0003] Secondly, traditional numerical models such as STWAVE and GENESIS are based on pre-set physical equations. These equations are derived based on idealized physical assumptions, ignoring complex physical processes such as nonlinear coupling of storm surge and wave, and multi-scale effects of sediment transport. For example, in the Shandao Bay beach restoration project, traditional models cannot accurately capture the complex wave shadow zone effect produced by the interaction of southeast waves and breakwaters.

[0004] Thirdly, the control equation parameters and structure of the beach system will exhibit time-varying characteristics due to factors such as seasonal changes and extreme weather events. Traditional static models require manual parameter adjustment to adapt to different environmental conditions, lack self-adaptive ability, and affect the prediction accuracy of long-term restoration effects.

[0005] Finally, even if the wave field distribution and coastline evolution prediction are obtained through traditional models, a large number of scheme comparisons (such as the 14 schemes in the Shandao Bay project) are still needed to determine the optimal restoration measures. There is a lack of systematic conversion methods to directly convert physical evolution rules into decision-making recommendations, and the decision-making process is inefficient and relies on experience. SUMMARY

[0006] The present application provides a beach restoration decision-making method and system based on multi-source data, which solves the limitations of traditional beach restoration decision-making methods in multi-source data fusion, complex physical process simulation, time-varying characteristics adaptation, and decision-making efficiency in related technologies, and cannot meet the engineering practice needs.

[0007] The present application provides a beach restoration decision-making method based on multi-source data, comprising the following steps: S100, acquiring multi-source data of beach monitoring, performing data fusion processing, and generating a feature tensor with unified spatio-temporal resolution; The multi-source data includes remote sensing image data, acoustic sounding data, GPS shoreline measurement data, wave buoy monitoring data, and tide station record data. S200, performing symbolic regression analysis based on the feature tensor to automatically find a control equation of beach evolution; S300, inputting the evolution equation into a decision optimization framework to generate a beach restoration decision scheme; The beach restoration decision scheme includes artificial sand supplement amount, sand blocking dike length, submerged dike elevation, number of groins, and inclination of groins.

[0008] Further, in S100, the following steps are specifically included: S110, multi-source data acquisition and preprocessing, acquiring remote sensing image data including visible light and infrared band images, performing atmospheric correction and geometric correction, acquiring acoustic sounding data, performing noise filtering and depth calibration, acquiring GPS shoreline measurement data, performing coordinate conversion and outlier removal, acquiring wave buoy monitoring data including wave height, period, and direction data quality control, acquiring tide station record data, performing tide level benchmark unification and data filling; S120, spatio-temporal alignment processing, using bilinear interpolation method to resample spatial data to 5-meter grid resolution, using cubic spline interpolation to resample time series to hourly time scale, establishing a unified spatio-temporal reference system to ensure alignment of all data sources in the same coordinate system, performing data synchronization test to ensure accurate correspondence of time stamps of different data sources; S130, physical feature calculation, using finite difference method to calculate spatial gradient of terrain elevation, using central difference format to calculate time derivative of wave height, using finite volume method to calculate divergence of sediment flux, constructing wave-tide coupling term and wave steepness-water depth ratio composite physical quantities; S140, feature tensor generation, designing data structure of feature tensor including spatial dimension, time dimension, and physical quantity dimension, filling aligned original observation data to corresponding positions of the tensor, integrating calculated physical derivative features into the tensor, performing data integrity check and outlier processing.

[0009] Further, in S200, the following steps are specifically included: S210, data preprocessing and initialization, performing standardization processing on the feature tensor to unify dimensions of each physical quantity, setting candidate function set, initializing genetic programming algorithm parameters, and constructing fitness function; S220, genetic programming evolution process, using tree structure coding method to represent candidate equations, and performing crossover operation and mutation operation; S230, sparse regression optimization, converting the candidate equation terms generated by genetic programming into a coefficient matrix, and applying a LASSO algorithm for sparse processing; S240, physical consistency verification, checking the dimensional consistency of the equation, verifying the law of conservation of sediment mass, and verifying the law of conservation of wave energy; S250, time-varying feature analysis, dividing the data sequence with a sliding window, identifying seasonal changes and extreme event periods, performing symbolic regression independently for each time period, generating a segmented evolution equation, and establishing a smooth transition mechanism between time periods; S260, causal structure identification, using the PC algorithm for causal discovery, calculating the structural edit distance of the DAG of adjacent periods, applying the PELT algorithm to detect structural mutation points, constructing variable combination rules based on the DAG, and updating the candidate set of equation terms; S270, coupling effect analysis, calculating the cross-correlation function of variable pairs, generating first-order and second-order coupling terms, and constructing composite physical quantities.

[0010] Further, the candidate function set includes: basic arithmetic operations; trigonometric functions; exponential functions; power functions.

[0011] Further, the genetic programming algorithm parameters include population size, iteration number, crossover probability, and mutation probability.

[0012] Further, in S220, the crossover operation includes randomly selecting two parent trees, randomly selecting a crossover point, and exchanging sub-trees to generate new individuals.

[0013] Further, in S220, the mutation operation is to randomly select a mutation point to replace it with a new operator / variable with a certain probability, and to use tournament selection method for iterative evolution.

[0014] Further, in S300, it specifically includes the following steps: S310, repair engineering parameter setting: determining the control variable set, setting the constraint condition, establishing the parameter association constraint, and generating an initial solution set that meets the constraint; S320, evolution prediction simulation: substituting the repair scheme parameters into the evolution equation; S330, multi-objective optimization function construction: combining the definition of beach stability index, the definition of engineering cost function, and the definition of ecological impact index to construct a comprehensive evaluation function; S340, genetic algorithm optimization solution: using real number coding to represent the repair scheme, setting the algorithm parameters, performing binary crossover and polynomial mutation simulation, and using fast non-dominated sorting; S350, scheme evaluation and screening: analyzing the characteristics of the optimal solution set, evaluating the engineering feasibility, calculating the cost-benefit ratio, and considering the decision preference; S360, uncertainty analysis: using Monte Carlo method, generating random samples, calculating evolution results, statistical confidence interval, calculating scheme robustness index; S370, target point identification: calculating shoreline change rate, identifying erosion and deposition area, evaluating terrain stability, determining control section; S380, intervention timing optimization: analyzing seasonal variation, identifying state transition point, designing implementation scheme based on state prediction model, distributing engineering quantity by stages, arranging construction time sequence, optimizing engineering progress; S390, decision suggestion generation; determine engineering combination, develop layout scheme, plan implementation time sequence, compile construction guide, output decision report.

[0015] Further, in S350, analyzing the characteristics of the optimal solution set includes calculating the uniformity of the solution distribution, evaluating the diversity index of the solution, and identifying the key parameter combination mode; evaluating the engineering feasibility includes construction difficulty index, material availability and environmental adaptability; calculating the cost benefit ratio includes stability improvement amount, engineering cost and maximum allowable cost, and selecting the optimal scheme.

[0016] The application also proposes a beach restoration decision system based on multi-source data, comprising: Data acquisition and preprocessing module: responsible for acquiring and preprocessing multi-source monitoring data, performing spatio-temporal alignment processing on the data, unifying to 5-meter grid resolution and hourly time scale, calculating physical characteristics and generating feature tensor containing multi-scale physical process information; Evolution equation discovery module: based on feature tensor, performing symbolic regression analysis, automatically discovering the control equation of beach evolution through genetic programming algorithm, performing sparse regression optimization and physical consistency verification, analyzing time-varying characteristics and causal structure, identifying key coupling effects, and finally generating evolution equation capable of describing complex physical processes; Decision optimization module: input evolution equation into decision optimization framework, set restoration engineering parameters and perform evolution prediction simulation, construct multi-objective optimization function, solve optimal scheme through genetic algorithm, perform scheme evaluation, uncertainty analysis and target point identification, optimize intervention timing, and finally generate complete beach restoration decision scheme.

[0017] The beneficial effects of the application are: The present application realizes accurate description of the evolution process of the coast and intelligent generation of the repair decision by the organic combination of multi-source data fusion, symbolic regression analysis and decision optimization framework; the scheme can make full use of the information of various monitoring data, automatically find the control equation of the evolution of the coast and its time-varying characteristics, and directly convert the physical law into specific engineering decision suggestions, thereby improving the prediction accuracy and decision efficiency of the coast repair and reducing the dependence on expert experience. Meanwhile, the scheme has strong adaptability and robustness, can cope with complex hydrodynamic conditions and extreme weather events, and provides strong support for scientific decision of the coast repair project. BRIEF DESCRIPTION OF DRAWINGS

[0018] Figure 1 is a flow chart of a coast repair decision method based on multi-source data proposed by the present application; Figure 2 is a structural block diagram of a coast repair decision system based on multi-source data proposed by the present application.

[0019] In the figure: 101, data acquisition and preprocessing module; 102, evolution equation discovery module; 103, decision optimization module. DETAILED DESCRIPTION

[0020] The subject matter described herein will now be discussed with reference to example implementations. It should be understood that the discussion of these implementations is merely meant to provide a better understanding of the subject matter described herein and can be changed in function and arrangement without departing from the scope of the present specification. Various processes or components can be omitted, substituted, or added according to different examples. In addition, features described with respect to some examples can also be combined in other examples.

[0021] As shown in Figure 1 A coast repair decision method based on multi-source data, comprising the following steps: S100, acquiring multi-source data of coast monitoring, performing data fusion processing, and generating a feature tensor with unified space-time resolution; In an embodiment of the present application, the following steps are specifically included: S110, multi-source data acquisition and preprocessing, acquiring remote sensing image data including visible light and infrared band images, performing atmospheric correction (using FLAASH model) and geometric correction (RMS error <0.5 pixels), acquiring acoustic sounding data, performing noise filtering (using median filtering, window size 3x3) and depth calibration (based on known control points, accuracy ±0.1 m), acquiring GPS shoreline measurement data, performing coordinate conversion (using 7-parameter conversion model) and outlier removal (3σ criterion), acquiring wave buoy monitoring data including wave height (Hs), period (T) and direction (θ) data quality control (based on physical constraints), acquiring tide station record data, performing tide level benchmark unification (using 85 elevation benchmark) and data filling (based on harmonic analysis method); S120, spatio-temporal alignment processing, using bilinear interpolation method to resample spatial data to 5-meter grid resolution: ; ; ; wherein, is the interpolation result at the target point (x, y), 、 、 、 are the coordinates of the surrounding four known points, is the interpolation coefficient in the x direction, , is the interpolation coefficient in the y direction, ; using cubic spline interpolation to resample the time series to hourly time scale: ; wherein, is the interpolation result at time t, 、 、 、 is the interpolation coefficient of the ith segment determined by boundary conditions, is the ith known time point, , is the time point to be interpolated, ; establishing a unified spatio-temporal reference system to ensure that all data sources are aligned in the same coordinate system (using CGCS2000 coordinate system), performing data synchronization test to ensure that the timestamps of different data sources correspond accurately (time error <1 min); S130, physical feature calculation, using finite difference method to calculate the spatial gradient of terrain elevation: ; ; in, Let (i, j) be the terrain elevation at grid point (i, j). , Let x and y be the spatial step sizes. For grid coordinates, ; The time derivative of the wave height was calculated using the central difference scheme: ; in, Let be the effective wave height at time t. For time step, For a point in time; Calculation of divergence of sediment flux based on the finite volume method: ; ; in, For sediment flux, , Let x and y be the components of the sediment flux. This represents the sediment transport coefficient. For the effective wave height, For the wave incident angle, ; Constructing composite physical quantities such as wave-tide coupling terms and wave steepness-depth ratio: Coupling terms: ; in, The coupling coefficient is... For the effective wave height, The tide level; Wave steepness ratio: ; in, For wave steepness ratio, For the effective wave height, For wavelength, For water depth; S140, Feature Tensor Generation: Designing the Data Structure for Feature Tensors ,Include: Spatial dimension ( ): 5m×5m grid, time dimension : 1-hour interval, physical quantity dimension 12 variables; Fill the corresponding positions in the tensor with the aligned original observation data: ; wherein, is a feature tensor with dimension [1000, 1000, 8760, 12], is a spatial grid index, , is a time index, , is a physical quantity index, , is an observation data mapping function, is a spatial coordinate, is a time point, is the lth physical quantity; the calculated physical derivative features are integrated into a tensor: ; wherein is a feature tensor with dimension [1000, 1000, 8760, 12], is an index of derivative features, m [1, 12] ， is an index of original features, [1, 12], is a spatial coordinate; data integrity check (missing rate <5%) and outlier processing (based on the MAD method) are performed; ; wherein, is a data point, the unit is consistent with the original variable, is a data set, the unit is consistent with the original variable, is a median operation, generating a final feature tensor (dimension: 1000x1000x8760x12) containing multi-scale physical process information; S200, based on the feature tensor, a symbolic regression analysis is performed to automatically find the control equation of the beach evolution; In an embodiment of the present application, the following steps are specifically included: S210, data preprocessing and initialization, standardizing the feature tensor to make the dimensions of each physical quantity uniform: ; wherein, is the standardized data, is the original data, is the mean, is the standard deviation; a candidate function set is set , including: basic arithmetic operations: ; Trigonometric functions: ; Exponential functions: ; Power functions: , ; Initialize genetic programming algorithm parameters: population size = 500, number of iterations = 1000, crossover probability = 0.8, mutation probability = 0.1; Construct fitness function: ; where = 0.7, = 0.2, = 0.1 are weight coefficients, is the mean square error, is the equation complexity, is the degree of violation of physical constraints.

[0022] S220, genetic programming evolution process, using tree structure coding method to represent candidate equations: Internal nodes ∈ (operator set) Leaf nodes ∈ {variable set ∪ constant set} Perform crossover operation: Randomly select two parent trees , ; Randomly select a crossover point , ; Exchange sub-trees to generate new individuals , ; Perform mutation operation: randomly select mutation point Replace with new operator / variable with probability ; Use tournament selection method: randomly select k individuals (k = 3) to select the individual with the highest fitness into the next generation, and iterate evolution until the termination condition is reached; Termination condition: the number of iterations reaches T or the optimal fitness does not improve for 50 consecutive generations; S230, sparse regression optimization, convert the candidate equation generated by genetic programming into a coefficient matrix , apply LASSO algorithm for sparse processing: ; where, for target variable, for coefficient vector, for regularization parameter ( =0.01), control equation complexity by L1 regularization term: ; where is the i-th component of coefficient vector, remove terms with small contribution: <0.001, get simplified equation expression; S240, physical consistency verification, check dimensional consistency of equation: [left side]=[right side], verify sediment mass conservation law: ; where is water depth, is sediment flux; verify wave energy conservation law: ; where, is wave energy density, is group velocity, is dissipation term; S250, time-varying feature analysis, divide data sequence with sliding window: window width w=90 days, sliding step s=30 days, identify seasonal changes and extreme event periods: analyze periodic characteristics by wavelet transform, identify extreme events by POT method, perform signed regression independently for each time period: apply S220-240 method to generate segmented evolution equation: ; ;

[0023] where, is shoreline position, is time, is evolution function / s for each period, is time interval; establish smooth transition mechanism between periods: use weight function w(t) to achieve smooth transition; S260, causal structure identification, use PC algorithm for causal discovery: significance level α=0.05 maximum conditional set size k=3 calculate structural edit distance of adjacent periods DAG: ; where, , is edge set, This represents the set difference operation; Applying the PELT algorithm to detect structural abrupt changes: penalty parameter β=0.1, minimum segment length l=30. Based on DAG, construct variable combination rules: direct causal relationship → first-order term; indirect causal relationship → higher-order term; update the candidate set of equation terms. S270, Coupling effect analysis, calculating the cross-correlation function of variable pairs: ; in, , It is a time series. , The mean, , Standard deviation, For time delay, For the expected operation, The cross-correlation coefficient; Generate first-order and second-order coupling terms: First-order terms: ; Second-order term: ; Constructing composite physical quantities: Wave height-tide level coupling term: ; Wave steepness-depth ratio: ; Assess the importance of each term: Through partial correlation analysis and variance contribution rate, generate the final composite equation, which is a complete evolution equation containing significant coupling terms; S300 inputs the evolution equation into the decision optimization framework to generate beach restoration decision schemes; In one embodiment of the present invention, the following steps are specifically included: S310, Repair engineering parameter settings, determine the set of control variables; ; in, This refers to the amount of sand added artificially. ∈[1×10 4 5×10 5 ], The length of the sand-retaining dike. ∈[100, 500], The elevation of the submerged dike. ∈[1, 5], For the number of groynes. ∈[1, 5], The inclination angle of the groyne. ∈[30, 90]; Set constraints: Total project cost ; Ecological impact index ; Structure spacing ; Wherein is the maximum allowable project cost, ecological impact index, wherein is the maximum allowable ecological impact, structure spacing, wherein is the spacing between adjacent structures; Establish parameter correlation constraints: Length of sand trap and sand replenishment ratio ; Submerged dike elevation and water depth ratio ; Generate an initial solution set that satisfies the constraints , wherein is the number of initial solutions; S320, evolution prediction simulation, substitute the repair scheme parameters into the evolution equation: ; Wherein, is the shoreline position, is the water depth, is the effective wave height, is the wave direction angle, is the tide level, is the time; Set the simulation time step =1h, the total duration =10 years, use the fourth-order Runge-Kutta method to solve the evolution equation: ; ; ; ; ; Wherein, record the shoreline position , , …, at key moments ; Calculate the evolution characteristic index: Rate of shoreline change ; Stability index ; Wave attenuation coefficient ; Wherein, is the shoreline change amount, is the maximum allowable variation, the wave attenuation coefficient, wherein , are the incident and transmitted wave heights, respectively; S330, multi-objective optimization function construction, define the beach stability index: ; wherein, is the shoreline variation, is the maximum allowable variation; Define the engineering cost function: ; wherein, is the unit sand replenishment cost, is the sand retaining dike unit price, is the sand retaining dike length, is the submerged dike unit price, is the submerged dike length; ; Define the ecological impact index: ; wherein, is the weight coefficient, is the sub-ecological impact factor; Set the objective function weight coefficient: = 0.4 (stability), = 0.3 (cost), = 0.3 (ecology); Construct the comprehensive evaluation function: ; S340, genetic algorithm optimization solution: adopt real number coding to represent the repair scheme: ; wherein, is the repair scheme, is the artificial sand replenishment amount, is the sand retaining dike length, is the submerged dike elevation, is the number of jetties, is the inclination angle of the jetties; Set the algorithm parameters: population size = 200, iteration number = 500, crossover probability = 0.8, mutation probability = 0.1; Perform simulated binary crossover (SBX): ; ; where β = 0.5, , ∈ [0, 1], = 20, distribution index, , is the parent individual, , is the child individual; Perform polynomial mutation: ; where, , ∈ [0, 1], = 20, distribution index, is the individual to be mutated, , are the upper and lower bounds; Use fast non-dominated sorting: calculate the number of domination and the dominated solution set according to hierarchical, update the crowded distance, update the Pareto optimal solution set ; S350, scheme evaluation and screening: Analyze the characteristics of the optimal solution set: calculate the distribution uniformity of the solution , evaluate the diversity index of the solution , identify the key parameter combination mode; Evaluate the engineering feasibility: construction difficulty index CI, material availability MI, environmental adaptability EI; Calculate the cost-benefit ratio: ; where, is the stability improvement amount, is the engineering cost, is the maximum allowable cost; Consider the decision preference: Utility function: ; where, is the weight coefficient, , is the utility function, is the decision variable, the unit is consistent with the original variable, and the optimal scheme is selected ; S360, uncertainty analysis, using Monte Carlo method: generate = 1000 sets of parameter samples parameter disturbance range ± 20%; Generate random samples: ; where, For the optimal solution, For random disturbance, The square of the variance; Calculate the evolution result: , statistical confidence interval: (t), (t)], calculate the robustness index of the scheme: ; S370, repair target identification, calculate the shoreline change rate: , identify the erosion and deposition area, where the strong erosion: <-2m / year; strong deposition: >1m / year.

[0024] Evaluate the terrain stability: ; Where, is the terrain elevation, is the Laplacian of the elevation, is the terrain stability index; Determine the control section: the location of the maximum erosion rate, the terrain mutation point, and the hydrodynamic concentration area.

[0025] S380, intervention timing optimization, analyze seasonal changes: wave height cycle Tide cycle Sediment transport cycle , identify state transition points: eigenvalue zero point, phase space trajectory inflection point; Predict the critical time: ; Where, is the initial time, is the prediction time interval, is the critical time; Based on the state prediction model, design the implementation scheme, allocate the engineering quantity by stages, arrange the construction sequence, and optimize the engineering progress: minimize the construction period , maximize the work efficiency η; S390, decision-making suggestion generation, determine the project combination: main project type, auxiliary measure configuration, monitor system layout, develop layout scheme: plan layout, cross-section design, construction zoning map, plan implementation time sequence: total construction period Sub-item construction period Construction sequence, prepare construction guide: technical standards, quality requirements, safety measures, output decision report: project implementation plan, expected effect, risk assessment.

[0026] As shown in Figure 2 , a shore beach repair decision system based on multi-source data includes the following modules: Data acquisition and preprocessing module 101: responsible for acquiring and preprocessing multi-source monitoring data, including remote sensing images, acoustic sounding, GPS shoreline measurement, wave buoy and tide station data. The data is processed for spatio-temporal alignment, unified to 5-meter grid resolution and hourly time scale. Physical characteristics are calculated and feature tensors containing multi-scale physical process information are generated.

[0027] Evolution equation discovery module 102: based on the feature tensor, symbolic regression analysis is performed to automatically discover the control equation of beach evolution through genetic programming algorithm. Sparse regression optimization and physical consistency verification are performed to analyze time-varying features and causal structure, identify key coupling effects, and finally generate evolution equations that can describe complex physical processes.

[0028] Decision optimization module 103: input the evolution equation into the decision optimization framework, set the repair engineering parameters and perform evolution prediction simulation. A multi-objective optimization function is constructed, and the optimal solution is obtained by genetic algorithm. Scheme evaluation, uncertainty analysis and repair target identification are performed to optimize the intervention time, and finally a complete beach repair decision scheme is generated.

[0029] Examples of S100-S300 of the beach repair decision method based on multi-source data, in an embodiment of the present application, through the steps of the above beach repair decision method based on multi-source data, and the modules in the system, the following examples are given: S100, multi-source data acquisition and preprocessing: in the Shandao Bay coastal zone protection and repair project, the multi-source data acquisition and preprocessing steps are implemented as follows: Ocean dynamic environment data acquisition: collect ocean dynamic environment observation data such as tide, wave, current, etc. in Shandao Bay area, use automatic observation station and field investigation combined mode to obtain continuous ocean dynamic environment parameters, establish tide model in Shandao Bay area, and obtain tide variation law; Topography data acquisition: obtain underwater topography data in Shandao Bay area through multi-beam sounding system, measure beach topography using RTK-GPS technology, obtain high-precision shoreline position and beach elevation data, collect historical topography data including historical charts, measurement data, etc.; Remote sensing image data acquisition: obtain multi-temporal high-resolution satellite images (such as GF-1, GF-2, etc.) in Shandao Bay area, collect unmanned aerial vehicle aerial image, obtain beach detail features, use historical remote sensing image data to establish shoreline change sequence; Sediment transport data acquisition: set up sediment sampling points at key sections, obtain sediment particle size, content, etc. parameters, measure suspended sediment concentration and transport flux, analyze sediment source and transport path; Socio-economic data collection: Collect socio-economic data such as population, industrial structure, land use, etc. in the surrounding area of Shandao Bay, investigate the current situation and planning of coastal zone development, evaluate the value of coastal zone resources and development intensity; The collected data is shown in Table 1; Table 1 Statistics of multi-source data collection of Shandao Bay coastal zone

[0030] Data preprocessing: Quality control of the collected raw data, eliminate abnormal values and error data, unify the time and space reference system of the data, establish a unified coordinate system, geometric correction, radiation correction and atmospheric correction of remote sensing images, interpolation processing of terrain data, generation of continuous digital elevation model, time series analysis of marine dynamic environment data, extraction of characteristic parameters; Through the above multi-source data collection and preprocessing, a complete basic database is established for the Shandao Bay coastal zone protection and restoration project, providing data support for subsequent beach evolution analysis and restoration decision-making.

[0031] S200, beach evolution mechanism analysis and model construction: In the Shandao Bay coastal zone protection and restoration project, the beach evolution mechanism analysis and model construction steps are implemented as follows: Beach evolution history analysis: Based on multi-temporal remote sensing images, extract the position change of Shandao Bay coastline from 1990 to 2020, calculate the coastline change rate, identify the erosion and deposition sections, analyze the spatio-temporal distribution characteristics of coastline change, and determine the key change areas; Influence factor identification and quantification: Analyze the influence of wave, tide, current and other natural dynamic factors on beach evolution, evaluate the impact of extreme events such as typhoon and storm surge on beach morphology, quantify the interference of human activities (such as reclamation, port construction, protection engineering, etc.) on beach evolution, and establish the correlation model between influence factors and beach evolution; Sediment transport process analysis: Based on field observation data, analyze the sediment transport characteristics of Shandao Bay, identify the main sediment sources and sinks, determine the sediment transport path, calculate the sediment budget balance, evaluate the regional sediment supply and demand situation, and analyze the interference mechanism of human activities on sediment transport process; Hydrodynamic numerical model construction: Establish a two-dimensional tidal current numerical model of Shandao Bay, construct a wave propagation model, simulate the wave field distribution, simulate the storm surge process under extreme weather conditions, and calibrate and verify the model through measured data; Beach evolution numerical model construction: Based on the hydrodynamic model, establish a sediment transport numerical model, construct a beach topography evolution model, simulate the beach surface change process, develop a coastline change prediction model, predict the future coastline position, and verify and optimize the model through historical data; Multi-scenario simulation and analysis: Design different sea level rise scenarios to simulate the impact of climate change on coastal evolution, simulate the impact of different intensity typhoons and storm surges on the coast, evaluate the impact of different human activity scenarios on the stability of the coast, analyze the response characteristics and evolution trend of the coast under various scenarios; Through the above analysis of the mechanism of coastal evolution and model construction, the Shandao Bay Coastal Zone Protection and Restoration Project reveals the internal mechanism of coastal evolution in the region and establishes a reliable prediction model, providing a theoretical basis and technical support for making scientific restoration decisions.

[0032] S300, Design and evaluation of coastal restoration schemes: In the Shandao Bay Coastal Zone Protection and Restoration Project, the design and evaluation of coastal restoration schemes are implemented as follows: Determination of restoration targets: Based on the results of coastal evolution analysis, determine the key restoration areas in Shandao Bay, set multi-dimensional restoration targets such as coastal stability, ecological function, disaster prevention and mitigation, and develop short-term (5 years), medium-term (10 years) and long-term (20 years) phased restoration targets, and clarify the expectations and requirements of all stakeholders for restoration results; Selection of restoration technologies: Collect domestic and foreign cases and best practices of coastal restoration technologies, evaluate the applicability of various restoration technologies in Shandao Bay, and select a combination of restoration technologies suitable for the characteristics of Shandao Bay, including beach replenishment and maintenance technology, ecological revetment technology, artificial reef construction technology, vegetation restoration technology, and sediment interception and regulation technology; Design of restoration schemes: Design Scheme One: Mainly engineering measures, including construction of offshore dikes, groin systems and beach replenishment; Design Scheme Two: Mainly ecological restoration, including artificial reefs, ecological revetments and vegetation restoration; Design Scheme Three: Comprehensive scheme combining engineering measures and ecological restoration; Refine the engineering layout, scale, materials and construction technology of each scheme.

[0033] Simulation and prediction of scheme effects: Use the numerical model constructed in S200 to simulate the response of the coast after the implementation of each restoration scheme, predict the trend of shoreline change and the evolution process of the beach under different schemes, evaluate the protection effect of each scheme under extreme weather conditions, and analyze the impact of each scheme on the regional sediment transport pattern; The results of simulation and prediction of scheme effects are shown in Table 2: Table 2: Degree of achievement of phased targets for Shandao Bay coastal restoration

[0034] Multi-criteria comprehensive evaluation: An evaluation index system including engineering effect, ecological impact, economic cost, social benefit and other factors is established, the weight of each evaluation index is determined by using analytic hierarchy process (AHP), multi-criteria comprehensive score is given to each repair scheme, the advantages and disadvantages of each scheme are analyzed, and the key restricting factors are identified; The evaluation results of the repair schemes are shown in Table 3: Table 3 Comparison and evaluation table of Shandao Bay beach repair schemes

[0035] Optimal scheme determination and optimization: Based on the comprehensive evaluation results, the optimal scheme of Shandao Bay beach repair is determined, the optimal scheme is locally optimized and adjusted to improve the implementation effect, the phased implementation plan and investment budget are formulated, and the repair effect monitoring and evaluation scheme is designed.

[0036] Through the above beach repair scheme design and evaluation, the Shandao Bay coastal zone protection and repair project finally determines the comprehensive repair scheme combining engineering measures and ecological restoration, which ensures the stability of the beach while focusing on the restoration of ecological system function, and has good economic feasibility and social acceptance, providing scientific technical support for the sustainable development of Shandao Bay coastal zone.

[0037] The embodiments of the present application are described above, but the present application is not limited to the specific embodiments described above, and the specific embodiments described above are only illustrative and not limiting, and those skilled in the art can make many forms under the inspiration of the present application, which all belong to the protection of the present application.

Claims

1. A method for shoreline restoration decision making based on multi-source data, characterized in that, The method comprises the following steps: S100, acquiring multi-source data of beach monitoring, performing data fusion processing, and generating a feature tensor with unified space-time resolution; The multi-source data comprises remote sensing image data, acoustic sounding data, GPS shoreline measurement data, wave buoy monitoring data, and tide station record data; S200, performing symbolic regression analysis based on the feature tensor to automatically find a control equation of beach evolution; S300, inputting the evolution equation into a decision optimization framework to generate a beach restoration decision scheme; The beach restoration decision scheme comprises artificial sand supplement amount, sand blocking dike length, submerged dike elevation, number of groins, and inclination angle of the groins.

2. The method of claim 1, wherein, In S100, the following steps are specifically included: S110, multi-source data acquisition and preprocessing, acquiring remote sensing image data including visible light and infrared band images, performing atmospheric correction and geometric correction, acquiring acoustic sounding data, performing noise filtering and depth calibration, acquiring GPS shoreline measurement data, performing coordinate conversion and outlier removal, acquiring wave buoy monitoring data including wave height, period, and direction data quality control, acquiring tide station record data, performing tide level benchmark unification and data filling; S120, time-space alignment processing, using a bilinear interpolation method to resample spatial data to a unified 5-meter grid resolution, using a cubic spline interpolation to resample time series to a unified hour time scale, establishing a unified time-space reference system to ensure that all data sources are aligned in the same coordinate system, performing data synchronization inspection to ensure that the timestamps of different data sources are accurately corresponded; S130, physical feature calculation, using a finite difference method to calculate the spatial gradient of terrain elevation, using a central difference format to calculate the time derivative of wave height, using a finite volume method to calculate the divergence of sediment flux, constructing a wave-tide coupling term and a composite physical quantity of wave steepness-water depth ratio; S140, feature tensor generation, designing a data structure of the feature tensor including spatial dimension, time dimension, and physical quantity dimension, filling the aligned original observation data to the corresponding positions of the tensor, integrating the calculated physical derivative features into the tensor, and performing data integrity inspection and outlier processing.

3. The method of claim 1, wherein, In S200, the following steps are specifically included: S210, data preprocessing and initialization, performing standardization processing on the feature tensor to unify the dimensions of each physical quantity, setting a candidate function set, initializing genetic programming algorithm parameters, and constructing a fitness function; S220, genetic programming evolution process, using a tree structure coding method to represent the candidate equation, and performing cross operation and mutation operation; S230, sparse regression optimization, converting the candidate equation items generated by genetic programming into a coefficient matrix, and applying a LASSO algorithm for sparse processing; S240, physical consistency verification, verifying the dimensional consistency of the equation, verifying the law of conservation of mass of sediment, and verifying the law of conservation of wave energy; S250, time-varying feature analysis, dividing the data sequence using a sliding window, identifying seasonal changes and extreme event periods, independently performing symbolic regression on each time period, generating a segmented evolution equation, and establishing a smooth transition mechanism between time periods; S260, Causal structure identification, PC algorithm is used for causal discovery, the structural edit distance of adjacent period DAG is calculated, PELT algorithm is applied to detect structural mutation points, variable combination rules are constructed based on DAG, and equation item candidate set is updated; S270, Coupling effect analysis, the cross-correlation function of variable pairs is calculated, first-order and second-order coupling terms are generated, and composite physical quantities are constructed.

4. The method of claim 3, wherein, The candidate function set includes: basic arithmetic operations; trigonometric functions; exponential functions; power functions.

5. The method of claim 4, wherein, The genetic programming algorithm parameters include population size, iteration number, crossover probability and mutation probability.

6. The method of claim 5, wherein, In S220, the crossover operation includes randomly selecting two parent trees, randomly selecting a crossover point, and exchanging sub-trees to generate new individuals.

7. The method of claim 6, wherein, In S220, the mutation operation is to replace the new operator / variable at the randomly selected mutation point with a probability, and the tournament selection method is used for iterative evolution.

8. The method of claim 1, wherein, In S300, it specifically includes the following steps: S310, Repair engineering parameter setting: determine the control variable set, set the constraint condition, establish the parameter association constraint, and generate the initial solution set satisfying the constraint; S320, Evolution prediction simulation: substitute the repair scheme parameters into the evolution equation; S330, Multi-objective optimization function construction: combine the definition of beach stability index, define the engineering cost function, and define the ecological impact index to construct the comprehensive evaluation function; S340, Genetic algorithm optimization solution: adopt real number coding to represent the repair scheme, set the algorithm parameters, perform simulated binary crossover and polynomial mutation, and adopt fast non-dominated sorting; S350, Scheme evaluation and screening: analyze the characteristics of the optimal solution set, evaluate the engineering feasibility, calculate the cost-benefit ratio, and consider the decision preference; S360, Uncertainty analysis: use the Monte Carlo method to generate random samples, calculate the evolution results, and calculate the confidence interval and the robustness index of the scheme; S370, Repair target point identification: calculate the shoreline change rate, identify the erosion and deposition area, evaluate the topographic stability, and determine the control section; S380, Intervention timing optimization: analyze seasonal changes, identify state transition points, design implementation schemes based on state prediction models, allocate engineering quantities by stages, arrange construction timing, and optimize engineering progress; S390, Decision suggestion generation; determine the engineering combination, develop the layout scheme, plan the implementation timing, prepare the construction guide, and output the decision report.

9. The method of claim 8, wherein, In S350, analyzing the characteristics of the optimal solution set includes calculating the uniformity of the solution distribution, evaluating the diversity index of the solution, and identifying the key parameter combination mode; evaluating the engineering feasibility includes construction difficulty index, material availability and environmental adaptability; calculating the cost-benefit ratio includes stability improvement amount, engineering cost and maximum allowed cost, and selecting the optimal scheme.

10. A multi-source data based beach restoration decision system, characterized in that, Performing a step in a beach restoration decision-making method based on multi-source data as claimed in any one of claims 1-9, comprising: A data acquisition and preprocessing module is responsible for acquiring and preprocessing multi-source monitoring data, performing spatio-temporal alignment processing on the data, unifying to a 5-meter grid resolution and an hourly time scale, calculating physical characteristics, and generating a feature tensor containing multi-scale physical process information; Evolution equation discovery module: based on the eigen tensor, the sign regression analysis is carried out, the control equation of the evolution of the beach is automatically discovered through the genetic programming algorithm, the sparse regression optimization and the physical consistency verification are executed, the time-varying characteristics and the causal structure are analyzed, the key coupling effect is identified, and finally the evolution equation capable of describing the complex physical process is generated; Decision optimization module: the evolution equation is input into the decision optimization framework, the repair engineering parameters are set and the evolution prediction simulation is carried out, the multi-objective optimization function is constructed, the optimal scheme is solved through the genetic algorithm, the scheme evaluation, the uncertainty analysis and the repair target identification are executed, the intervention time is optimized, and finally the complete beach repair decision scheme is generated.

Images