A marine environment quality evaluation method and system

Through the correction and feature extraction of marine environmental data, a marine environmental quality calculation model is built, and a digital twin algorithm and convolutional neural network are used, combined with dynamic convolution and self-attention mechanisms, the limited coverage and high cost of marine environmental quality evaluation technology are solved. The problem of this paper realizes a comprehensive analysis and evaluation of marine ecosystems and provides a more comprehensive environmental quality assessment.

CN118570739BActive Publication Date: 2024-11-19SECOND INST OF OCEANOGRAPHY MNR
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411054411.9
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-08-02
Publication Date
2024-11-19
Estimated Expiration
2044-08-02

AI Technical Summary

Technical Problem

The existing marine environmental quality evaluation technology faces problems such as limited coverage, difficulty in fully reflecting the conditions of a large-scale marine environment, lack of comprehensive evaluation of the overall health of marine ecosystems, and high costs.

Method used

By obtaining marine environmental data, performing data correction and feature extraction, building a marine environmental quality calculation model, using digital twin algorithms and convolutional neural networks, combining dynamic convolution and self-attention mechanisms, comprehensive analysis and evaluation of marine environmental quality can be achieved.

Benefits of technology

It realizes monitoring of large-scale sea areas, provides a more comprehensive environmental quality assessment, reduces human resource investment, and improves the processing effect of complex marine environmental data.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118570739B_ABST
    Figure CN118570739B_ABST
Patent Text Reader

Abstract

The present invention relates to the technical field of environmental quality evaluation, and specifically to a method and system for evaluating the quality of a marine environment, wherein the evaluation method comprises: acquiring marine environmental data, performing data correction on the acquired data, and constructing the corrected data into a marine environmental data set; using the data set to build a marine environmental quality calculation model; and realizing analysis and evaluation of the marine environmental quality according to the calculation results of the marine environmental quality calculation model; the present invention constructs a virtual marine environment by using a digital twin algorithm, which can dynamically reflect the changes in the actual marine environment in real time and provide a more comprehensive environmental quality assessment; by introducing dynamic convolution and self-attention mechanisms, the improved convolutional neural network can dynamically adjust the convolution kernel weights and capture the dependency between different positions in the input data, thereby improving the feature extraction capability of the neural network and enhancing its processing effect on complex marine environmental data.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of environmental quality assessment, and in particular to a method and system for assessing marine environmental quality. Background Art

[0002] Marine environmental quality assessment refers to the systematic evaluation of the physical, chemical, biological and ecological characteristics of the marine environment through the comprehensive use of a variety of technical means. Although the existing marine environmental quality assessment technology has made significant progress in practical application, it still faces the following major problems:

[0003] 1. Existing monitoring technologies usually rely on point sampling, with limited coverage, and are unable to fully reflect the overall status of a large marine environment, especially in deep sea and remote waters, where data acquisition is more difficult;

[0004] 2. Existing technologies mainly focus on monitoring a single indicator or a few indicators, lacking a comprehensive evaluation of the overall health and function of the marine ecosystem. This method is difficult to accurately reflect the complexity and diversity of the marine ecosystem;

[0005] 3. The existing monitoring technology and equipment are relatively expensive, especially when long-term, continuous monitoring is required, which requires large economic costs and human resources, which limits the widespread application of marine environmental quality assessment on a larger scale. Summary of the invention

[0006] In view of the above-mentioned problems, the present invention is proposed.

[0007] To solve the above technical problems, the present invention provides the following technical solutions: a method for evaluating marine environmental quality, comprising the following steps:

[0008] Acquire marine environment data, perform data correction on the acquired data, and construct the corrected data into a marine environment data set, including using sensors to acquire image data and marine water quality parameters in the marine environment data, and perform data correction and feature extraction;

[0009] Use the data set to build a marine environmental quality calculation model, specifically: build a marine environmental quality calculation model based on different characteristics in the marine environmental data, including the calculation of the normalized water body index for the relevant characteristics of the image data and the calculation of the comprehensive index for the relevant characteristics of the marine water quality parameters. Finally, use the digital twin algorithm to build the marine environmental quality calculation model;

[0010] According to the calculation results of the marine environment quality calculation model, the analysis and evaluation of the marine environment quality is realized, including using the calculation results of the marine environment quality calculation model and building an intelligent analysis model based on a convolutional neural network, while introducing dynamic convolution and self-attention mechanisms, specifically:

[0011] Dynamic convolution kernel weights are generated according to the input data, and a preliminary feature map of the ocean environment is obtained through convolution operation. The generated preliminary feature map of the ocean environment is processed using the self-attention mechanism, and a pooling operation is performed on the final feature map of the ocean environment processed by the self-attention mechanism to generate high-level features. Finally, the ocean water quality data is analyzed based on the generated high-level ocean environment feature map.

[0012] As a preferred solution of the marine environment quality assessment method described in the present invention, the feature extraction is specifically as follows:

[0013] When extracting features from image data, first read the RGB image data and convert the RGB image into a grayscale image using the weighted average method to obtain the red channel pixel value of the image data. , green channel pixel value And the blue channel pixel value , and then use Gaussian filtering technology to reduce the noise of grayscale image data to improve the accuracy of the data. At the same time, the Sobel operator is used to calculate the gradient of the filtered image. The specific implementation formula is as follows:

[0014]

[0015]

[0016]

[0017] ,

[0018] in, represents a Gaussian filter, represents the standard deviation of the image data, , Represent the gradients in the horizontal and vertical directions respectively. represents the gradient direction, Represents the gradient amplitude. According to the extracted gradient amplitude and direction, the data features can be extracted.

[0019] When extracting features of ocean water quality parameters, the data parameters collected by the sensor are combined into a set of ocean water quality parameter features according to the time series. ,and It represents the characteristic parameters obtained by different sensors, including water turbidity, dissolved oxygen content, chemical oxygen demand, seawater temperature, seawater salinity, seawater suspended matter concentration and seawater phytoplankton distribution.

[0020] As a preferred solution of the marine environment quality assessment method described in the present invention, the normalized water index calculation for the image data related features is specifically as follows:

[0021]

[0022]

[0023]

[0024] in, , Respectively represent the maximum and minimum values ​​of the red channel pixel value in the image data, , Respectively represent the maximum and minimum values ​​of the green channel pixel value in the image data, , Respectively represent the normalized reflectance of the red light band and the green light band, represents the reflectivity in the near-infrared band, Represents the calculated normalized water index.

[0025] As a preferred solution of the marine environment quality evaluation method described in the present invention, the construction of the marine environment quality calculation model using the digital twin algorithm is specifically as follows:

[0026] Based on the ocean water quality parameters, a physical simulation model is built.

[0027]

[0028] in, represents the composite index calculated under different time series, express Virtual ocean environment at all times;

[0029] Constructing the relationship model between the normalized water index and the comprehensive index, we have:

[0030]

[0031] in, , Respectively represent the coefficients of the constructed relationship model, Represents the error term. For the coefficients and error terms of the relationship model, by substituting the normalized water index and comprehensive index under the same time series into the relationship model, it can be obtained that: express Normalized water index calculated at all times;

[0032] The converted virtual ocean state and virtual ocean environment are dynamically combined to complete the construction of the ocean environment quality calculation model, then,

[0033]

[0034] in, express Virtual ocean environment at all times, , Respectively represent the coefficients of the constructed relationship model, Represents the error term. For the coefficients and error terms of the relationship model, by substituting the normalized water index and comprehensive index under the same time series into the relationship model, it can be obtained that: express The normalized water index calculated at each moment, The marine environment quality calculation model is The calculation result of time, , It represents the impact weight of the comprehensive index and normalized water index on the marine environmental quality.

[0035] As a preferred solution of the marine environment quality assessment method described in the present invention, the preliminary characteristic map of the marine environment obtained by the convolution operation is specifically as follows:

[0036] Generate dynamic convolution kernel weights based on input data,

[0037] Dynamic convolution operation,

[0038] in, Represents the dynamically generated convolution kernel weights, which are dynamically adjusted according to the input data. represents the convolutional layer bias vector, Represents the generated preliminary characteristic map of the ocean environment.

[0039] As a preferred solution of the marine environment quality assessment method described in the present invention, the processing of the generated preliminary feature map of the marine environment using the self-attention mechanism is specifically as follows:

[0040] The self-attention mechanism is used to calculate the query, key, and value from the preliminary feature map of the ocean environment after the convolution operation. , ,

[0041] Based on the calculated query and key, the similarity between the two is calculated, and then the attention weight is calculated.

[0042] The values ​​are weighted averaged according to the calculated attention weights,

[0043] in, , , denote the query, key, and value computed from the preliminary feature graph, respectively. , , Represent the weight matrices of query, key, and value in the self-attention mechanism, respectively. represents the calculated attention weight, Represents the final characteristic map of the ocean environment, represents a weighted frequency sequence, Represents the total number of weighted times.

[0044] As a preferred solution of the marine environment quality assessment method described in the present invention, the pooling operation of the final feature map of the marine environment processed by the self-attention mechanism is specifically as follows:

[0045]

[0046] The final feature map of the pooled ocean environment is passed through the fully connected layer for feature fusion and nonlinear transformation to generate high-level features.

[0047]

[0048] in, represents the ocean environment feature map after the pooling layer reduces the size. Represents a high-level ocean environment feature map after nonlinear transformation, represents the weight matrix of the fully connected layer, Represents the bias vector of the fully connected layer.

[0049] As a preferred solution of the marine environment quality evaluation system described in the present invention, it includes: a marine environment data set construction module, a marine environment quality calculation model construction module, and a marine environment quality evaluation module. The marine environment data set construction module is used to obtain marine environment data, and perform data correction on the acquired data, and construct the corrected data into a marine environment data set; the marine environment quality calculation model construction module builds a marine environment quality calculation model based on different features in the marine environment data set; the marine environment quality evaluation module is used to use an intelligent analysis model to realize analysis and evaluation of marine environment quality.

[0050] The beneficial effects of the present invention are as follows: by introducing the normalized water index, the present invention can realize the monitoring of a large range of sea areas and avoid the limitations of traditional point sampling; by using the digital twin algorithm to construct a virtual ocean environment, it can dynamically reflect the changes in the actual ocean environment in real time and provide a more comprehensive environmental quality assessment; by introducing dynamic convolution and self-attention mechanisms, the traditional convolutional neural network is improved, so that the improved convolutional neural network can dynamically adjust the convolution kernel weights and capture the dependencies between different positions in the input data, thereby improving the feature extraction capability of the neural network and enhancing its processing effect on complex ocean environment data; by constructing an intelligent analysis model, a comprehensive analysis of the ocean environment can be performed based on different parameters of ocean environment quality; through the application of drone remote sensing and sensors, real-time monitoring of ocean environment data can be achieved, reducing human resource investment. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. Among them:

[0052] Figure 1 This is a schematic diagram of the overall method steps of a marine environment quality assessment method of the present invention.

[0053] Figure 2 The figure is a schematic diagram of the overall structure of a marine environment quality evaluation system of the present invention. DETAILED DESCRIPTION

[0054] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0055] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0056] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0057] The present invention is described in detail with reference to schematic diagrams. When describing the embodiments of the present invention, for the sake of convenience, the cross-sectional diagrams showing the device structure will not be partially enlarged according to the general scale, and the schematic diagrams are only examples, which should not limit the scope of protection of the present invention. In addition, in actual production, the three-dimensional dimensions of length, width and depth should be included.

[0058] Meanwhile, in the description of the present invention, it should be noted that the terms "first, second or third" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance.

[0059] In the present invention, unless otherwise clearly specified and limited, the terms "install, connect, connect" should be understood in a broad sense, for example: it can be a fixed connection, a detachable connection or an integral connection; it can also be a mechanical connection, an electrical connection or a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0060] Example 1

[0061] Reference Figure 1 , as the first embodiment of the present invention, provides a method for evaluating marine environmental quality, comprising the following steps,

[0062] S1: Acquire marine environment data, perform data correction on the acquired data, and construct the corrected data into a marine environment data set.

[0063] Specifically, the acquisition of marine environmental data is to use sensors to collect environmental data of the ocean in the monitoring area. The sensors include drone remote sensing and ocean buoys. The drone remote sensing is to use drones to collect ocean image data in the monitoring area. The ocean buoys are deployed in the monitoring area, and multi-parameter sensors are built in the ocean buoys to monitor the ocean water quality parameters in real time. At the same time, a sensor network is deployed in the monitoring area, and multiple sensors are combined to monitor the marine environment.

[0064] Furthermore, the data correction performed on the acquired data includes correction of image data and correction of ocean water quality parameters. The correction of the image data is performed through the steps of radiation correction, geometric correction and image stitching. Radiation correction, geometric correction and image stitching are existing well-known technologies that can be queried by personnel in this field and are not described in this embodiment. The correction of the ocean water quality parameters is performed through the steps of data cleaning, data deduplication and outlier processing. Data cleaning, data deduplication and outlier processing are existing well-known technologies that can be queried by personnel in this field and are not described in this embodiment.

[0065] Furthermore, the corrected data is constructed into a marine environment data set by converting the corrected image data into and ocean water quality parameters Perform feature extraction operations, and use the extracted features to construct a marine environment data set. The feature extraction is specifically as follows:

[0066] When extracting features from image data, first read the RGB image data and convert the RGB image into a grayscale image using the weighted average method to obtain the red channel pixel value of the image data. , green channel pixel value And the blue channel pixel value , and then use Gaussian filtering technology to reduce the noise of grayscale image data to improve the accuracy of the data. At the same time, the Sobel operator is used to calculate the gradient of the filtered image. The specific implementation formula is as follows:

[0067]

[0068]

[0069]

[0070] ,

[0071] in, represents a Gaussian filter, represents the standard deviation of the image data, , Represent the gradients in the horizontal and vertical directions respectively. represents the gradient direction, Represents the gradient amplitude. According to the extracted gradient amplitude and direction, the data features can be extracted.

[0072] When extracting features of ocean water quality parameters, the data parameters collected by the sensor are combined into a set of ocean water quality parameter features according to the time series. ,and It represents the characteristic parameters obtained by different sensors, including water turbidity, dissolved oxygen content, chemical oxygen demand, seawater temperature, seawater salinity, seawater suspended matter concentration, and seawater phytoplankton distribution.

[0073] It should be noted that the marine environment data set is a data set for calculating the marine environment quality by secondary fusion of the extracted features. , then there is, ,and Represents the feature index, which is the total number of features in the data set.

[0074] S2: Use the data set to build a marine environmental quality calculation model.

[0075] Specifically, the use of data sets to build a marine environment quality calculation model is to use the characteristic values ​​in the marine environment data set to build a marine environment quality calculation model, realize automatic detection of marine environment data, and mark and remind abnormal values ​​at the same time.

[0076] Furthermore, the construction of the marine environment quality calculation model is based on different characteristic values ​​in the marine environment set, as follows:

[0077] For the image data related features in the data set, the normalized water index is used to build a marine environmental quality calculation model, as follows:

[0078] The normalized water index is based on the red channel pixel value of the image data. , green channel pixel value Calculated, then,

[0079]

[0080]

[0081]

[0082] in, , Respectively represent the maximum and minimum values ​​of the red channel pixel value in the image data, , Respectively represent the maximum and minimum values ​​of the green channel pixel value in the image data, , Respectively represent the normalized reflectance of the red light band and the green light band, represents the reflectivity in the near-infrared band, represents the calculated normalized water index;

[0083] For the relevant characteristics of ocean water quality parameters in the data set, a comprehensive index is used to build a marine environmental quality calculation model, as follows:

[0084] For the parameters collected by different sensors, the comprehensive index is calculated, then:

[0085]

[0086] in, represents the calculated composite index, Indicates The weight of each characteristic parameter is set by the implementer according to the application scenario based on the degree of impact on ocean water quality in historical data. Indicates feature parameters.

[0087] Furthermore, the marine environment quality calculation model uses a digital twin algorithm to construct a virtual marine environment based on seawater quality parameters, and maps the calculated normalized water index to the constructed virtual marine environment, thereby realizing automatic detection of marine environmental data, and marking and reminding abnormal values;

[0088] Specifically, the virtual ocean environment uses a digital twin algorithm and builds a physical simulation model based on ocean water quality parameters.

[0089]

[0090] in, represents the composite index calculated under different time series, express Virtual ocean environment at all times;

[0091] Mapping the calculated normalized water index to the constructed virtual ocean environment is to convert the normalized water index into a virtual ocean state, and converting the normalized water index into a virtual ocean state is to construct a relationship model between the normalized water index and the comprehensive index, then,

[0092]

[0093] in, , Respectively represent the coefficients of the constructed relationship model, Represents the error term. For the coefficients and error terms of the relationship model, by substituting the normalized water index and comprehensive index under the same time series into the relationship model, it can be obtained that: express Normalized water index calculated at all times;

[0094] The ocean environment quality calculation model is to dynamically combine the converted virtual ocean state and the virtual ocean environment to complete the construction of the ocean environment quality calculation model, then,

[0095]

[0096] in, express Virtual ocean environment at all times, , Respectively represent the coefficients of the constructed relationship model, Represents the error term. For the coefficients and error terms of the relationship model, by substituting the normalized water index and comprehensive index under the same time series into the relationship model, it can be obtained that: express The normalized water index calculated at each moment, The marine environment quality calculation model is The calculation result of time, , The weights of the comprehensive index and the normalized water index on the marine environmental quality are set by the implementers based on the calculation results of the two, which will not be elaborated here.

[0097] S3: Based on the calculation results of the marine environment quality calculation model, the analysis and evaluation of the marine environment quality is realized.

[0098] Specifically, the analysis and evaluation of the marine environment quality is achieved by utilizing the calculation results of the marine environment quality calculation model to construct an intelligent analysis model, thereby achieving the analysis and evaluation of the marine environment quality. The intelligent analysis model uses the calculation results of the marine environment quality calculation model as the input of the model, and uses the historical data of the marine environment as the training set of the intelligent analysis model. The training set is used to optimize the training of the intelligent analysis model to improve the accuracy of the intelligent analysis model.

[0099] Furthermore, the intelligent analysis model is built based on a convolutional neural network, and combined with dynamic convolution and self-attention mechanisms, the calculation results of the marine environment quality calculation model are used as the input of the intelligent analysis model, and the marine environment historical data are used as the training set of the intelligent analysis model. The intelligent analysis model is optimized and trained, thereby realizing the analysis and evaluation of the marine environment quality, as follows:

[0100] Constructing an input data set and an input training set of an intelligent analysis model, wherein the input data set is composed of calculation results of a marine environment quality calculation model, including calculation results based on current seawater quality parameters, calculation results based on current marine sediment parameters, calculation results based on current marine biological quality, and calculation results of current marine image data; the input training set is composed of historical data of the marine environment, including historical data with different time nodes but the same as current seawater quality parameters, historical data with different time nodes but the same as current marine sediment parameters, historical data with different time nodes but the same as current marine biological quality, and historical data with different time nodes but the same as current marine image data;

[0101] For the current ocean environment data input, the dynamic convolution kernel weights are generated using the input data, and then the dynamically generated convolution kernel is applied for convolution operation to obtain the preliminary feature map of the ocean environment. Then,

[0102] Generate dynamic convolution kernel weights based on input data,

[0103] Dynamic convolution operation,

[0104] in, Represents dynamically generated weights, which are dynamically adjusted based on input data. represents the convolutional layer bias vector, Represents the generated preliminary characteristic map of the ocean environment;

[0105] After the convolution operation, the self-attention mechanism is used to process the generated preliminary feature map of the ocean environment, and then,

[0106] The self-attention mechanism is used to calculate the query, key, and value from the preliminary feature map of the ocean environment after the convolution operation. , ,

[0107] Based on the calculated query and key, the similarity between the two is calculated, and then the attention weight is calculated.

[0108] The values ​​are weighted averaged according to the calculated attention weights,

[0109] in, , , denote the query, key, and value computed from the preliminary feature graph, respectively. , , Represent the weight matrices of query, key, and value in the self-attention mechanism, respectively. represents the calculated attention weight, Represents the final characteristic map of the ocean environment, represents a weighted frequency sequence, represents the total number of weighted times;

[0110] The final feature map of the ocean environment processed by the self-attention mechanism is pooled to reduce the size of the final feature map of the ocean environment while retaining the information. Then,

[0111]

[0112] The final feature map of the pooled ocean environment is passed through the fully connected layer for feature fusion and nonlinear transformation to generate high-level features.

[0113]

[0114] in, represents the ocean environment feature map after the pooling layer reduces the size. Represents a high-level ocean environment feature map after nonlinear transformation, represents the weight matrix of the fully connected layer, Represents the bias vector of the fully connected layer;

[0115] The current model is trained using the training set of the intelligent analysis model, and the model is iteratively optimized based on the training results until the training results meet the actual needs of the implementer.

[0116] Furthermore, the analysis and evaluation of the marine environmental quality is to analyze the marine water quality data according to the generated high-level marine environmental characteristic map, as follows:

[0117] Before conducting an analysis of marine environmental quality, use historical marine environmental data to set environmental standard thresholds ;

[0118] Using the generated high-level marine environmental characteristic map and environmental standard threshold Compare and judge whether the current water quality meets the expected standards, and conduct data statistics and evaluation;

[0119] Using the marine sediment-related features in the generated high-level marine environmental characteristic map, the environmental data of marine sediments are analyzed to assess the pollution level and change trend of sediments;

[0120] Using the marine life-related features in the generated high-level marine environmental characteristics map, the quality data of marine organisms are analyzed to assess the health status of marine life and the deterioration of biodiversity;

[0121] By using the marine resource-related characteristics in the generated high-level marine environmental characteristic map, the changes in marine ecological resources are analyzed and the health status of the ecosystem and resource utilization are evaluated.

[0122] Example 2

[0123] Reference Figure 2 , which is a second embodiment of the present invention, provides a marine environment quality evaluation system, including a marine environment data set building module, a marine environment quality calculation model building module, and a marine environment quality evaluation module.

[0124] Specifically, the marine environment data set construction module is used to acquire marine environment data, perform data correction on the acquired data, and construct the corrected data into a marine environment data set; the marine environment quality calculation model construction module is used to build a marine environment quality calculation model based on different features in the marine environment data set; the marine environment quality evaluation module is used to use an intelligent analysis model to realize analysis and evaluation of marine environment quality.

[0125] Furthermore, the marine environment data set construction module is used to acquire and process data that will serve as basic input data for building a marine environment quality calculation model, providing reliable data support for subsequent quality calculations and evaluations. It includes a data acquisition unit, a data correction unit, and a data storage unit. The data acquisition unit uses drones, satellite remote sensing equipment, and ocean buoys to acquire image data and water quality parameters of the marine environment. The data correction unit uses grayscale processing, noise elimination, and data calibration technology to correct the acquired data to ensure data accuracy. The data storage unit constructs the corrected data into a data set according to different characteristics, stores it in a database, and provides basic data for subsequent analysis.

[0126] The marine environment quality calculation model building module uses the data set constructed in the previous module to generate a marine environment quality calculation model to provide basic calculation results for subsequent environmental quality evaluation. Specifically, key features are extracted from the marine environment data set, including NDWI, water quality parameters, etc., and the marine environment quality calculation model is constructed by using the digital twin algorithm and combining the extracted features. The calculation model is trained and optimized based on historical data and real-time data to ensure the accuracy and reliability of the model.

[0127] The marine environment quality evaluation module uses the results provided by the calculation model and combines historical data for comprehensive analysis and evaluation, builds an intelligent analysis model based on a convolutional neural network, and uses the calculation results and historical data provided by the calculation model for comprehensive analysis. Finally, based on the analysis results, the marine environment quality is evaluated and early warning information is generated.

[0128] Furthermore, if the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0129] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0130] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering or, if necessary, processing in another suitable manner, and then stored in a computer memory.

[0131] Additionally, in order to provide a concise description of exemplary embodiments, all features of an actual embodiment may not be described (ie, those features that are not relevant to the best mode presently contemplated for carrying out the invention or those features that are not relevant to implementing the invention).

[0132] It will be appreciated that in the development of any actual implementation, as in any engineering or design project, numerous implementation-specific decisions may be made. Such a development effort may be complex and time-consuming, but will be a routine task of design, fabrication, and production for those of ordinary skill having the benefit of this disclosure without undue experimentation.

[0133] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for evaluating marine environmental quality, characterized in that: The following steps are included: Acquire marine environment data, perform data correction on the acquired data, and construct the corrected data into a marine environment data set, including using sensors to acquire image data and marine water quality parameters in the marine environment data, and perform data correction and feature extraction; The feature extraction is specifically as follows: When extracting features from image data, first read the RGB image data and convert the RGB image into a grayscale image using the weighted average method to obtain the red channel pixel value of the image data. , green channel pixel value And the blue channel pixel value , and then use Gaussian filtering technology to reduce the noise of grayscale image data to improve the accuracy of the data. At the same time, the Sobel operator is used to calculate the gradient of the filtered image. The specific implementation formula is as follows: , in, represents a Gaussian filter, represents the standard deviation of the image data, , Represent the gradients in the horizontal and vertical directions respectively. represents the gradient direction, Represents the gradient amplitude. According to the extracted gradient amplitude and direction, the data features can be extracted. When extracting features of ocean water quality parameters, the data parameters collected by the sensor are combined into a set of ocean water quality parameter features according to the time series. ,and Indicates the characteristic parameters obtained by different sensors, including water turbidity, dissolved oxygen content, chemical oxygen demand, seawater temperature, seawater salinity, seawater suspended matter concentration, and seawater phytoplankton distribution; Use the data set to build a marine environmental quality calculation model, specifically: build a marine environmental quality calculation model based on different characteristics in the marine environmental data, including the calculation of the normalized water body index for the relevant characteristics of the image data and the calculation of the comprehensive index for the relevant characteristics of the marine water quality parameters. Finally, use the digital twin algorithm to build the marine environmental quality calculation model; The calculation of the normalized water index for the image data related features is specifically as follows: in, , Respectively represent the maximum and minimum values ​​of the red channel pixel value in the image data, , Respectively represent the maximum and minimum values ​​of the green channel pixel value in the image data, , Respectively represent the normalized reflectance of the red light band and the green light band, represents the reflectivity in the near-infrared band, represents the calculated normalized water index; For the relevant characteristics of ocean water quality parameters in the data set, a comprehensive index is used to build a marine environmental quality calculation model, as follows: For the parameters collected by different sensors, the comprehensive index is calculated, then: in, represents the calculated composite index, Indicates The weight of each characteristic parameter is set by the implementer according to the application scenario based on the degree of impact on ocean water quality in historical data. Indicates characteristic parameters; The construction of the marine environmental quality calculation model using the digital twin algorithm is as follows: Based on the ocean water quality parameters, a physical simulation model is built. in, represents the composite index calculated under different time series, express Virtual ocean environment at all times; Constructing the relationship model between the normalized water index and the comprehensive index, we have: in, , Respectively represent the coefficients of the constructed relationship model, Represents the error term. For the coefficients and error terms of the relationship model, by substituting the normalized water index and comprehensive index under the same time series into the relationship model, it can be obtained that: express Normalized water index calculated at all times; The converted virtual ocean state and virtual ocean environment are dynamically combined to complete the construction of the ocean environment quality calculation model, then, in, express Virtual ocean environment at all times, , Respectively represent the coefficients of the constructed relationship model, Represents the error term. For the coefficients and error terms of the relationship model, by substituting the normalized water index and comprehensive index under the same time series into the relationship model, it can be obtained that: express The normalized water index calculated at each moment, The marine environmental quality calculation model is The calculation result of time, , Indicates the impact weight of the comprehensive index and the normalized water index on the quality of the marine environment; According to the calculation results of the marine environment quality calculation model, the analysis and evaluation of the marine environment quality is realized, including using the calculation results of the marine environment quality calculation model and building an intelligent analysis model based on a convolutional neural network, while introducing dynamic convolution and self-attention mechanisms, specifically: Dynamic convolution kernel weights are generated according to the input data, and a preliminary feature map of the ocean environment is obtained through convolution operation. The generated preliminary feature map of the ocean environment is processed using the self-attention mechanism, and a pooling operation is performed on the final feature map of the ocean environment processed by the self-attention mechanism to generate high-level features. Finally, the ocean water quality data is analyzed based on the generated high-level ocean environment feature map.

2. A method for evaluating marine environmental quality according to claim 1, characterized in that: The preliminary characteristic map of the ocean environment obtained by the convolution operation is as follows: Generate dynamic convolution kernel weights based on input data, Dynamic convolution operation, in, Represents the dynamically generated convolution kernel weights, which are dynamically adjusted according to the input data. represents the convolutional layer bias vector, Represents the generated preliminary characteristic map of the ocean environment.

3. A method for evaluating marine environmental quality as claimed in claim 2, characterized in that: The self-attention mechanism is used to process the generated preliminary feature map of the ocean environment as follows: The self-attention mechanism is used to calculate the query, key, and value from the preliminary feature map of the ocean environment after the convolution operation. , , Based on the calculated query and key, the similarity between the two is calculated, and then the attention weight is calculated. The values ​​are weighted averaged according to the calculated attention weights, in, , , denote the query, key, and value computed from the preliminary feature graph, respectively. , , Represent the weight matrices of query, key, and value in the self-attention mechanism, respectively. represents the calculated attention weight, Represents the final characteristic map of the ocean environment, represents a weighted frequency sequence, Represents the total number of weighted times.

4. A method for evaluating marine environmental quality as claimed in claim 3, characterized in that: The pooling operation of the final feature map of the ocean environment processed by the self-attention mechanism is as follows: The final feature map of the pooled ocean environment is passed through the fully connected layer for feature fusion and nonlinear transformation to generate high-level features. in, represents the ocean environment feature map after the pooling layer reduces the size. Represents a high-level ocean environment feature map after nonlinear transformation, represents the weight matrix of the fully connected layer, Represents the bias vector of the fully connected layer.

5. A system using the marine environment quality assessment method according to any one of claims 1 to 4, characterized in that: It includes: marine environment data set construction module, marine environment quality calculation model construction module, marine environment quality evaluation module, The marine environment data set construction module is used to obtain marine environment data, perform data correction on the obtained data, and construct the corrected data into a marine environment data set; The marine environment quality calculation model building module builds a marine environment quality calculation model based on different features in the marine environment data set; The marine environment quality evaluation module is used to analyze and evaluate the marine environment quality using an intelligent analysis model.