A layout method for deep-sea offshore observation points based on multi-element fusion
By using a multi-factor fusion method, a standardized multidimensional spatiotemporal dataset is constructed. Improved GMM-p and KM-D algorithms are used for clustering. Combined with centroid fusion and weighted fusion strategies, the problem of insufficient representativeness of observation point layout in existing technologies is solved, and more robust and comprehensive marine environmental monitoring is achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- EAST CHINA SEA FORECAST CENT OF THE STATE OCEANIC ADMINISTRATION
- Filing Date
- 2026-04-03
- Publication Date
- 2026-08-04
AI Technical Summary
Existing methods for setting up observation points are mostly based on the analysis of a single environmental element, failing to fully consider differences in water depth, spatial and temporal environmental changes, and ecologically sensitive areas. This results in insufficient representativeness of the monitoring data and makes it difficult to reflect the comprehensive impact of multiple marine environmental elements.
By using a multi-factor fusion method, a standardized multidimensional spatiotemporal dataset is constructed. An improved GMM-p algorithm and KM-D algorithm are used for clustering. By combining centroid fusion and weighted fusion strategies, the cohesion index is calculated to select the optimal layout location.
This improves the scientific rigor and stability of the observation point layout, enabling a more comprehensive reflection of the multi-element characteristics of the marine environment and enhancing the representativeness of the data and the robustness of the clustering results.
Smart Images
Figure CN121960237B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of ocean observation point layout technology, specifically to a layout method for deep-sea offshore observation points based on multi-element fusion. Background Technology
[0002] Deep-sea areas are influenced by complex ocean dynamic processes, resulting in significant spatiotemporal variations in their hydrology, meteorology, and marine ecological environment. To achieve long-term, continuous observation of the deep-sea environment, it is typically necessary to deploy offshore observation points in the target sea area to collect data on various marine environmental elements, such as ocean currents, waves, temperature, salinity, and meteorological parameters. However, the construction and maintenance costs of deep-sea observation points are high, and adjustments to their locations are difficult once selected. Therefore, rationally planning the layout of observation points is crucial for improving the coverage and data representativeness of the marine observation network.
[0003] Existing methods for deploying observation points are mostly based on the analysis of single environmental factors or rely solely on expert experience. Traditional buoy deployment does not consider differences in water depth, spatiotemporal environmental changes, and ecologically sensitive areas, resulting in insufficient representativeness of monitoring data and an inability to fully reflect the comprehensive impact of multiple marine environmental factors. Furthermore, marine environmental factors exhibit significant nonlinear, multi-scale, and high-dimensional characteristics, making it difficult to obtain stable and physically meaningful results through direct clustering and deployment analysis based on raw data. Summary of the Invention
[0004] (a) Purpose of the invention The purpose of this invention is to provide a method for the layout of deep-sea offshore observation points based on multi-factor fusion. This method achieves scientific and intelligent layout of deep-sea observation points through multi-factor spatiotemporal data standardization, improved clustering algorithms, and multi-strategy fusion. An agglomeration index is introduced to screen for the optimal layout scheme; this method comprehensively reflects the multi-factor characteristics of the marine environment, improving the stability and representativeness of the clustering and fusion results.
[0005] (II) Technical Solution To address the above problems, this invention provides a method for the layout of offshore observation points in deep-sea areas based on multi-element fusion, comprising: The observation data and remote sensing data of various marine environmental elements in the target sea area are divided into time domains to obtain standardized spatiotemporal datasets of each marine environmental element. Based on a preset clustering algorithm, the standardized spatiotemporal dataset is clustered to obtain initial cluster centers corresponding to each marine environmental element. The preset clustering algorithm includes an improved GMM-p algorithm and an improved KM-D algorithm. Based on a preset fusion strategy, the initial cluster centers are fused using multiple factors to obtain fused cluster centers. The preset fusion strategy includes a centroid fusion strategy and a weighted fusion strategy. Each clustering algorithm is matched one-to-one with each fusion strategy to obtain multiple clustering combinations; Based on a preset agglomeration algorithm, the spatial deviation between the initial cluster centers and the fused cluster centers corresponding to the agglomeration combination is calculated to obtain the agglomeration index; The fusion cluster center corresponding to the fusion combination with the smallest fusion index is determined as the layout location of the deep-sea offshore observation point.
[0006] In another aspect of the present invention, preferably, the various marine environmental elements include: sea temperature, salinity, wind speed, pH value, and wave height; The observation data and remote sensing data of various marine environmental elements in the target sea area are divided into time domains to obtain standardized spatiotemporal datasets for each marine environmental element, including: The observation data and remote sensing data are preprocessed, including the removal of outlier data, to obtain preprocessed data. Preprocessed data from different sources are divided into temporal and spatial scales to obtain spatiotemporal datasets; The spatiotemporal dataset is standardized to obtain a standardized spatiotemporal dataset.
[0007] In another aspect of the present invention, preferably, the improved GMM-p algorithm includes: Dimensionality reduction processing is performed on the standardized spatiotemporal datasets corresponding to each marine environmental element; A Gaussian mixture model is constructed based on the dimensionality-reduced data, and a probabilistic optimization strategy is used for clustering calculations. Based on the clustering results of the Gaussian mixture model, the first initial cluster centers with a preset number of cluster centers are obtained.
[0008] In another aspect of the present invention, preferably, the improved KM-D algorithm includes: The K-means++ method was used to initialize the standardized spatiotemporal dataset to obtain candidate cluster centers; Based on the candidate cluster centers, the DTW distance algorithm is used to measure the similarity between each time series data and each candidate cluster center in the standardized spatiotemporal dataset, and each time series data is assigned to the cluster of the candidate cluster center with the smallest distance. The DBA algorithm is used to calculate the mean sequence of time series data within each cluster in order to obtain new candidate cluster centers. The above steps are iteratively executed based on the new candidate cluster centers until the clustering results converge, obtaining the second initial cluster centers with the preset number of cluster centers.
[0009] In another aspect of the present invention, preferably, the center-of-gravity fusion strategy includes: Based on the initial cluster centers corresponding to each marine environmental element, the contour coefficients corresponding to each marine environmental element are calculated, and the contour coefficients are normalized to obtain the corresponding first element weights. The coordinates of the initial cluster centers of each marine environmental element are arithmetically weighted with the corresponding first element weights to obtain the coordinates of the centroid fusion cluster centers.
[0010] In another aspect of the present invention, preferably, the coordinates of the centroid fusion cluster center are calculated using the following formula: ; ; in, The longitude coordinates of the centroid fusion cluster center are represented. The latitudinal coordinates of the centroid fusion cluster center are represented. This represents the first element weight of the j-th marine environmental element. This represents the longitude coordinates of the initial cluster center of the j-th marine environmental element. Represents the latitude coordinates of the initial cluster center of the j-th marine environmental element.
[0011] In another aspect of the present invention, preferably, the weighted fusion strategy includes: The coordinates of the initial cluster centers corresponding to each marine environmental element are standardized to obtain standardized initial cluster centers; Allocate the same weight for the second factor to each marine environmental element; The weighted fusion cluster centers are obtained by arithmetic averaging based on the standardized initial cluster centers and the weights of the second element.
[0012] In another aspect of the present invention, preferably, the preset coagulation algorithm is based on the Haversine formula; The method based on a preset agglomeration algorithm calculates the spatial deviation between the initial cluster centers and the fused cluster centers corresponding to the agglomeration combination, and obtains the agglomeration index, including: Based on the Havesay formula, the spherical distance between the initial cluster center and the fusion cluster center corresponding to each marine environmental element is calculated according to the difference in latitude and longitude. The agglomeration index is obtained by summing the spherical distances from the initial cluster centers of each marine environmental element to the corresponding fusion cluster centers.
[0013] In another aspect of the invention, preferably, the spherical distance is expressed using the following formula: ; in, R represents the spherical distance, R represents the Earth's radius, and c represents the angle of a great circle.
[0014] In another aspect of the present invention, preferably, the large arc angle is calculated using the following formula: ; ; in, , Δlat represents the latitude of two points, Δlon represents the latitude difference between the two points, and a represents the intermediate calculation quantity of the Haversian formula, corresponding to the semi-versus relationship in spherical trigonometry.
[0015] (III) Beneficial Effects The above-described technical solution of the present invention has the following beneficial technical effects: This invention simultaneously incorporates multiple marine environmental elements to construct a standardized multidimensional spatiotemporal dataset, elevating layout analysis from a single-element framework to a multi-element comprehensive evaluation level. This allows for a more comprehensive reflection of the dynamic processes and environmental change characteristics of the target sea area. Improved GMM-p and KM-D algorithms are employed to cluster the multi-element data separately. In the fusion stage, two strategies—centroid fusion and balanced weight fusion—are used. These different strategies differ in their handling of element weight allocation. By constructing multiple agglomerative combinations using multiple clustering algorithms and fusion strategies, and calculating the agglomeration index of these combinations, the optimal combination is selected. This comprehensively reflects the multi-element characteristics of the marine environment, improving the stability and representativeness of the clustering and fusion results, making the layout results more robust and comprehensive. Attached Figure Description
[0016] Figure 1 This is an overall flowchart of one embodiment of the present invention. Detailed Implementation
[0017] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments and the accompanying drawings. It should be understood that these descriptions are merely exemplary and not intended to limit the scope of the invention. Furthermore, descriptions of well-known structures and techniques are omitted in the following description to avoid unnecessarily obscuring the concept of the invention.
[0018] Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.
[0019] In the description of this invention, it should be noted that the terms "first," "second," and "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.
[0020] Furthermore, the technical features involved in the different embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.
[0021] The invention will now be described in more detail with reference to the accompanying drawings. In the various drawings, the same elements are indicated by similar reference numerals. For clarity, the various parts in the drawings are not drawn to scale.
[0022] Example 1 A method for the layout of offshore observation points in deep sea based on multi-element fusion. Figure 1 An overall flowchart of one embodiment of the present invention is shown, as follows: Figure 1 As shown, it includes: Observational and remote sensing data of various marine environmental elements in the target sea area are divided into time-domain data to obtain standardized spatiotemporal datasets for each marine environmental element. These elements include sea surface temperature, salinity, wind speed, pH value, and wave height. Time-domain division allows for the acquisition of standardized spatiotemporal datasets for each element at different time periods, eliminating differences caused by variations in scale, unit, and time, achieving a unified representation of multi-element data, and providing a reliable foundation for subsequent analysis. In this embodiment, the time-domain division of observational and remote sensing data of various marine environmental elements in the target sea area to obtain standardized spatiotemporal datasets for each marine environmental element includes: The observational and remote sensing data are preprocessed, including the removal of outliers, to obtain preprocessed data. Outlier removal can be achieved through statistical analysis methods, such as threshold judgment based on mean and standard deviation or consistency checks based on neighboring spatiotemporal data, to remove observations that significantly deviate from the normal range, thus ensuring data quality reliability. In this embodiment, the 3σ criterion is used to remove outliers.
[0023] Preprocessed data from different sources are divided into temporal and spatial scales to obtain spatiotemporal datasets. Temporal scale division can be based on hours, days, ten-day periods, or months, segmenting continuous observation data and remote sensing image data to form time-series spatiotemporal data blocks. Spatial scale division can be based on gridded marine areas or functional zoning of key regions, mapping observation data and remote sensing data to unified spatial units for spatial comparison and analysis. In this embodiment, a unified "hourly" temporal resolution and a 0.25°×0.25° spatial grid are used.
[0024] The spatiotemporal dataset is standardized to obtain a standardized spatiotemporal dataset.
[0025] Based on a pre-defined clustering algorithm, the standardized spatiotemporal dataset is clustered to obtain initial cluster centers corresponding to each marine environmental element. The pre-defined clustering algorithm includes an improved GMM-p algorithm and an improved KM-D algorithm. The improved GMM-p algorithm models the data distribution using a Gaussian mixture model, improving its ability to identify complex spatial patterns. The improved KM-D algorithm combines K-means++ initialization and DTW distance metric, optimizing the cluster centers of time-series data through dynamic time warping, thus more accurately reflecting the spatial and temporal characteristics of marine environmental elements. The initial cluster centers are used to preliminarily determine the representative spatial location of each marine environmental element.
[0026] Furthermore, in this embodiment, the improved GMM-p algorithm includes: Dimensionality reduction is performed on the standardized spatiotemporal datasets corresponding to various marine environmental elements. Because marine environmental elements have significant spatiotemporal dimensionality characteristics, the original data has a high dimensionality, and direct clustering can easily lead to high computational cost and poor clustering stability. Therefore, this embodiment employs dimensionality reduction methods such as Principal Component Analysis (PCA), Locally Linear Embedding (LLE), or Multidimensional Scaling Analysis (MDS) to reduce the dimensionality of the standardized spatiotemporal data, extracting core feature vectors that best preserve data variability and structural characteristics. The dimensionality of the data is significantly reduced after dimensionality reduction, facilitating the subsequent establishment of Gaussian mixture models and improving model convergence efficiency and clustering performance.
[0027] A Gaussian mixture model is constructed based on the dimensionality-reduced data, and a probabilistic optimization strategy is used for clustering calculations. The Gaussian mixture model models the probability distribution of spatiotemporal data through a linear combination of multiple Gaussian distributions, effectively characterizing the multimodal distribution features of marine environmental elements. In this embodiment, a probabilistic optimization strategy is introduced to solve for the model parameters. This strategy includes the Expectation-Maximization (EM) iterative algorithm, regularization constraints, and a dynamic parameter adjustment mechanism. The EM algorithm iteratively updates the mean, covariance matrix, and mixing coefficients, enabling the model to fully fit the probability distribution of the data. By adding a regularization term to the objective function, overfitting is avoided, improving the robustness of the clustering results. Furthermore, by setting a dynamic adjustment strategy, the learning rate or weight parameters can be automatically adjusted according to the model's convergence trend during iteration, thereby improving the model's convergence speed and stability.
[0028] Based on the clustering results of the Gaussian mixture model, a first initial cluster center with a preset number of cluster centers is obtained. Based on the clustering probability results output by the optimized Gaussian mixture model, the category to which each sample point belongs is determined, and the first initial cluster center with a preset number of cluster centers is calculated accordingly. Specifically, multiple initial cluster centers can be obtained by using the mean vector corresponding to each Gaussian component as the center position of that category; or the cluster centers can be further refined by weighted averaging of the model allocation results, making the cluster centers closer to the actual data distribution. This embodiment, by selecting a preset number of cluster centers, can effectively summarize the spatial distribution of data for various marine environmental elements, thereby providing accurate representative locations for subsequent multi-element fusion and observation point layout. Using the improved GMM-p algorithm described above, this embodiment can obtain more stable and reliable cluster centers while fully preserving the spatiotemporal characteristics of marine environmental elements, improving the scientific rigor and accuracy of the multi-element fusion results.
[0029] Furthermore, in this embodiment, the improved KM-D algorithm includes: The K-means++ method is used to initialize the standardized spatiotemporal dataset to obtain candidate cluster centers. K-means++ introduces a distance-based probability sampling strategy during initialization, ensuring good dispersion of initial cluster centers in the data space. Compared to traditional K-means random initialization, this reduces the risk of clustering getting trapped in local optima and improves the stability and accuracy of the final clustering results. In this embodiment, it is used for initial center selection of the standardized spatiotemporal dataset, providing a more reasonable starting point for subsequent DTW-based time series clustering.
[0030] Based on the candidate cluster centers, the DTW (Dynamic Time Warping) distance algorithm is used to measure the similarity between each time series data point and each candidate cluster center in the standardized spatiotemporal dataset, and each time series data point is assigned to the cluster of the candidate cluster center with the smallest distance. DTW distance allows for flexible matching on the time axis and is suitable for time series data with drift, stretching, or local deformation characteristics, such as sea surface temperature, salinity, and wind speed, thus more accurately characterizing the true similarity between time series data. By calculating the DTW distance between each time series data point and all candidate cluster centers, each time series data point is assigned to the cluster corresponding to the candidate cluster center with the smallest distance, achieving clustering based on the morphological similarity of time series data.
[0031] The DBA (DTW Barycenter Averaging) algorithm is used to calculate the mean sequence of time series data within each cluster to obtain new candidate cluster centers. The DBA algorithm calculates the mean sequence of all time series data within a cluster based on DTW alignment, resulting in a mean sequence that represents the morphological characteristics of that type of time series, which can then be used as new candidate cluster centers. Compared to traditional point-by-point averaging methods, DBA can calculate the mean of time series data under the premise of sequence stretching and alignment, more accurately reflecting typical time series patterns, thereby improving the representativeness of cluster centers.
[0032] The above steps are iteratively executed based on the new candidate cluster centers until the clustering results converge, obtaining a second initial cluster center with a preset number of cluster centers. Based on the updated candidate cluster centers, the clustering step based on DTW distance and the center update step based on DBA are repeated, forming an iterative process. In each iteration, the sequence clustering is recalculated and the centers are updated until the local cluster structure no longer changes, the cluster center update magnitude is lower than a preset threshold, or the maximum number of iterations is reached, at which point the clustering results are considered converged. After the clustering results converge, a second initial cluster center with a preset number of cluster centers is obtained. This cluster center can accurately reflect the typical patterns of temporal changes in various marine environmental elements, has good temporal representativeness and interpretability, and provides a reliable basis for subsequent multi-element fusion.
[0033] Based on a preset fusion strategy, the initial cluster centers are fused using multiple factors to obtain fused cluster centers. The preset fusion strategy includes a centroid fusion strategy and a weighted fusion strategy. The centroid fusion strategy calculates the weighted centroid of each factor's cluster center to reflect the spatial location under the combined influence of multiple factors. The weighted fusion strategy assigns equal weights to different factors, highlighting comprehensive representativeness and balance.
[0034] Furthermore, in this embodiment, the center-of-gravity fusion strategy includes: Based on the initial cluster centers corresponding to each marine environmental element, the silhouette coefficient corresponding to each marine environmental element is calculated using the following formula: ; Where a(i) is the distance from vector i to all other points in its cluster, and b(i) is the average distance from vector i to all points in the nearest cluster that does not contain it. The silhouette coefficient is then normalized to obtain the corresponding first element weight. The silhouette coefficient can be used to evaluate the tightness and separation of the cluster structure. The larger the value, the better the clustering effect and the stronger the representativeness of the element.
[0035] The coordinates of the initial cluster centers of each marine environmental element are arithmetically weighted with their corresponding first-element weights to obtain the coordinates of the centroid fusion cluster center. By weighted summing of the longitude and latitude coordinates of the initial cluster centers of each element, a spatial location that simultaneously considers the characteristics of multiple elements is obtained. The weighting process reflects the contribution of each marine environmental element to the construction of the comprehensive cluster center, making the fusion result more spatially balanced and representative.
[0036] The coordinates of the centroid fusion cluster center are calculated using the following formula: ; ; in, The longitude coordinates of the centroid fusion cluster center are represented. The latitudinal coordinates of the centroid fusion cluster center are represented. This represents the first element weight of the j-th marine environmental element. This represents the longitude coordinates of the initial cluster center of the j-th marine environmental element. This represents the latitude coordinates of the initial cluster center of the j-th marine environmental element. By weighted averaging the initial cluster center coordinates of all marine environmental elements, the resulting centroid fusion cluster center can comprehensively express the spatial variation characteristics of elements such as sea temperature, salinity, wind speed, pH value, and wave height, and is more holistic and representative than the cluster center of a single element.
[0037] Furthermore, in this embodiment, the weighted fusion strategy includes: The coordinates of the initial cluster centers corresponding to each marine environmental element are standardized to obtain standardized initial cluster centers. Since the initial cluster centers of different marine environmental elements may have different data ranges, spatial scales, or numerical distributions, to avoid imbalances in the final fusion location due to differences in numerical scales during direct fusion, this embodiment standardizes the latitude and longitude coordinates of all elements separately. Standardization methods may include range standardization, Z-score standardization, or vector normalization, ensuring that the processed coordinates of each element have similar scale ranges, which helps maintain the fairness and stability of the fusion process.
[0038] Each marine environmental element is assigned the same weight as the second factor; if there are J marine environmental elements in total, each element is assigned a weight of 1 / J, ensuring that all elements have the same contribution in the fusion calculation. This equal-weighting strategy can directly achieve balanced fusion without relying on clustering quality evaluation indicators, ensuring that the fusion process is simple and uniform.
[0039] A weighted fusion cluster center is obtained by arithmetic averaging the standardized initial cluster centers and the weights of the second element. The latitude and longitude coordinates of the standardized initial cluster centers of each element are then weighted and summed. Since all weights are equal, the weighted summation is equivalent to a simple average. The final weighted fusion cluster center coordinates comprehensively reflect the common spatial characteristics of all marine environmental elements, thus forming a representative fusion result that occupies an intermediate position in the multi-element space.
[0040] Each clustering algorithm is matched one-to-one with each fusion strategy to obtain multiple agglomeration combinations; each agglomeration combination corresponds to a specific clustering algorithm and fusion strategy, such as the improved KM-D algorithm and the weighted fusion strategy as an agglomeration combination.
[0041] Based on a preset agglomeration algorithm, the spatial deviation between the initial cluster centers and the fused cluster centers corresponding to the agglomeration combination is calculated to obtain a cohesion index. The cohesion index is used to quantify the stability and consistency of each combination in terms of spatial representativeness; the smaller the value, the better the fusion effect of the combination and the smaller the spatial deviation. In this embodiment, the preset agglomeration algorithm is based on the Haversay formula, which can accurately calculate the spherical distance between two latitude and longitude points in a spherical coordinate system. The calculation of the spatial deviation between the initial cluster centers and the fused cluster centers corresponding to the agglomeration combination based on the preset agglomeration algorithm to obtain the cohesion index includes: Based on the Havelsein formula, the spherical distance between the initial cluster centers and the merged cluster centers corresponding to each marine environmental element is calculated according to the latitude and longitude differences. Since both the initial cluster centers and the merged cluster centers are represented by geographic latitude and longitude coordinates, to ensure the accuracy of the distance calculation, this embodiment uses the Havelsein formula to obtain the great circle distance between the two points. The Havelsein formula can effectively take into account the curvature of the Earth.
[0042] The spherical distance is expressed using the following formula: ; in, R represents the spherical distance, R represents the Earth's radius, and c represents the angle of a great circle.
[0043] The large circle angle is calculated using the following formula: ; ; in, , Let Δlat represent the latitude of the two points, Δlon represent the latitude difference between the two points, and a represent the intermediate calculation quantity of the Haversine formula, which corresponds to the semi-versus relationship in spherical trigonometry.
[0044] The aggregation index is obtained by summing the spherical distances from the initial cluster centers of each marine environmental element to the corresponding fusion cluster centers. This allows for an accurate evaluation of the aggregation effect of each aggregation combination at the geographic spatial level, thereby ensuring that the final selected fusion cluster centers are more representative and reasonable in spatial location, and providing a scientific and reliable criterion for the layout of offshore observation points in deep seas.
[0045] The fusion cluster center corresponding to the fusion combination with the smallest fusion index is determined as the layout location of the deep-sea offshore observation point.
[0046] This invention simultaneously incorporates multiple marine environmental elements to construct a standardized multidimensional spatiotemporal dataset, elevating layout analysis from a single-element framework to a multi-element comprehensive evaluation level. This allows for a more comprehensive reflection of the dynamic processes and environmental change characteristics of the target sea area. Improved GMM-p and KM-D algorithms are employed to cluster the multi-element data separately. In the fusion stage, two strategies—centroid fusion and balanced weight fusion—are used. These different strategies differ in their handling of element weight allocation. By constructing multiple agglomerative combinations using multiple clustering algorithms and fusion strategies, and calculating the agglomeration index of these combinations, the optimal combination is selected. This comprehensively reflects the multi-element characteristics of the marine environment, improving the stability and representativeness of the clustering and fusion results, making the layout results more robust and comprehensive.
[0047] It should be understood that the specific embodiments described above are merely illustrative or explanatory of the principles of the invention and do not constitute a limitation thereof. Therefore, any modifications, equivalent substitutions, improvements, etc., made without departing from the spirit and scope of the invention should be included within the protection scope of the invention. Furthermore, the appended claims are intended to cover all variations and modifications falling within the scope and boundaries of the appended claims, or equivalent forms of such scope and boundaries.
[0048] The present invention has been described above with reference to embodiments thereof. However, these embodiments are merely illustrative and not intended to limit the scope of the invention. The scope of the invention is defined by the appended claims and their equivalents. Various substitutions and modifications can be made by those skilled in the art without departing from the scope of the invention, and all such substitutions and modifications should fall within the scope of the invention.
[0049] Although embodiments of the present invention have been described in detail, it should be understood that various changes, substitutions, and modifications can be made to the embodiments of the present invention without departing from the spirit and scope of the invention.
[0050] Obviously, the above embodiments are merely illustrative examples for clear explanation and are not intended to limit the implementation. Those skilled in the art will recognize that other variations or modifications can be made based on the above description. It is neither necessary nor possible to exhaustively list all possible implementations here. However, obvious variations or modifications derived therefrom are still within the scope of protection of this invention.
Claims
1. A method for the layout of offshore observation points in deep sea based on multi-element fusion, characterized in that, include: The observation data and remote sensing data of various marine environmental elements in the target sea area are divided into time domains to obtain standardized spatiotemporal datasets of each marine environmental element. Based on a preset clustering algorithm, the standardized spatiotemporal dataset is clustered to obtain initial cluster centers corresponding to each marine environmental element. The preset clustering algorithm includes the GMM-p algorithm and the KM-D algorithm. Based on a preset fusion strategy, the initial cluster centers are fused using multiple factors to obtain fused cluster centers. The preset fusion strategy includes a centroid fusion strategy and a weighted fusion strategy. Each clustering algorithm is matched one-to-one with each fusion strategy to obtain multiple clustering combinations; Based on a preset agglomeration algorithm, the spatial deviation between the initial cluster centers and the fused cluster centers corresponding to the agglomeration combination is calculated to obtain the agglomeration index; The fusion cluster center corresponding to the fusion combination with the smallest fusion index is determined as the layout location of the deep-sea offshore observation point. The GMM-p algorithm includes: Dimensionality reduction processing is performed on the standardized spatiotemporal datasets corresponding to each marine environmental element; A Gaussian mixture model is constructed based on the dimensionality-reduced data, and a probabilistic optimization strategy is used for clustering calculations. Based on the clustering results of the Gaussian mixture model, the first initial cluster centers with a preset number of cluster centers are obtained; The KM-D algorithm includes: The K-means++ method was used to initialize the standardized spatiotemporal dataset to obtain candidate cluster centers; Based on the candidate cluster centers, the DTW distance algorithm is used to measure the similarity between each time series data and each candidate cluster center in the standardized spatiotemporal dataset, and each time series data is assigned to the cluster of the candidate cluster center with the smallest distance. The DBA algorithm is used to calculate the mean sequence of time series data within each cluster in order to obtain new candidate cluster centers. Iterative execution is performed based on the new candidate cluster centers until the clustering results converge, obtaining a second initial cluster center with a preset number of cluster centers; The center-of-gravity fusion strategy includes: Based on the initial cluster centers corresponding to each marine environmental element, the contour coefficients corresponding to each marine environmental element are calculated, and the contour coefficients are normalized to obtain the corresponding first element weights. The coordinates of the initial cluster centers of each marine environmental element are arithmetically weighted with the corresponding first element weights to obtain the coordinates of the centroid fusion cluster centers. The weighted fusion strategy includes: The coordinates of the initial cluster centers corresponding to each marine environmental element are standardized to obtain standardized initial cluster centers; Allocate the same weight for the second factor to each marine environmental element; The weighted fusion cluster centers are obtained by arithmetic averaging based on the standardized initial cluster centers and the weights of the second element. The preset coagulation algorithm is based on the Havelsein formula; The method based on a preset agglomeration algorithm calculates the spatial deviation between the initial cluster centers and the fused cluster centers corresponding to the agglomeration combination, and obtains the agglomeration index, including: Based on the Havesay formula, the spherical distance between the initial cluster center and the fusion cluster center corresponding to each marine environmental element is calculated according to the difference in latitude and longitude. The agglomeration index is obtained by summing the spherical distances from the initial cluster centers of each marine environmental element to the corresponding fusion cluster centers.
2. The method for laying out deep-sea offshore observation points based on multi-element fusion according to claim 1, characterized in that, The various marine environmental factors include: sea temperature, salinity, wind speed, pH value, and wave height; The observation data and remote sensing data of various marine environmental elements in the target sea area are divided into time domains to obtain standardized spatiotemporal datasets for each marine environmental element, including: The observation data and remote sensing data are preprocessed, including the removal of outlier data, to obtain preprocessed data. Preprocessed data from different sources are divided into temporal and spatial scales to obtain spatiotemporal datasets; The spatiotemporal dataset is standardized to obtain a standardized spatiotemporal dataset.
3. The method for laying out deep-sea offshore observation points based on multi-element fusion according to claim 2, characterized in that, The coordinates of the centroid fusion cluster center are calculated using the following formula: in, The longitude coordinates of the centroid fusion cluster center are represented. The latitudinal coordinates of the centroid fusion cluster center are represented. This represents the first element weight of the j-th marine environmental element. This represents the longitude coordinates of the initial cluster center of the j-th marine environmental element. Represents the latitude coordinates of the initial cluster center of the j-th marine environmental element.
4. The method for layout of deep-sea offshore observation points based on multi-element fusion according to claim 3, characterized in that, The spherical distance is expressed using the following formula: in, R represents the spherical distance, R represents the Earth's radius, and c represents the angle of a great circle.
5. The method for laying out deep-sea offshore observation points based on multi-element fusion according to claim 4, characterized in that, The large circle angle is calculated using the following formula: in, , Δlat represents the latitude of two points, Δlon represents the latitude difference between the two points, and a represents the intermediate calculation quantity of the Haversian formula, corresponding to the semi-versus relationship in spherical trigonometry.