Water vapor networking observation method and system based on GNSS

Abstract

The invention provides a GNSS-based water vapor networking observation method and system, and relates to the technical field of meteorological observation, and the method comprises the steps: obtaining first information and second information; obtaining a first feature vector corresponding to each GNSS receiver; obtaining a second feature vector; determining an association degree value of the first feature vector of each GNSS receiver and a second feature vector of the atmospheric water vapor information of the target area; and obtaining atmospheric water vapor information in a future preset time period. According to the invention, effective feature extraction and fusion are carried out on the satellite data collected by the GNSS receiver in the water vapor network and the atmospheric water vapor information, so that the observation of the atmospheric water vapor is more accurate and real-time. The atmospheric influence factors are extracted through principal component analysis, and the feature fusion model is combined, so that the capability of observing water vapor information under the complex atmospheric condition is enhanced, the regional atmospheric water vapor change can be observed in real time, and compared with a traditional meteorological observation station, the observation method provided by the invention has remarkable advantages in spatial distribution and time precision.

Description

Technical Field

[0001] The present invention relates to the technical field of meteorological observation, and more particularly, to a GNSS-based water vapor networking observation method and system. Background Art

[0002] With the rapid development of Global Navigation Satellite System (GNSS) technology, GNSS-based atmospheric water vapor monitoring methods have gradually emerged. When GNSS signals pass through the atmosphere, they are affected by atmospheric water vapor and delayed. By accurately measuring and analyzing this delay, the atmospheric water vapor content can be retrieved. However, most of the existing GNSS-based water vapor monitoring is based on observations at a single station, which can only obtain local water vapor information at the observation point and cannot grasp the distribution and variation law of atmospheric water vapor in the region as a whole.

[0003] To achieve comprehensive, accurate, and real-time monitoring of atmospheric water vapor, it is of great practical significance and urgent need to construct a GNSS-based water vapor networking observation system and method. By forming a network of multiple GNSS receivers, collaborating in observations, and fusing and processing data, it is expected to overcome the limitations of traditional observation methods and single-station GNSS observations and provide richer and more accurate atmospheric water vapor information for atmospheric science research and related application fields. Summary of the Invention

[0004] The purpose of the present invention is to provide a GNSS-based water vapor networking observation method and system to improve the above problems. To achieve the above purpose, the technical solutions adopted by the present invention are as follows:

[0005] In a first aspect, the present application provides a GNSS-based water vapor networking observation method, including:

[0006] Obtaining first information and second information, where the first information is satellite data collected by the water vapor network in the target area and its corresponding time series, and the second information is the atmospheric water vapor data of the target area, and the water vapor network includes multiple GNSS receivers;

[0007] Performing principal component analysis on the first information to extract atmospheric influence factors, and inputting the atmospheric influence factors into a preset feature transformation model for feature transformation to obtain a first feature vector corresponding to each GNSS receiver;

[0008] Inputting the second information into a preset feature fusion model for feature fusion to obtain a second feature vector;

[0009] Performing correlation degree analysis based on the first feature vector and the second feature vector to determine the correlation degree value between the first feature vector of each GNSS receiver and the second feature vector of the atmospheric water vapor information in the target area;

[0010] Send the first eigenvector, the second eigenvector, and the correlation degree value of each GNSS receiver in the water vapor group to a preset water vapor observation model for observation to obtain the atmospheric water vapor information of the target area in a preset time period in the future.

[0011] In a second aspect, the present application also provides a GNSS-based water vapor networking observation system, including:

[0012] An acquisition unit, configured to acquire first information and second information, where the first information is satellite data collected by a water vapor network in a target area and its corresponding time series, and the second information is atmospheric water vapor data of the target area, and the water vapor network includes multiple GNSS receivers;

[0013] An extraction unit, configured to perform principal component analysis on the first information to extract atmospheric influence factors, and input the atmospheric influence factors into a preset feature transformation model for feature transformation to obtain a first eigenvector corresponding to each GNSS receiver;

[0014] A fusion unit, configured to input the second information into a preset feature fusion model for feature fusion to obtain a second eigenvector;

[0015] A first analysis unit, configured to perform correlation degree analysis based on the first eigenvector and the second eigenvector to determine the correlation degree value between the first eigenvector of each GNSS receiver and the second eigenvector of the atmospheric water vapor information of the target area;

[0016] An observation unit, configured to send the first eigenvector, the second eigenvector, and the correlation degree value of each GNSS receiver in the water vapor network to a preset water vapor observation model for observation to obtain the atmospheric water vapor information of the target area in a preset time period in the future.

[0017] The beneficial effects of the present invention are as follows:

[0018] By effectively extracting and fusing the satellite data and atmospheric water vapor information collected by the GNSS receivers in the water vapor network, the present invention makes the observation of atmospheric water vapor more accurate and real-time. By extracting atmospheric influence factors through principal component analysis and combining with a feature fusion model, it can better process multi-source data, enhance the observation ability of water vapor information under complex atmospheric conditions, and can observe the changes of regional atmospheric water vapor in real time. The observation method of the present application has significant advantages in both spatial distribution and time accuracy compared with traditional meteorological observation stations.

[0019] Other features and advantages of the present invention will be described in the subsequent description, and some will become obvious from the description, or can be understood by implementing the embodiments of the present invention. Description of the Drawings

[0020] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention, and thus should not be regarded as limiting the scope. For those of ordinary skill in the art, without creative efforts, other related drawings can also be obtained based on these drawings.

[0021] Figure 1 It is a schematic flow diagram of the GNSS-based water vapor networking observation method described in the embodiments of the present invention;

[0022] Figure 2 It is a schematic structural diagram of the GNSS-based water vapor networking observation system described in the embodiments of the present invention.

[0023] Markings in the figure: 10, acquisition unit; 20, extraction unit; 30, fusion unit; 40, first analysis unit; 50, observation unit. Specific embodiments

[0024] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. Usually, the components of the embodiments of the present invention described and shown in the accompanying drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed present invention, but merely represents the selected embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts fall within the scope of protection of the present invention.

[0025] It should be noted that: similar reference numerals and letters denote similar items in the following drawings. Therefore, once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings. At the same time, in the description of the present invention, terms such as "first" and "second" are only used for distinguishing descriptions and cannot be understood as indicating or implying relative importance.

[0026] Embodiment 1:

[0027] This embodiment provides a GNSS-based water vapor networking observation method.

[0028] See Figure 1 , the figure shows that this method includes steps S10, S20, S30, S40, and S50.

[0029] Step S10. Obtain the first information and the second information. The first information is the satellite data collected by the water vapor network in the target area and its corresponding time series, and the second information is the atmospheric water vapor data of the target area. The water vapor network includes multiple GNSS receivers;

[0030] Specifically, in this application, the water vapor network is composed of multiple GNSS receivers deployed at multiple ground stations. In a relatively large area, the atmospheric water vapor content usually varies. Therefore, the spatial distribution of GNSS receivers will directly affect the estimation accuracy of atmospheric water vapor information, and the relative positions between GNSS receivers will also affect their observation effects. A denser GNSS receiver layout can capture the spatial variation of atmospheric water vapor more accurately, while a sparser GNSS receiver layout may lead to the lack or inaccuracy of atmospheric water vapor information in some areas. In order to better observe the atmospheric water vapor information in the target area, it is necessary to comprehensively consider the spatial distribution characteristics of each GNSS receiver in the water vapor network in the target area and their different contributions to the atmospheric water vapor information. Therefore, for the observation of the atmospheric water vapor information in the target area, this application considers the acquisition information of all GNSS receivers in the water vapor network in the target area to ensure that the data of each receiver can be fully utilized, thereby improving the accuracy and comprehensiveness of the observation.

[0031] Among them, the first information is the satellite signal data obtained by all GNSS receivers included in the water vapor network in the target area and the time series information derived from these data. These data are helpful for analyzing the propagation delay of GNSS signals and atmospheric effects. The second information refers to the specific atmospheric water vapor information of the target area, which is usually used to reflect the meteorological conditions in the target area, especially the water vapor concentration. The combination of the two provides basic data for subsequent atmospheric water vapor observation and analysis.

[0032] Step S20. Perform principal component analysis on the first information to extract the atmospheric influence factors, and input the atmospheric influence factors into a preset feature transformation model for feature transformation to obtain the first feature vector corresponding to each GNSS receiver;

[0033] Specifically, the satellite data collected by each GNSS receiver in the water vapor network contains a lot of data. Among them, some data can effectively reflect the water vapor information in the current atmosphere. Therefore, the most representative influence factors, especially the factors related to water vapor, can be extracted from these multi-dimensional data, and the extracted atmospheric influence factors are input into a preset feature transformation model to perform steps such as data standardization and dimensionality reduction to improve the usability and accuracy of the data.

[0034] Specifically, step S20 specifically includes steps S21 to S25:

[0035] Step S21. Perform principal component analysis on the satellite data in the first information. By calculating the covariance matrix of each feature in the satellite data and performing eigenvalue decomposition, a simplified set of principal component features is obtained;

[0036] Specifically, by performing principal component analysis on the satellite data in the first information, calculating the covariance matrix of the features and performing eigenvalue decomposition, the main influencing factors related to the atmosphere can be effectively extracted, the remaining irrelevant factors can be removed, and redundant and noise information can be reduced.

[0037] S22. Based on the distance differences and time differences between the sample points in the principal component feature set, construct a k-nearest neighbor graph. The nodes in the k-nearest neighbor graph represent the sample points, and the weight of the edge represents the similarity degree value between the corresponding sample points;

[0038] Specifically, constructing a k-nearest neighbor graph based on the distance differences and time differences between the sample points in the principal component feature set can effectively reveal the similarity relationship between the sample points. The k-nearest neighbor graph can not only reflect the similarity in space of the sample points, but also combine the time dimension and consider the mutual influence of the sample points in the time series.

[0039] Specifically, step S22 specifically includes steps S221 to S223:

[0040] Step S221. Calculate the distance differences and time differences between the sample points in the principal component feature set and construct vectors to obtain a distance vector and a time vector respectively;

[0041] Specifically, step S221 specifically includes steps S2211 to S2218:

[0042] Step S2211. Calculate the Euclidean distance between the sample point and the other sample points to obtain a plurality of first distances;

[0043] Step S2212. Determine the values within a preset distance range from the plurality of first distances as the reference distance;

[0044] Step S2213. Calculate the difference between the first distance corresponding to the sample point and the reference distance to obtain a plurality of second distances;

[0045] Step S2214. Based on the second distances, form a distance vector;

[0046] Step S2215. Calculate the time differences between the sample point and the other sample points to obtain a plurality of first differences;

[0047] Step S2216. Determine the values within a preset time range from the plurality of first differences as the reference difference;

[0048] Step S2217. Calculate the difference between the first difference corresponding to the sample point and the reference difference to obtain multiple second differences;

[0049] Step S2218. Construct a time vector based on the second differences;

[0050] Specifically, by calculating the Euclidean distance and time difference between sample points, determining a reference value according to a preset range, and using the difference between the difference and the reference value as a vector, it can effectively improve the accuracy of similarity measurement, enhance the ability to capture spatio-temporal correlations, improve calculation efficiency, and enhance the robustness of the model, thereby providing more accurate and effective support in complex spatio-temporal data analysis tasks. In this embodiment, the preset distance range is [9, 13], and the preset time range is [2, 8]. The above values can be adjusted according to the actual situation and are not particularly limited here.

[0051] Step S222. Calculate the similarity degree value between two sample points based on the distance vectors and time vectors respectively corresponding to the two sample points;

[0052] Specifically, the calculation formula for the similarity degree value between two sample points includes:

[0053]

[0054] where S ab is the similarity degree value between the a-th sample point and the b-th sample point; is the distance vector corresponding to the a-th sample point; is the time vector corresponding to the a-th sample point; is the distance vector corresponding to the b-th sample point; is the time vector corresponding to the b-th sample point; a ≠ b, a, b ∈ N, and N is the number of sample points.

[0055] Step S223. Construct a k-nearest neighbor graph based on the similarity degree value, and the length of the edge in the k-nearest neighbor graph is a set threshold;

[0056] Specifically, in this step, by analyzing the similarity degree value of the distance and time factors of each sample point, and constructing a k-nearest neighbor graph, the system can more intuitively display the relationship between feature data, which is helpful for subsequent structural feature extraction.

[0057] Step S23. Perform a graph convolution operation based on the k-nearest neighbor graph, transfer the information of each sample point to adjacent sample points, and update the features by combining the information of adjacent sample points to obtain the updated sample point information;

[0058] Specifically, this step uses a structural feature extraction neural network algorithm to perform graph convolution operations. For the constructed k-nearest neighbor graph, this algorithm allows the information of each sample point to be passed to its adjacent sample points and updates the information by combining the information of adjacent sample points.

[0059] The sample point information update formula includes:

[0060]

[0061] Where s’ i is the updated sample point information of the i-th sample point; s i is the sample point information of the i-th sample point; s j is the sample point information of the j-th sample point; X ij is the correlation degree value between the i-th sample point and the j-th sample point; W is the weight matrix; δ is the activation function; N is the number of sample points.

[0062] Specifically, step S23 specifically includes steps S231 to S233:

[0063] Step S231. Use the similarity degree value matrix as the weight matrix and input the weight matrix and a preset activation function into the input layer of the graph convolutional neural network.

[0064] Specifically, using the similarity degree value matrix as the weight matrix means quantifying the similarity between sample points as the strength of the connections in the graph convolutional network. In the graph convolutional neural network, these similarity degree values determine the degree of information transfer between sample points, which is manifested as the edge weights in the graph. By inputting this weight matrix together with the preset activation function into the input layer of the graph convolutional neural network, the network can dynamically update the feature representation of each sample point according to the similarity between sample points. The activation function helps the model introduce non-linear relationships in this process, making the feature update more flexible, thereby enhancing the network's ability to express complex data patterns.

[0065] Step S232. Perform weighted summation on the information of each sample point and the information transferred from adjacent sample points, and perform a linear transformation through the weight matrix to obtain the linearly transformed feature data information.

[0066] Specifically, this step can effectively aggregate local neighborhood information, enabling each sample point to not only retain its own feature information but also update its own features by combining the information of its adjacent sample points.

[0067] Step S233. Perform non-linear mapping on the linearly transformed feature data information based on the activation function to obtain the updated sample point information.

[0068] Specifically, through the non-linear mapping of the activation function, the system can learn and capture more complex patterns and features in the data, effectively improving the expressive ability of the neural network.

[0069] Step S24. Perform a pooling operation on all the updated sample point information to obtain multiple overall sample information;

[0070] Step S25. Send the overall sample information after the pooling operation to the fully connected layer to obtain the first feature vector;

[0071] Specifically, performing a pooling operation on all the updated sample point information, by extracting the maximum value or average value of the features globally, reduces the dimension of the data and retains the important feature information, making the features more concise and more representative. The fully connected layer further processes the pooled features, through linear transformation and non-linear activation function, and finally generates the first feature vector of each GNSS receiver. The first feature vector represents a refined expression of all the sample information in the GNSS receiver and can be used as the input for subsequent tasks.

[0072] Step S30. Input the second information into a preset feature fusion model for feature fusion to obtain the second feature vector;

[0073] Specifically, inputting the second information into a preset feature fusion model for feature fusion can effectively integrate the atmospheric water vapor information from multiple sources or dimensions, thereby extracting the most representative features. Through feature fusion, the model can reduce redundancy and noise while retaining the key content in the information, improving the expressive ability of the data.

[0074] Specifically, step S30 specifically includes steps S31 to S33:

[0075] Step S31. Analyze and process the second information. Among them, divide the second information into multiple levels, and there is an inclusion relationship from top to bottom for each level. And based on the comparison of the importance of the data corresponding to two levels, construct a judgment matrix;

[0076] Step S32. Calculate the weights of each level in the judgment matrix based on the AHP algorithm to obtain the relative importance values of each data in the second information;

[0077] Step S33. Sort the relative importance values in ascending order, and perform feature extraction and fusion on the data whose sorting is before the set position to obtain the second feature vector;

[0078] Specifically, the second information is divided into multiple levels, and the weights of the data at each level are calculated based on the AHP algorithm, which can scientifically evaluate the relative importance of each data in the entire information system, so that reasonable sorting can be carried out according to the importance of the data. Subsequently, the data features with higher importance are extracted and fused in the set order to obtain a concise and discriminative second feature vector. This approach not only improves the efficiency of feature selection but also ensures the retention of the most critical data, helps to improve the accuracy and robustness of subsequent observations and analyses, and at the same time reduces the interference of redundant information.

[0079] Step S40. Perform correlation degree analysis based on the first feature vector and the second feature vector to determine the correlation degree value between the first feature vector of each GNSS receiver and the second feature vector of the atmospheric water vapor information in the target area;

[0080] Specifically, performing correlation degree analysis based on the first feature vector and the second feature vector can effectively evaluate the relationship between the geographical information in the area where the GNSS receiver is located and the atmospheric water vapor information.

[0081] Specifically, step S40 specifically includes steps S41 to S43:

[0082] Step S41. Perform dimensionless processing on the first feature vector and the second feature vector based on the mean transformation method to obtain the dimensionless first feature vector and the second feature vector respectively;

[0083] Specifically, the differences between all the first feature vectors and the corresponding second feature vectors in each water vapor network are eliminated through dimensionless processing. The dimensionless processing includes maximum-minimum normalization, standard deviation normalization, maximum absolute value normalization, etc., and there is no special limitation here.

[0084] Step S42. Calculate the correlation coefficient based on the dimensionless first feature vector and the second feature vector to obtain the correlation coefficient between the dimensionless first feature vector and the second feature vector;

[0085] Step S43. Calculate the correlation degree value between the dimensionless first feature vector and the second feature vector based on the correlation coefficient;

[0086] Specifically, the calculation formula for the correlation coefficient is:

[0087]

[0088] Among them, x1 is the first dimensionless eigenvector; x2 is the second dimensionless eigenvector; t1 is the information of the first information collection time corresponding to the first eigenvector; t2 is the information of the collection time of the second information corresponding to the second eigenvector; σ is the resolution coefficient, and its value range is [1, 0]; γ is the relationship coefficient between the first dimensionless eigenvector and the second dimensionless eigenvector.

[0089] By quantifying the correlation degree between the first eigenvector and the second eigenvector, it provides an effective basis for subsequent observation, modeling or decision-making, and helps the model better understand the internal relationship between these two eigenvectors.

[0090] Step S50. Send the first eigenvector, the second eigenvector and the correlation degree value of each GNSS receiver in the water vapor network to a preset water vapor observation model for observation, and obtain the atmospheric water vapor information in the area where the GNSS receiver is located in a preset future time period.

[0091] Specifically, send the first eigenvectors of multiple GNSS receivers in the same water vapor network, as well as the corresponding second eigenvectors, to a preset water vapor observation model. The model will analyze and calculate the water vapor situation at the location of the GNSS receiver in the water vapor network based on the input eigenvectors and the correlation degree value by using specific algorithms and principles, so as to obtain the atmospheric water vapor information in the area where the receiver is located in a preset future time period, such as data on water vapor content, vertical distribution of water vapor, etc. The obtained prediction data can provide an important decision-making basis for multiple fields, especially in dealing with extreme weather, resource management, environmental protection and public safety.

[0092] Embodiment 2:

[0093] As Figure 2 shown, this embodiment provides a GNSS-based water vapor network observation system, which includes:

[0094] An acquisition unit 10, configured to acquire first information and second information, where the first information is satellite data collected by a water vapor network in a target area and its corresponding time series, and the second information is atmospheric water vapor data of the target area, and the water vapor network includes multiple GNSS receivers;

[0095] An extraction unit 20, configured to perform principal component analysis on the first information to extract atmospheric influence factors, and input the atmospheric influence factors into a preset feature transformation model for feature transformation to obtain a first eigenvector corresponding to each GNSS receiver;

[0096] A fusion unit 30, configured to input the second information into a preset feature fusion model for feature fusion to obtain a second eigenvector;

[0097] The first analysis unit 40 is configured to perform correlation degree analysis based on the first feature vector and the second feature vector, and determine the correlation degree value between the first feature vector of each GNSS receiver and the second feature vector of the atmospheric water vapor information in the target area;

[0098] The observation unit 50 is configured to send the first feature vector, the second feature vector, and the correlation degree value of each GNSS receiver in the water vapor network to a preset water vapor observation model for observation, so as to obtain the atmospheric water vapor information in the target area in a preset time period in the future.

[0099] In a specific implementation manner disclosed in the present application, the extraction unit 20 includes:

[0100] The second analysis unit is configured to perform principal component analysis processing on the satellite data in the first information, calculate the covariance matrix of each feature in the satellite data, perform eigenvalue decomposition, and obtain a simplified principal component feature set;

[0101] The first construction unit is configured to construct a k-nearest neighbor graph based on the distance difference and time difference between sample points in the principal component feature set. The nodes in the k-nearest neighbor graph represent sample points, and the weight of the edge represents the similarity degree value between the corresponding sample points;

[0102] The transfer unit is configured to perform a graph convolution operation based on the k-nearest neighbor graph, transfer the information of each sample point to adjacent sample points, and update the features by combining the information of adjacent sample points, so as to obtain the updated sample point information;

[0103] The pooling unit is configured to perform a pooling operation on all the updated sample point information to obtain multiple overall sample information;

[0104] The sending unit is configured to send the overall sample information after the pooling operation to a fully connected layer to obtain the first feature vector.

[0105] In a specific implementation manner disclosed in the present application, the first construction unit includes:

[0106] The first calculation unit is configured to calculate the distance difference and time difference between sample points in the principal component feature set and construct vectors, respectively obtaining a distance vector and a time vector;

[0107] The second calculation unit is configured to calculate the similarity degree value between two sample points based on the distance vector and time vector respectively corresponding to the two sample points;

[0108] The second construction unit is configured to construct a k-nearest neighbor graph based on the similarity degree value, and the length of the edge in the k-nearest neighbor graph is a set threshold.

[0109] In a specific implementation manner disclosed in the present application, the fusion unit 30 includes:

[0110] A third analysis unit for analyzing and processing the second information, wherein the second information is divided into multiple levels, and there is a sequential inclusion relationship from top to bottom for each level, and a judgment matrix is constructed based on the comparison of the importance of the data corresponding to two levels;

[0111] A judgment unit for calculating the weight of each level in the judgment matrix based on the AHP algorithm to obtain the relative importance value of each data in the second information;

[0112] A sorting unit for sorting the relative importance values in order of magnitude and performing feature extraction and fusion on the data whose sorting is before the set position to obtain a second feature vector.

[0113] In a specific implementation manner disclosed in the present application, the first calculation unit includes:

[0114] A third calculation unit for calculating the Euclidean distance between a sample point and the remaining sample points to obtain a plurality of first distances;

[0115] A first determination unit for determining the values within a preset distance range from the plurality of first distances as the reference distance;

[0116] A fourth calculation unit for calculating the difference between the first distance corresponding to the sample point and the reference distance to obtain a plurality of second distances;

[0117] A first composition unit for forming a distance vector based on the second distances;

[0118] A fifth calculation unit for calculating the time difference between a sample point and the remaining sample points to obtain a plurality of first differences;

[0119] A second determination unit for determining the values within a preset time range from the plurality of first differences as the reference difference;

[0120] A sixth calculation unit for calculating the difference between the first difference corresponding to the sample point and the reference difference to obtain a plurality of second differences;

[0121] A second composition unit for forming a time vector based on the second differences.

[0122] In a specific implementation manner disclosed in the present application, the transfer unit includes:

[0123] A "as" unit for using the similarity degree value matrix as a weight matrix and inputting the weight matrix and a preset activation function into the input layer of the graph convolutional neural network;

[0124] A weighting unit, configured to perform weighted summation on the information of each sample point and the information transmitted by adjacent sample points, and perform a linear transformation through a weight matrix to obtain the feature data information after the linear transformation;

[0125] A mapping unit, configured to perform a non-linear mapping on the feature data information after the linear transformation based on an activation function to obtain the updated sample point information.

[0126] In a specific embodiment disclosed in the present application, the first analysis unit 40 includes:

[0127] A processing unit, configured to perform dimensionless processing on the first feature vector and the second feature vector based on the mean conversion method to respectively obtain the dimensionless first feature vector and the second feature vector;

[0128] A seventh calculation unit, configured to calculate a relationship coefficient based on the dimensionless first feature vector and the second feature vector to obtain the relationship coefficient between the dimensionless first feature vector and the second feature vector;

[0129] An eighth calculation unit, configured to calculate an association degree value of the dimensionless first feature vector and the second feature vector based on the relationship coefficient.

[0130] It should be noted that regarding the system in the above embodiments, the specific manners in which each module performs operations have been described in detail in the embodiments related to the method, and will not be elaborated herein.

[0131] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention can have various changes and modifications. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

[0132] The above is only the specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or replacements within the technical scope disclosed by the present invention, and all should be covered by the protection scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.

Claims

1. A GNSS-based water vapor network observation method, characterized in that: include: Acquire first information and second information, wherein the first information is satellite data collected by a water vapor network in a target area and its corresponding time series, and the second information is atmospheric water vapor data in the target area, wherein the water vapor network includes multiple GNSS receivers; Performing principal component analysis based on the first information to extract atmospheric impact factors, and inputting the atmospheric impact factors into a preset feature conversion model for feature conversion to obtain a first feature vector corresponding to each GNSS receiver; Inputting the second information into a preset feature fusion model for feature fusion to obtain a second feature vector; Performing a correlation analysis based on the first eigenvector and the second eigenvector to determine a correlation value between the first eigenvector of each GNSS receiver and the second eigenvector of the atmospheric water vapor information of the target area; The first eigenvector, the second eigenvector and the correlation degree value of each GNSS receiver in the water vapor network are sent to a preset water vapor observation model for observation to obtain the atmospheric water vapor information of the target area in a preset time period in the future.

2. The GNSS-based water vapor network observation method according to claim 1 is characterized in that , based on the first information, principal component analysis is performed to extract atmospheric influence factors, and the atmospheric influence factors are input into a preset feature conversion model for feature conversion to obtain a first feature vector corresponding to each GNSS receiver, including: Performing principal component analysis on the satellite data in the first information, calculating the covariance matrix of each feature in the satellite data, and performing eigenvalue decomposition to obtain a simplified principal component feature set; Based on the distance difference and time difference between each sample point in the principal component feature set, a k nearest neighbor graph is constructed, wherein the nodes in the k nearest neighbor graph represent the sample points, and the edge weights represent the similarity values ​​between the corresponding sample points; Performing a graph convolution operation based on the k-nearest neighbor graph, transferring information of each sample point to adjacent sample points, and performing feature update in combination with information of adjacent sample points to obtain updated sample point information; All updated sample point information is pooled to obtain multiple overall sample information; The overall sample information after the pooling operation is sent to the fully connected layer to obtain the first feature vector.

3. The GNSS-based water vapor network observation method according to claim 2 is characterized in that ,Based on the distance difference and time difference between each sample point in the principal component feature set, a k nearest neighbor graph is constructed, including: Calculating the distance difference and time difference between each sample point in the principal component feature set and constructing vectors to obtain a distance vector and a time vector respectively; Based on the distance vector and time vector corresponding to the two sample points, the similarity value of the two sample points is calculated; The k nearest neighbor graph is constructed based on the similarity value, and the length of the edge in the k nearest neighbor graph is a set threshold.

4. The GNSS-based water vapor network observation method according to claim 1, characterized in that , inputting the second information into a preset feature fusion model for feature fusion to obtain a second feature vector, including: Analyze and process the second information, wherein the second information is divided into multiple levels, each level has a sequential inclusion relationship from top to bottom, and compare the importance of data corresponding to each of the two levels to construct a judgment matrix; Calculate the weight of each level in the judgment matrix based on the AHP algorithm to obtain the relative importance value of each data in the second information; The relative importance values ​​are sorted in order of size, and feature extraction and fusion are performed on the data that is sorted before the set position to obtain the second feature vector.

5. The GNSS-based water vapor network observation method according to claim 3 is characterized in that , calculate the distance difference and time difference between each sample point in the principal component feature set and construct a vector, and obtain the distance vector and time vector respectively, including: Calculating the Euclidean distance between the sample point and other sample points to obtain multiple first distances; Determine a value within a preset distance range from the plurality of first distances as a reference distance; Calculating the difference between the first distance corresponding to the sample point and the reference distance to obtain a plurality of second distances; constructing the distance vector based on the second distance; Calculating the time difference between the sample point and the remaining sample points to obtain a plurality of first difference values; Determine a value within a preset time range from the plurality of first difference values ​​as a reference difference value; Calculate the difference between the first difference value corresponding to the sample point and the reference difference value to obtain a plurality of second difference values; The time vector is constructed based on the second difference.

6. The GNSS-based water vapor network observation method according to claim 3 is characterized in that , based on the k nearest neighbor graph, a graph convolution operation is performed, the information of each sample point is transferred to the adjacent sample points, and the features are updated in combination with the information of the adjacent sample points to obtain the updated sample point information, including: Using the similarity value matrix as a weight matrix, and inputting the weight matrix and a preset activation function into an input layer in a graph convolutional neural network; The information of each sample point is weighted and summed with the information transmitted by the adjacent sample points, and linearly transformed through the weight matrix to obtain the feature data information after linear transformation; Based on the activation function, the linearly transformed feature data information is nonlinearly mapped to obtain the updated sample point information.

7. The GNSS-based water vapor network observation method according to claim 1, characterized in that , performing a correlation degree analysis based on the first feature vector and the second feature vector to determine a correlation degree value between the first feature vector of each GNSS receiver and the second feature vector of the target area, including: Performing dimensionless processing on the first eigenvector and the second eigenvector based on a mean conversion method to obtain dimensionless first eigenvector and second eigenvector respectively; Calculating the relationship coefficient based on the dimensionless first eigenvector and the second eigenvector to obtain the relationship coefficient between the dimensionless first eigenvector and the second eigenvector; The dimensionless correlation degree value between the first eigenvector and the second eigenvector is calculated based on the relationship coefficient.

8. The GNSS-based water vapor network observation method according to claim 6, characterized in that ,The sample point information update formula includes: Among them, s' i is the updated sample point information of the i-th sample point; s i is the sample point information of the i-th sample point; s j is the sample point information of the jth sample point; X ij is the updated sample point information correlation value of the i-th sample point; W is the weight matrix; δ is the activation function; N is the number of sample points.

9. The GNSS-based water vapor network observation method according to claim 7, characterized in that , the calculation formula of the relationship coefficient is: Among them, x1 is the first eigenvector after dimensionless conversion; x2 is the second eigenvector after dimensionless conversion; t1 is the first information collection time information corresponding to the first eigenvector; t2 is the second information collection time information corresponding to the second eigenvector; σ is the resolution coefficient, and its value range is [1,0]; γ is the relationship coefficient between the first eigenvector and the second eigenvector after dimensionless conversion.

10. A GNSS-based water vapor network observation system, characterized in that: include: an acquisition unit, configured to acquire first information and second information, wherein the first information is satellite data collected by a water vapor network in a target area and its corresponding time series, and the second information is atmospheric water vapor data in the target area, wherein the water vapor network includes a plurality of GNSS receivers; an extraction unit, configured to perform principal component analysis based on the first information to extract atmospheric influence factors, and input the atmospheric influence factors into a preset feature conversion model for feature conversion to obtain a first feature vector corresponding to each GNSS receiver; A fusion unit, used for inputting the second information into a preset feature fusion model for feature fusion to obtain a second feature vector; A first analysis unit is used to perform a correlation degree analysis based on the first eigenvector and the second eigenvector to determine a correlation degree value between the first eigenvector of each GNSS receiver and the second eigenvector of the atmospheric water vapor information of the target area; The observation unit is used to send the first eigenvector, the second eigenvector and the correlation degree value of each GNSS receiver in the water vapor network to a preset water vapor observation model for observation, so as to obtain the atmospheric water vapor information of the target area in a preset time period in the future.

Images