A design and optimization method for underwater submerged dikes based on weighted K-means method

Abstract

The present invention discloses a method for designing and optimizing underwater submerged dikes based on a weighted K-means method, comprising the following steps: conducting tests according to a test combination derived from a D-optimal design method and collecting test data; preprocessing the collected test data, calculating the transmission coefficient, and determining factors affecting the wave-breaking performance of the submerged dike; clustering the transmission coefficient using a weighted K-means clustering method, analyzing the clustering results obtained, and determining a parameter selection range that can maximize the wave-breaking performance of the submerged dike. The present invention adopts the above-mentioned method for designing and optimizing underwater submerged dikes based on a weighted K-means method, which can more accurately analyze marine environmental data, improve the accuracy of submerged dike design, effectively reduce engineering costs, and improve the stability and durability of submerged dikes. It has strong versatility and is applicable to different types of underwater submerged dike design and optimization, greatly reducing the amount of calculation and complexity, and can draw conclusions in a short time.

Description

Technical Field

[0001] The present invention relates to the technical fields of marine engineering, coastal protection and ecological restoration, and in particular to an underwater submerged dike design and optimization method based on a weighted K-means method. Background Art

[0002] Oyster reefs, part of so-called ecosystem engineering, are widespread in temperate and subtropical estuaries and shallow marine areas. These reefs contribute significantly to water purification, fish stock enhancement, shoreline erosion mitigation, carbon sequestration, biodiversity enhancement, and overall ecosystem stability. They play a crucial role in stabilizing intertidal sediments and influencing hydrodynamic patterns in coastal environments.

[0003] While traditional rigid protective structures (such as concrete embankments) can effectively resist wave erosion, they also have negative impacts on the ecological environment. In recent years, natural ecological protective structures (such as oyster reefs and coral reefs) have gradually gained attention because they not only reduce wave energy but also promote biodiversity, improve water quality, and restore ecosystems.

[0004] In practical applications, natural underwater dikes should meet the following conditions: (1) have good wave-absorbing performance; (2) not cause too much negative impact on the local ecological environment; (3) the cost and construction difficulty should be controlled within a reasonable range; (4) the structure can be guaranteed to be stable and not damaged in the long term. In order to meet the above conditions, the design and optimization of underwater dikes are very important. When sufficient data is collected, how to quickly distinguish which factors will have a greater impact on the wave-absorbing performance of the dike, which factors have a smaller impact, and the interaction between them is very critical.

[0005] Traditional methods rely on empirical formulas and numerical simulations. Empirical formulas are typically based on experimental data or historical engineering cases under specific conditions, and their applicability is limited. When marine environmental conditions change, the predictive accuracy of empirical formulas decreases significantly. While numerical simulation techniques can simulate complex marine environments, their accuracy is highly dependent on the selection of model parameters and the setting of boundary conditions. Furthermore, numerical simulations are computationally intensive, resulting in low computational efficiency, especially when processing large-scale, multi-scale marine environmental data.

[0006] In the existing technology, cluster analysis is an unsupervised learning method that can automatically discover potential patterns and structures from large amounts of data. In marine engineering, cluster analysis can be used to identify key features of the marine environment, thereby providing a scientific basis for submerged dike design. Although the traditional K-means clustering method is simple and effective, it has certain limitations when processing complex data, such as sensitivity to noise and difficulty in handling non-uniform data distribution. The weighted K-means method can better reflect the importance or influence of the data by assigning weights to data points, thereby improving the accuracy of clustering. In submerged dike design, the weighted K-means method can assign weights according to the degree of influence of different environmental parameters on the design, thereby more accurately identifying key design parameters and optimizing design solutions.

[0007] Therefore, based on the physical model test, the present invention adopts the weighted K-means clustering method and proposes an underwater submerged dike design and optimization method based on the weighted K-means method to realize the design and optimization of ecological submerged dike. Summary of the Invention

[0008] The purpose of the present invention is to provide an underwater submerged dike design and optimization method based on the weighted K-means method. By introducing the weighted K-means clustering method, it can more accurately analyze marine environmental data and improve the accuracy of submerged dike design; by optimizing design parameters, it can effectively reduce engineering costs and improve the stability and durability of submerged dikes; it can greatly reduce the amount of calculation and complexity, and can draw conclusions in a short time. It can be applied to the design and optimization of different types of underwater submerged dikes.

[0009] To achieve the above object, the present invention provides an underwater submerged dike design and optimization method based on the weighted K-means method, comprising the following steps:

[0010] Step S1: Conducting tests according to the test combination obtained by the D-optimal design method and collecting test data;

[0011] Step S2: pre-processing the collected test data, calculating the transmission coefficient under each working condition, and determining the main factors affecting the wave-breaking performance of the submerged dike;

[0012] Step S3: cluster the preprocessed data according to the transmission coefficient using the weighted K-means clustering method, analyze the obtained clustering results, and determine the parameter selection range that can maximize the wave absorption performance of the submerged dike.

[0013] Preferably, in step S1, a test combination is given according to the D-optimal design method, and the wave generator parameters are adjusted to simulate different ocean environments for testing, while the wave height meter data is collected and recorded for subsequent analysis;

[0014] The collected test data specifically include: transmission wave height H t Relative water depth kh, wave steepness kH / 2, relative immersion depth of oyster reef d / h, relative length of oyster reef kL / (2π), relative distance between oyster reefs kL s / (2π), riverbed slope s and the number of oyster reefs N.

[0015] Preferably, in step S2, the collected test data is preprocessed to calculate the transmission coefficient K under each working condition. t , determine the main factors affecting the wave-absorbing performance of the submerged dike. The specific process is as follows:

[0016] Step S21, pre-processing the collected test data;

[0017] The collected data contains incident wave signals and reflected wave signals, so the data is separated and processed as shown below:

[0018]

[0019] η1=A1 cosωt+B1 sinωt;

[0020] η2=A2 cosωt+B2 sinωt;

[0021] Among them, a i is the incident wave amplitude, a r is the reflected wave amplitude, k is the wave number, Δl is the distance between wave height meters 1 and 2; η1 and η2 are the measured data of wave height meters 1 and 2, respectively, from which the values ​​of parameters A1, B1 and A2, B2 are obtained;

[0022] Step S22: Calculate the transmission coefficient K under each working condition t ;

[0023] After separating the incident wave and the reflected wave, the transmission wave height is obtained and the transmission coefficient K is obtained. t , as shown below:

[0024]

[0025] Among them, K t is the transmission coefficient; H t is the transmission wave height measured in the experiment; H i is the incident wave height; a t is the transmitted amplitude;

[0026] Step S23: determining the main factors affecting the wave absorbing performance;

[0027] According to the transmission coefficient obtained in step S22, the Box-plot method is used to analyze the effect of a single factor on the transmission coefficient K t By judging the change of the median as the parameter value changes, the degree of influence of the parameter on the wave elimination performance can be obtained;

[0028] Step S24: Perform t-SNE dimension reduction and visualization processing on the independent variable parameters;

[0029] According to the basic principle of t-sne, the relative water depth kh, wave steepness kH / 2, relative immersion depth of oyster reef d / h, relative length of oyster reef kL / (2π), relative distance between oyster reefs kL s / (2π), riverbed slope s, number of oyster reefs N and transmission coefficient K t Reduce the dimension to two-dimensional space for subsequent clustering operations.

[0030] Preferably, in step S3, the weighted K-means clustering method is used to cluster the pre-processed data according to the transmission coefficient K t Perform clustering and analyze the clustering results to determine the parameter selection range that can maximize the wave absorption performance of the submerged dike. The specific process is as follows:

[0031] Step S31, determining the number of cluster centers;

[0032] According to the basic principle of K-means clustering, the goal of clustering is to make the total square error of each sample point to the cluster center closest to it, then the square error sum of the data set SSE is as follows:

[0033]

[0034] Among them, the size of SSE indicates the quality of clustering results; x is the data object; C i is the i-th cluster center; K is the number of cluster centers;

[0035] Among them, the maximum value of K does not exceed the square root of the sample size N, as shown below:

[0036]

[0037] Step S32: introducing a weight vector w and assigning weights according to the degree of influence of different environmental parameters on the design, thereby identifying key design parameters and optimizing the design scheme;

[0038] After considering the influence of each independent variable parameter, the weight distribution is selected as follows:

[0039] w=[0.04,0.04,0.04,0.04,0.04,0.0.4,0.04,0.72];

[0040] Step S33: Weighted K-means optimizes the distribution of cluster centers and data points through iteration until convergence, thereby achieving clustering operation.

[0041] Preferably, in step S33, the clustering process is as follows:

[0042] Step S331: assigning data points to the nearest cluster center;

[0043] For each data point x i , calculate its relationship with each cluster center C j The weighted distance is as follows:

[0044] d ij =w i ·||x i -C j ||;

[0045] Among them, d ij is the data point x i and cluster center C j The weighted distance of i is the data point x i The weight of ||x i -C j || is the data point x i and cluster center C j The Euclidean distance between

[0046] Step S332: Update the cluster center;

[0047] For each cluster j, recalculate its cluster center C' by taking the weighted average of all current data points j , as shown below:

[0048]

[0049] Among them, x i ∈Cluster j Represents all data points x that currently belong to the jth cluster i ;

[0050] Step S333: Check convergence;

[0051] The change between the current cluster center and the cluster center of the previous iteration is calculated. If the change is less than a preset threshold, the algorithm converges and the iteration stops; otherwise, return to step S331 to continue iteration.

[0052] Preferably, in step S4, the geometry, material selection, and structural arrangement of the submerged dike are optimized according to the cluster analysis results. The specific process is as follows:

[0053] The clustering results were analyzed and the Box-plot method was used to obtain the parameters with greater influence. Under the condition of reducing the transmission wave height, the value range of the corresponding independent variable parameters was obtained by drawing a graph. The weighted K-means method was used to obtain the K t The main factors are clustered and the interaction effects between the independent variable parameters are comprehensively considered to obtain the parameter value range with the best wave-breaking performance. Based on the conclusions, the structure, material and layout of the underwater submerged dike are optimized to achieve the best wave-breaking performance.

[0054] Therefore, the present invention adopts the above-mentioned underwater submerged dike design and optimization method based on the weighted K-means method, and the beneficial effects are as follows:

[0055] (1) Accuracy: By introducing the weighted K-means clustering method, the present invention can more accurately analyze marine environmental data and improve the accuracy of submerged dike design;

[0056] (2) Economical: The present invention can effectively reduce engineering costs and improve the stability and durability of the submerged dike by optimizing design parameters;

[0057] (3) Universality: The method of the present invention has strong versatility and can be applied to the design and optimization of different types of underwater submerged dikes;

[0058] (4) Speed: Compared with traditional optimization methods, the present invention can greatly reduce the amount of calculation and complexity and can draw conclusions in a short time.

[0059] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 This is a simplified experimental model diagram corresponding to the present invention;

[0061] Figure 2 The distribution diagram of sample points tested at different d / h;

[0062] Figure 3 It is a schematic diagram of the principle of the Box-plot method;

[0063] Figure 4 The Box-plot method is applied to the seven independent variable parameters in this invention; a is the relative water depth kh; b is the wave steepness kH / 2; c is the relative immersion depth of the oyster reef d / h; d is the relative length of the oyster reef kL / (2π); e is the relative distance between oyster reefs kLs / (2π); f is the riverbed slope s; g is the number of oyster reefs N;

[0064] Figure 5 This is the cluster visualization effect diagram after applying t-sne technology;

[0065] Figure 6 is the sum of squared errors (SSE) of the clustering in the present invention;

[0066] Figure 7 It is the feature map of each class after clustering is completed;

[0067] Figure 8 is the distribution of clustering results under seven independent variable parameters; where a is the relative water depth kh; b is the wave steepness kH / 2; c is the relative immersion depth of oyster reef d / h; d is the relative length of oyster reef kL / (2π); e is the relative distance between oyster reefs kL s / (2π); f is the riverbed slope s; g is the number of oyster reefs N;

[0068] Figure 9 is the value range of the three main parameters; among them, a is the relative immersion depth of oyster reef d / h; b is the relative length of oyster reef kL / (2π); and c is the number of oyster reefs N. DETAILED DESCRIPTION

[0069] The technical solution of the present invention is further described below with reference to the accompanying drawings and embodiments.

[0070] The present invention provides an underwater submerged dike design and optimization method based on a weighted K-means method, comprising the following steps:

[0071] Step S1: Conduct an experiment according to the experimental combination obtained by the D-optimal design method and collect experimental data.

[0072] Step S2: Pre-process the collected test data and calculate the transmission coefficient K under each working condition. t , determine the main factors affecting the wave-breaking performance of the submerged dike, and normalize the independent variable parameters for subsequent clustering operations.

[0073] Step S3: Use the weighted K-means clustering method to cluster the pre-processed data according to the transmission coefficient K t Clustering is performed, and the obtained clustering results are analyzed to determine the parameter selection range that can maximize the wave absorption performance of the submerged dike.

[0074] Step S4: Optimize the geometry, material selection, and structural layout of the submerged dike based on the cluster analysis results.

[0075] Example

[0076] Step S1: Conduct an experiment according to the experimental combination obtained by the D-optimal design method and collect experimental data.

[0077] Taking oyster reefs as the research object, the experiment was carried out in the laboratory according to the experimental combination given by the D-optimal design method. By adjusting the wave machine parameters, different marine environments were simulated for experiments. At the same time, data from four wave height meters were collected and recorded for subsequent analysis.

[0078] The collected test data specifically include: transmission wave height H t The relative water depth (kh), wave steepness (kH / 2), relative immersion depth of oyster reefs (d / h), relative length of oyster reefs (kL / (2π)), relative distance between oyster reefs (kL s / (2π)), riverbed slope (s) and number of oyster reefs (N).

[0079] The present invention uses the D-optimal design method to ensure that the test combination is as comprehensive as possible, while reducing the number of design groups. In the present invention, it is reduced to 0.5% of the original number, which greatly reduces the time for physical model experiments and numerical simulation calculations.

[0080] Step S2: Pre-process the collected test data and calculate the transmission coefficient K under each working condition. t , determine the main factors affecting the wave-breaking performance of the submerged dike, and normalize the independent variable parameters for subsequent clustering operations.

[0081] Step S21: pre-process the collected test data.

[0082] like Figure 1 As shown in the figure, the data of the water surface height measured by the four wave height meters over time are collected. Since the underwater reef has a certain reflection effect on the waves, the data collected by wave height meters 1 and 2 contain not only the incident wave signal but also the reflected wave signal. Therefore, they need to be separated and processed as shown below:

[0083]

[0084] η1=A1 cosωt+B1 sinωt;

[0085] η2=A2 cosωt+B2 sinωt;

[0086] Among them, a i is the incident wave amplitude, a r is the reflected amplitude, k is the wave number, Δl is the distance between wave height meters No. 1 and No. 2; η1 and η2 are the measured data of wave height meters No. 1 and No. 2 respectively, from which the values ​​of parameters A1, B1 and A2, B2 are calculated.

[0087] Step S22: Calculate the transmission coefficient K under each working condition t .

[0088] After separating the incident wave and the reflected wave, the transmission wave height is obtained and the transmission coefficient K is obtained. t , as shown below:

[0089]

[0090] Among them, K t is the transmission coefficient; H t is the transmission wave height measured in the experiment; H i is the incident wave height; a t is the transmission amplitude.

[0091] Step S23: determine the main factors affecting the wave absorbing performance.

[0092] According to the transmission coefficient obtained in step S22, we can know the quality of wave absorption performance under different working conditions, but we cannot know the degree of influence of a single factor on the wave absorption performance of the underwater submerged dike. Therefore, in order to determine the main factors affecting the wave absorption performance, the Box-plot method can be used to quickly analyze the influence of a single factor on the transmission coefficient K. t The influence of the test results is eliminated, thereby eliminating the independent variable factors that have little impact on the test results, achieving the purpose of quickly reducing the number of independent variables in the test, thereby reducing the subsequent workload and saving test time.

[0093] like Figure 2 As shown, the transmission coefficient K corresponding to different values ​​of a single parameter is t Draw in the figure, the specific operation process of the Box-plot method is as follows Figure 3 As shown, the sample points drawn can show the upper and lower bounds, median, upper and lower quartiles, and outliers of the sample. t The median is used to determine the influence of the parameter on the wave-absorbing performance as the parameter value changes.

[0094] like Figure 4 As shown in the figure, the relative water depth (kh), wave steepness (kH / 2), relative immersion depth of oyster reefs (d / h), relative length of oyster reefs (kL / (2π)), relative distance between oyster reefs (kL s / (2π)), riverbed slope (s), and number of oyster reefs (N) are seven independent variable parameters. The transmission coefficient change curve is drawn as the seven independent variable parameters change, and the median of the transmission coefficient corresponding to all working conditions is obtained as the individual factor changes.

[0095] By judging the change of the median as the parameter values ​​change, the degree of influence of the parameters on the wave-breaking performance is obtained. Several parameters whose medians change significantly as the parameter values ​​change are taken as the main influencing factors, and their influence is mainly considered to achieve the purpose of reducing independent variables and simplifying the model.

[0096] Therefore, the analysis shows that among the seven independent variable parameters, the relative immersion depth of oyster reefs (d / h), the relative length of oyster reefs (kL / (2π)) and the number of oyster reefs (N) have the greatest impact on the results of the transmission coefficient, while the other independent variable parameters have no significant impact on the wave absorption performance of the submerged dike. Therefore, the focus is on the above three independent variable factors.

[0097] Step S24: perform t-SNE dimensionality reduction and visualization processing on the independent variable parameters.

[0098] According to the basic principle of t-SNE, it is a nonlinear dimensionality reduction technology, which is mainly used to reduce high-dimensional data to two-dimensional or three-dimensional space in order to better observe and visualize the structure and characteristics of the data.

[0099] Therefore, the t-SNE dimensionality reduction technique was used to reduce the relative water depth (kh), wave steepness (kH / 2), relative immersion depth of oyster reefs (d / h), relative length of oyster reefs (kL / (2π)), relative distance between oyster reefs (kL s / (2π)), riverbed slope (s), number of oyster reefs (N) and transmission coefficient K t Dimensionality reduction to two-dimensional space Figure 5 As shown, in order to carry out subsequent clustering operations.

[0100] Step S3: Use the weighted K-means clustering method to cluster the pre-processed data according to the transmission coefficient K t Clustering is performed, and the obtained clustering results are analyzed to determine the parameter selection range that can maximize the wave absorption performance of the submerged dike.

[0101] Step S31: Determine the number of cluster centers.

[0102] According to the basic principle of K-means clustering, the goal of clustering is to make the total sum of squared errors from each sample point to its nearest cluster center, that is, the clustering cost function SSE, as small as possible. Therefore, the sum of squared errors SSE of the entire dataset is as follows:

[0103]

[0104] Among them, the size of SSE indicates the quality of clustering results; x is the data object; C i is the i-th cluster center; K is the number of cluster centers.

[0105] Theoretically, as K increases, SSE decreases monotonically, because the increase in the number of cluster centers means that some sample points will save the next distance by being assigned to new cluster centers, until K = N, when each sample becomes a separate class, at which point the SSE value drops to 0. However, in order to minimize SSE while minimizing the number of cluster centers, it is generally ensured that the maximum value of K does not exceed the square root of the sample size N, as shown below:

[0106]

[0107] like Figure 6 As shown in FIG, it can be clearly seen that when K=6, the drop rate of the SSE image decreases rapidly, so K=6 is preferably used as the number of cluster centers.

[0108] Step S32: Introduce a weight vector w and assign weights according to the degree of influence of different environmental parameters on the design, so as to more accurately identify key design parameters and optimize the design solution.

[0109] In order to better reflect the importance or influence of data and improve clustering accuracy, we add weight w to K-means clustering. We can assign weights based on the degree of influence of different environmental parameters on the design, thereby more accurately identifying key design parameters and optimizing the design scheme. After fully considering the influence of each independent variable parameter, the weight distribution is selected as follows:

[0110] w=[0.04,0.04,0.04,0.04,0.04,0.0.4,0.04,0.72].

[0111] Step S33: Weighted K-means optimizes the distribution of cluster centers and data points through iteration until convergence, thereby achieving clustering operation.

[0112] The clustering process can be performed iteratively using a computer. The specific process is as follows:

[0113] Step S331: Assign data points to the nearest cluster center.

[0114] For each data point x i , calculate its relationship with each cluster center C j The weighted distance is as follows:

[0115] d ij =w i ·||x i -C j ||;

[0116] Among them, d ij is the data point x i and cluster center Cj The weighted distance of i is the data point x i The weight of ||x i -C j || is the data point x i and cluster center C j The Euclidean distance between them is calculated as follows:

[0117]

[0118] Step S332: Update the cluster center.

[0119] For each cluster, recalculate its cluster center C' by taking the weighted average of all current data points j , as shown below:

[0120]

[0121] Among them, x i ∈Cluster j Represents all data points x that currently belong to the jth cluster i .

[0122] Step S333: Check convergence.

[0123] Calculate the change between the current cluster center and the cluster center of the previous iteration. If the change is less than a preset threshold, the algorithm converges and stops iterating; otherwise, return to step S331 to continue iterating.

[0124] Step S4: Optimize the geometry, material selection, and structural layout of the submerged dike based on the cluster analysis results.

[0125] Analyze the clustering results obtained, focusing on the transmission coefficient K t The smallest class, combined with the parameters with greater influence obtained by the Box-plot method, can be used to obtain the value range of the corresponding independent variable parameters by drawing a graph while minimizing the transmission wave height. t The main factors are clustered and the interaction effects between the independent variable parameters are comprehensively considered to obtain the parameter value range with the best wave elimination performance.

[0126] Based on the conclusions drawn, the structure, materials and layout of the underwater dike can be optimized to achieve the best wave-breaking performance.

[0127] like Figure 7 As shown in Figure 1, all samples are divided into six categories according to K=6. As shown in Table 1, the transmission coefficient of the first category is the smallest, so we focus on the value range of the first category parameters.

[0128] Table 1 The value range of each parameter after clustering is completed

[0129] Clustering <![CDATA[K t <!-- 7 -->]]> Cluster 1 [0.19，0.52) Cluster 2 [0.52，0.70) Cluster 3 [0.70，0.82) Cluster 4 [0.82，0.90) Cluster 5 [0.90，0.95) Cluster 6 [0.95，1.00]

[0130] like Figure 8 As shown, through data analysis, it can be seen that: when the independent variables of interest are the three independent variable parameters of oyster reef relative immersion depth (d / h), oyster reef relative length (kL / (2π)), and oyster reef number (N), the distribution of the first category is obviously limited to a certain range of values, while the six clustering results of the other independent variable parameters are roughly evenly distributed. This also confirms to a certain extent that the three independent variable parameters of oyster reef relative immersion depth (d / h), oyster reef relative length (kL / (2π)), and oyster reef number (N) have the greatest impact on wave absorption performance, while the impact of other independent variables can be ignored. Through analysis, it is concluded that when the wave absorption performance is optimal, such as Figure 9 As shown in the figure, the value ranges of the three main independent variable parameters are:

[0131] relative immersion depth of oyster reef (d / h)∈[0.55, 0.70];

[0132] Oyster reef relative length (kL / (2π))∈[0.22, 0.30];

[0133] Number of oyster reefs N = 3 or 4;

[0134] Based on the conclusions drawn, the structure, materials and layout of the underwater dike can be optimized to achieve the best wave-breaking performance, which is of great significance in practical engineering applications.

[0135] Therefore, the present invention adopts the above-mentioned underwater submerged dike design and optimization method based on the weighted K-means method. By introducing the weighted K-means clustering method, it can more accurately analyze marine environmental data and improve the accuracy of submerged dike design; by optimizing design parameters, it can effectively reduce engineering costs and improve the stability and durability of submerged dikes; the method proposed in the present invention has strong versatility and can be applied to different types of underwater submerged dike design and optimization; compared with traditional optimization methods, it can greatly reduce the amount of calculation and complexity, and can draw conclusions in a short time.

[0136] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit the same. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that they can still modify or replace the technical solutions of the present invention with equivalents, and these modifications or equivalent replacements cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for designing and optimizing underwater submerged dikes based on the weighted K-means method, characterized in that: The following steps are involved: Step S1: Conducting tests according to the test combination obtained by the D-optimal design method and collecting test data; According to the D-optimal design method, a test combination is given. By adjusting the wave generator parameters, different ocean environments are simulated for testing. At the same time, the wave height meter data is collected and recorded for subsequent analysis. The collected test data specifically include: transmission wave height H t Relative water depth kh, wave steepness kH / 2, relative immersion depth of oyster reef d / h, relative length of oyster reef , relative distance between oyster reefs , the relationship between the seven parameters of riverbed slope s and the number of oyster reefs N; Step S2: pre-process the collected test data, calculate the transmission coefficient under each working condition, use the Box-plot method to analyze the effect of a single parameter on the transmission coefficient, and determine the main parameters that affect the wave-breaking performance of the submerged dike; Step S3: clustering the preprocessed data according to the transmission coefficient using a weighted K-means clustering method; Step S4: Determine the parameter selection range that can maximize the wave-absorbing performance of the submerged dike based on the cluster analysis results, and optimize the geometry, material selection, and structural layout of the submerged dike; Analyze the clustering results obtained, focusing on the transmission coefficient The smallest class, combined with the main parameters obtained by the Box-plot method, can obtain the value range of the corresponding parameters by drawing a graph while minimizing the transmission wave height; Based on the conclusions drawn, the structure, materials and layout of the underwater dike are optimized to achieve the best wave-breaking performance.

2. The underwater submerged dike design and optimization method based on the weighted K-means method according to claim 1, characterized in that: In step S2, the collected test data is preprocessed to calculate the transmission coefficient under each working condition. , determine the main parameters that affect the wave-absorbing performance of the submerged dike. The specific process is as follows: Step S21, pre-processing the collected test data; Since the collected data contains incident wave signals and reflected wave signals, the data is separated and processed to obtain the incident wave amplitude and reflected wave amplitude, as shown below: ； ； ； ； in, is the incident wave amplitude, is the reflected amplitude, is the wave number, is the distance between wave height meters No. 1 and No. 2; and The measured data of wave height meters 1 and 2 are respectively, and the parameters are obtained from them 、 and 、 The value of Step S22: Calculate the transmission coefficient under each working condition ; After separating the incident wave and the reflected wave, the transmission wave height is obtained and the transmission coefficient is calculated. , as shown below: ； in, is the transmission coefficient; is the transmission wave height measured in the experiment; is the incident wave height; is the transmitted amplitude; is the incident wave amplitude; Step S23, determining the main parameters affecting the wave absorbing performance; According to the transmission coefficient obtained in step S22, the Box-plot method is used to analyze the effect of a single parameter on the transmission coefficient. By judging the change of the median as the parameter value changes, the degree of influence of the parameter on the wave elimination performance can be obtained; Step S24: perform t-SNE dimension reduction and visualization processing on the parameters; According to the basic principle of t-sne, the relative water depth kh, wave steepness kH / 2, relative immersion depth of oyster reef d / h, and relative length of oyster reef are calculated. , relative distance between oyster reefs , riverbed slope s, number of oyster reefs, and the transmission coefficient Reduce the dimension to two-dimensional space for subsequent clustering operations.

3. The underwater submerged dike design and optimization method based on the weighted K-means method according to claim 1, characterized in that: In step S3, the weighted K-means clustering method is used to cluster the preprocessed data according to the transmission coefficient Clustering is performed, and the specific process is as follows: Step S31, determining the number of cluster centers; According to the basic principle of K-means clustering, the goal of clustering is to make the total square error of each sample point to the cluster center closest to it, then the square error sum of the data set SSE is as follows: ； in, The size of indicates the quality of the clustering result; For data objects; is the i-th cluster center; is the number of cluster centers; Among them, the maximum value of K does not exceed the square root of the sample size N, as shown below: ； Step S32: Introduce weight vector , assign weights according to the degree of influence of different environmental parameters on the design, thereby identifying key design parameters and optimizing the design scheme; After considering the influence of each independent variable parameter, the weight distribution is selected as follows: ； Step S33: Weighted K-means optimizes the distribution of cluster centers and data points through iteration until convergence, thereby achieving clustering operation.

4. The underwater submerged dike design and optimization method based on the weighted K-means method according to claim 3, characterized in that: In step S33, the clustering process is as follows: Step S331: assigning data points to the nearest cluster center; For each data point x i , calculate its relationship with each cluster center The weighted distance is as follows: ； in, is a data point and cluster centers The weighted distance of is a data point The weight of is a data point and cluster centers The Euclidean distance between Step S332: Update the cluster center; For each cluster, recalculate its cluster center by taking the weighted average of all current data points , as shown below: ； in, Indicates that the current All data points of the cluster ; Step S333: Check convergence; The change between the current cluster center and the cluster center of the previous iteration is calculated. If the change is less than a preset threshold, the algorithm converges and the iteration stops; otherwise, return to step S331 to continue iteration.

Images