Method for acquiring data of different depths of seawater based on salinity sensor

Abstract

The invention relates to the technical field of marine science, in particular to a method for acquiring data of different depths of seawater based on a salinity sensor, and the method comprises the following steps: S1, collecting salinity data of seawater at different depths in real time; s2, preprocessing the collected salinity data; s3, constructing a depth-salinity mathematical model by adopting a Gaussian process regression algorithm; s4, acquiring environmental data in real time, and dynamically correcting parameters of the depth-salinity model according to the environmental data; and S5, inputting the preprocessed salinity data in the S2 into the dynamically corrected depth-salinity mathematical model, and calculating corresponding seawater depth information. According to the method, a nonlinear mathematical model of the depth and the salinity is constructed through a Gaussian process regression algorithm and a dynamic correction technology, and accurate prediction and dynamic adaptation of the relationship between the salinity and the depth in a complex marine environment are realized in combination with data preprocessing and inverse calculation.

Description

Technical Field

[0001] The invention relates to the field of marine science and technology, and in particular to a method for acquiring data of different depths of seawater based on a salinity sensor. Background Art

[0002] With the growing demand for marine resource development and environmental monitoring, accurately obtaining salinity data at different ocean depths has become an important technical issue. Salinity data can not only reflect the chemical properties of marine water bodies, but also provide key support for marine ecosystem research, ocean current analysis, and climate change prediction. However, due to the complexity and dynamics of the marine environment, the salinity distribution at different depths is affected by many factors, including water temperature, ocean current speed, and seasonal changes. The collection of salinity data and depth correlation analysis have high technical challenges. Traditional salinity measurement methods usually rely on single-point sampling or mechanical depth sounding equipment. These methods have obvious limitations in coverage, real-time and data accuracy, and it is difficult to meet the monitoring needs in complex marine environments.

[0003] In the prior art, the research on the depth-salinity relationship mainly focuses on empirical formulas or simple linear regression models. Such methods cannot effectively capture the nonlinear relationship between depth and salinity. In addition, the dynamic change characteristics of salinity in different ocean environments have not been fully considered, resulting in poor adaptability of the prediction results. Especially in the real-time changing ocean environment, the influence of environmental factors such as water temperature and ocean current speed on salinity distribution is often ignored, resulting in the actual application effect of the model being limited. Summary of the invention

[0004] Based on the above purpose, the present invention provides a method for acquiring data of different depths of seawater based on a salinity sensor.

[0005] The method for acquiring data of different depths of seawater based on a salinity sensor comprises the following steps: S1: Deploy salinity sensors at different depths in the target ocean to collect salinity data of seawater at different depths in real time; S2: preprocessing the collected salinity data, including noise removal, outlier removal and missing data interpolation; S3: Based on long-term historical salinity data, a depth-salinity mathematical model was constructed using the Gaussian process regression algorithm; S4: Real-time collection of environmental data, including water temperature and ocean current speed, and dynamic correction of the parameters of the depth-salinity model based on the environmental data, adjusting the kernel function and related hyperparameters in the mathematical model of depth and salinity to adapt to the real-time changing ocean environment; S5: Input the salinity data pre-processed in S2 into the dynamically corrected depth-salinity mathematical model to calculate the corresponding seawater depth information.

[0006] Optionally, the S1 specifically includes: S11: Determine the sampling depth location according to the water depth distribution of the target sea area, specifically select several depth points in the surface layer of 0-50 meters, the middle layer of 50-200 meters and the deep layer below 200 meters; S12: at each selected depth point, deploy a salinity sensor, each of which is fixed at a corresponding depth by a mooring cable or an anchoring device; S13: Use the depth meter of the salinity sensor to monitor and record its depth information in real time to ensure that the collected salinity data matches the corresponding depth one by one.

[0007] Optionally, the S2 specifically includes: S21: noise removal processing is performed on the collected salinity data, and the salinity signal is filtered using a bandpass filter to remove high-frequency noise and low-frequency interference; S22: Remove outliers from salinity data, and use the Z-Score algorithm to detect the salinity value of each data point. If the Z value of a data point exceeds a set threshold, the data point is considered an outlier and is removed; S23: Interpolate and complete missing data. Linear interpolation or spline interpolation is used to fill in missing salinity data in the time series to ensure data continuity and consistency.

[0008] Optionally, the S3 specifically includes: S31: Obtain long-term historical salinity data of the target sea area, including salinity observations at different periods and depths, covering a period of at least two years; S32: Normalize the historical salinity data and adjust the salinity data at different depths to the standardized range. Specifically, the minimum-maximum normalization method is used to map the salinity values ​​to the interval [0, 1] to eliminate the magnitude differences of data at different depths and observation periods. S33: Based on the historical salinity data normalized in S32, the depth of each data record is used as an input variable and the salinity is used as an output variable to construct an input-output pair of depth-salinity to form a training data set; S34: Based on the training data set formed by S33, the Gaussian process regression algorithm is used to train the depth-salinity mathematical model. Specifically, the radial basis function is selected as the kernel function, and the hyperparameters of the kernel function are optimized by the maximum likelihood estimation method, so that the depth-salinity mathematical model can capture the nonlinear relationship between depth and salinity.

[0009] Optionally, the S33 specifically includes: S331: Extract each data record from the historical salinity data normalized by S32, including the depth and normalized salinity , where i represents the sequence number of the data; S332: Depth As the input variable of the model, construct the input vector X, which is expressed as: , where n is the total number of data records, Indicates the depth value of each record; S333: Normalize salinity value As the output variable of the model, construct the output vector Y, which is expressed as: ,in, Indicates the salinity value corresponding to each record; S334: Pair the input vector X and the output vector Y to form a training data set T, which is expressed as: ; S335: In order to improve the coverage of the training data set, the training data is enhanced; specifically, a linear interpolation method is used to fill in the missing data within the continuous depth range.

[0010] Optionally, the S34 specifically includes: S341: Initialize the Gaussian process regression model and select the radial basis function as the kernel function to measure the similarity between different depths in the training data; S342: Based on the training data set, calculate the similarity of each pair of depth data to generate a covariance matrix, where each element in the covariance matrix represents the correlation between two depth positions; S343: Calculate the predicted salinity value at any test depth according to the covariance matrix and the salinity value in the training data, and obtain the predicted value corresponding to the test depth through the similarity between the training depth and the test depth and the observation noise parameter; S344: Optimizing the similarity parameters and observation noise parameters of the kernel function. The optimization process is based on maximizing the probability value of the training data under the model to ensure that the model can best fit the training data, thereby obtaining a depth-salinity mathematical model.

[0011] Optionally, the S4 specifically includes: S41: The water temperature and ocean current speed data of the target area are collected in real time through the deployed water temperature sensors and ocean current speed sensors; the water temperature sensors are used to accurately monitor the temperature changes of seawater; the ocean current speed sensors are used to measure the flow speed of the ocean surface or at different depths; at the same time, the collected data are pre-processed to remove noise and complete the missing data to ensure the integrity of the data; S42: Based on the pre-processed environmental data, analyze the effects of water temperature and ocean current speed on salinity changes, and determine the relationship between environmental data and the depth-salinity model; water temperature affects the solubility of salinity, and ocean current speed affects the mixing and distribution of salinity; S43: Based on the analysis results of S42, the kernel function and related hyperparameters in the depth-salinity model are adjusted; the temperature factor in the kernel function is adjusted specifically according to the change in water temperature to optimize the response of the model under different water temperatures; and the velocity factor in the kernel function is dynamically adjusted according to the change in ocean current speed.

[0012] Optionally, the S42 specifically includes: S421: extracting time series data of water temperature T and ocean current velocity v based on the preprocessed environmental data; S422: Analyze the effect of water temperature T on salinity solubility, and describe the relationship between water temperature change and salinity through the solubility of salinity. The formula is: ,in, represents the solubility of salinity under the influence of water temperature, Indicates standard water temperature The salinity value at is the temperature influence factor; S423: Analyze the effect of ocean current velocity v on salinity mixing and distribution. Use the salinity diffusion model in fluid dynamics to describe the effect of ocean current velocity changes on salinity gradients. The expression is: ,in, represents the rate of change of salinity distribution caused by ocean current velocity, is the ocean current influence factor, is the depth gradient of salinity; S424: The effects of water temperature and ocean current speed are comprehensively modeled to construct a comprehensive relationship expression of environmental data on salinity changes: ,in, It represents the salinity value under the combined effect of water temperature and ocean current speed at a specified depth d.

[0013] Optionally, the S43 specifically includes: S431: Based on the comprehensive relationship expression of environmental data on salinity changes, extract the impact of environmental factors on the salinity model, including water temperature factors and flow rate factor ; S432: Define the kernel function of the Gaussian process regression model and set the temperature factor and flow rate factor As a hyperparameter of the Gaussian process kernel function; S433: Dynamically adjust the hyperparameters in the kernel function based on the comprehensive relationship expression of environmental data to salinity changes; specifically adjust the temperature factor through the optimization algorithm that maximizes the log-likelihood function and flow rate factor , to minimize the prediction error; S434: Obtain the optimal temperature factor by maximizing the optimization of the log-likelihood function and flow rate factor The optimized parameters are fed back into the Gaussian process regression model to dynamically correct the depth-salinity mathematical model.

[0014] Optionally, the S5 specifically includes: S51: inputting the salinity data pre-processed in S2 into the dynamically corrected depth-salinity mathematical model; S52: Using the inverse function of the depth-salinity mathematical model, the corresponding seawater depth is calculated according to the salinity value; S53: Perform confidence interval analysis on the calculation results, quantify the uncertainty of the calculation results through the covariance matrix of the model, and generate predicted values ​​and confidence intervals for each depth to ensure the credibility of the results; S54: All predicted depth results are sorted into a depth set, matched with actual salinity data, and a final salinity-depth relationship table is output.

[0015] Beneficial effects of the present invention: The present invention, by combining the Gaussian process regression algorithm with the dynamic correction mechanism, accurately constructs a nonlinear mathematical model between depth and salinity, can comprehensively process salinity data at different depths and environmental factors such as water temperature and ocean current speed collected in real time, and models historical salinity data by using the Gaussian process regression algorithm, and dynamically adjusts the kernel function and hyperparameters of the model under real-time environmental changes, thereby achieving high-precision prediction of salinity distribution, effectively solving the problem that traditional models cannot adapt to changes in complex marine environments, and greatly improving the prediction accuracy and adaptability of the relationship between salinity and depth.

[0016] The present invention realizes high-precision derivation of seawater depth information from salinity data through data preprocessing, depth inverse calculation and confidence interval analysis technology. The preprocessing technology ensures the integrity and consistency of the data. The inverse calculation method combined with the dynamically adjusted mathematical model can accurately calculate the seawater depth information corresponding to the salinity data, providing reliable technical support for marine ecological research, climate change analysis and other related fields. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only for the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0018] Figure 1 A schematic diagram of a method for acquiring data at different depths of seawater according to an embodiment of the present invention; Figure 2 Schematic diagram of a method for dynamically correcting parameters according to an embodiment of the present invention. DETAILED DESCRIPTION

[0019] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. At the same time, it is explained here that in order to make the embodiments more detailed, the following embodiments are the best and preferred embodiments, and those skilled in the art may also adopt other alternatives to implement some known technologies; and the accompanying drawings are only for more specific description of the embodiments, and are not intended to specifically limit the present invention.

[0020] like Figure 1-Figure 2 As shown, a method for acquiring data of different depths of seawater based on a salinity sensor includes the following steps: S1: Deploy salinity sensors at different depths in the target ocean to collect salinity data of seawater at different depths in real time; S2: preprocessing the collected salinity data, including noise removal, outlier removal, and missing data interpolation, to make the data suitable for further analysis; S3: Based on long-term historical salinity data, the Gaussian process regression (GPR) algorithm is used to construct a depth-salinity mathematical model. The model models the complex nonlinear relationship between salinity and depth by selecting an appropriate kernel function, which can effectively adapt to the hydrological characteristics and seasonal changes of different regions; S4: Real-time collection of environmental data, including water temperature and ocean current speed, and dynamic correction of the parameters of the depth-salinity model based on the environmental data, adjusting the kernel function and related hyperparameters in the mathematical model of depth and salinity to adapt to the real-time changing ocean environment; S5: Input the salinity data pre-processed in S2 into the dynamically corrected depth-salinity mathematical model to calculate the corresponding seawater depth information.

[0021] S1 specifically includes: S11: Determine the sampling depth location according to the water depth distribution of the target sea area, specifically select several depth points in the surface layer of 0-50 meters, the middle layer of 50-200 meters and the deep layer below 200 meters; S12: At each selected depth point, a salinity sensor is deployed. Each salinity sensor is fixed at the corresponding depth by a mooring cable or an anchoring device to ensure its stability and measurement accuracy; S13: Use the depth meter of the salinity sensor to monitor and record its depth information in real time to ensure that the collected salinity data matches the corresponding depth one by one; through clear sampling depth selection and sensor deployment methods, ensure that accurate salinity data is obtained at different seawater depths, providing a reliable foundation for subsequent data analysis and model building.

[0022] The preprocessing of the collected salinity data in S2 specifically includes: S21: Perform noise removal on the collected salinity data, use a bandpass filter to filter the salinity signal, remove high-frequency noise and low-frequency interference, and ensure the smoothness and effectiveness of the data; the cutoff frequency of the filter is set according to the fluctuation characteristics of the target sea area; S22: Remove outliers from salinity data. Use the Z-Score algorithm to detect the salinity value of each data point. If the Z value of a data point exceeds the set threshold, the data point is considered an outlier and is removed. The threshold is set to 3 to ensure that most abnormal fluctuations are effectively identified and eliminated. S23: Interpolate and complete missing data. Use linear interpolation or spline interpolation to fill in the missing salinity data in the time series to ensure data continuity and consistency. The specific interpolation method is selected according to the location of the missing data and the rules of adjacent data. Through preprocessing steps such as noise removal, outlier removal and missing data interpolation, the quality of salinity data can be significantly improved, unnecessary interference information can be removed, and the accuracy and reliability of subsequent data analysis and model training can be ensured.

[0023] S3 specifically includes: S31: Obtain long-term historical salinity data of the target sea area, including salinity observations at different periods and depths, covering a period of at least two years, to ensure that the data reflects the seasonal and hydrological characteristics of the sea area; S32: Normalize the historical salinity data and adjust the salinity data at different depths to the standardized range. Specifically, the minimum-maximum normalization method is used to map the salinity values ​​to the interval [0, 1] to eliminate the magnitude differences of data at different depths and observation periods. S33: Based on the historical salinity data after normalization in S32, the depth of each data record is used as the input variable and the salinity is used as the output variable to construct the input-output pair of depth-salinity to form a training data set; ensuring that the input and output data can accurately correspond to each other to facilitate model fitting; S34: Based on the training data set formed in S33, a Gaussian process regression algorithm is used to train a depth-salinity mathematical model. Specifically, a radial basis function is selected as a kernel function, and the hyperparameters of the kernel function are optimized by a maximum likelihood estimation method, so that the depth-salinity mathematical model can capture the nonlinear relationship between depth and salinity. The Gaussian process regression algorithm is used to construct a depth-salinity mathematical model through clear steps. The present invention can accurately describe the nonlinear relationship between depth and salinity, and lay a reliable data foundation for subsequent model correction and depth calculation.

[0024] S33 specifically includes: S331: Extract each data record from the historical salinity data normalized by S32, including the depth and normalized salinity , where i represents the sequence number of the data, ensuring that the data has temporal consistency and spatial distribution consistency; S332: Depth As the input variable of the model, construct the input vector X, which is expressed as: , where n is the total number of data records, Indicates the depth value of each record; S333: Normalize salinity value As the output variable of the model, construct the output vector Y, which is expressed as: ,in, Indicates the salinity value corresponding to each record; S334: Pair the input vector X and the output vector Y to form a training data set T, which is expressed as: ; Ensure that each input depth data accurately matches the corresponding salinity data; S335: In order to improve the coverage of the training data set, the training data is enhanced; specifically, a linear interpolation method is used to fill in the missing data within the continuous depth range, and the interpolation calculation formula is: ,in, and are the depths of the front and back points respectively, and are the corresponding salinity values, is the interpolated salinity value, is the depth of the interpolation point; the depth-salinity input-output pairs are constructed through clear steps, and the training data set is enhanced in combination with the interpolation algorithm to ensure the integrity and continuity of the data, providing rich and accurate training samples for the Gaussian process regression algorithm.

[0025] The training depth-salinity mathematical model in S34 specifically includes: S341: Initialize the Gaussian process regression model and select the radial basis function as the kernel function to measure the similarity between different depths in the training data; S342: Based on the training data set, calculate the similarity of each pair of depth data to generate a covariance matrix, where each element in the covariance matrix represents the correlation between two depth positions; S343: Calculate the predicted salinity value at any test depth according to the covariance matrix and the salinity value in the training data, and obtain the predicted value corresponding to the test depth through the similarity between the training depth and the test depth and the observation noise parameter; S344: Optimizing the similarity parameters and observation noise parameters of the kernel function. The optimization process is based on maximizing the probability value of the training data under the model to ensure that the model can best fit the training data, thereby obtaining a depth-salinity mathematical model.

[0026] The specific training steps for training the depth-salinity mathematical model are as follows: First, initialize the Gaussian process regression model and select the radial basis function (RBF) as the kernel function. The kernel function is used to measure the similarity between different depths. The calculation formula is: ,in, and represents the i-th and j-th depth positions in the depth dataset, is the length scale parameter of the kernel function, which is used to control the decay rate of the similarity between depths; Then, based on the training data set T, the covariance matrix K of Gaussian process regression is constructed. The matrix elements Indicates depth and depth Similarities between: , where i and j represent any two records in the data set; Next, the test depth is calculated using the covariance matrix K and the salinity value S of the training data Predicted salinity under , the formula is: ,in, Indicates the test depth and the kernel function value vector of the training depth set, represents the variance of the observation noise, and I represents the identity matrix; Then, the kernel function hyperparameters are optimized by maximum likelihood estimation. and the observation noise variance , so that the likelihood function of the training data set under a given model is maximized. The likelihood function calculation formula is: , where L represents the likelihood function, represents the determinant of the covariance matrix, is the number of records in the training data set; Finally, according to the optimized hyperparameters and ,Using the depth and salinity pairs in the training data set T, the Gaussian process regression model is trained, and the depth-salinity mathematical model is obtained.

[0027] S4 specifically includes: S41: Through the deployed water temperature sensors and ocean current speed sensors, the water temperature and ocean current speed data of the target area are collected in real time; the water temperature sensor is used to accurately monitor the temperature changes of seawater; the ocean current speed sensor is used to measure the flow speed of the ocean surface or at different depths; at the same time, the collected data is pre-processed to remove noise and complete the missing data to ensure the integrity of the data; S42: Based on the pre-processed environmental data, analyze the effects of water temperature and ocean current speed on salinity changes, and determine the relationship between environmental data and the depth-salinity model; water temperature affects the solubility of salinity, and ocean current speed affects the mixing and distribution of salinity; S43: Based on the analysis results of S42, the kernel function and related hyperparameters in the depth-salinity model are adjusted; the temperature factor in the kernel function is adjusted specifically according to the change in water temperature to optimize the response of the model at different water temperatures; and the velocity factor in the kernel function is dynamically adjusted according to the change in ocean current speed, so that the depth-salinity model can better adapt to the changes in salinity distribution caused by changes in ocean current speed; by collecting environmental data (such as water temperature and ocean current speed) in real time and adjusting the kernel function and hyperparameters in the model, the depth-salinity model can be adapted to different marine environmental conditions, thereby improving the accuracy and flexibility of salinity prediction and ensuring that the prediction model can maintain high accuracy and reliability in various actual environments.

[0028] S42 specifically includes: S421: Based on the pre-processed environmental data, extract the time series data of water temperature T and ocean current velocity v, corresponding to different depth positions , forming a data pair ; S422: Analyze the effect of water temperature T on salinity solubility, and describe the relationship between water temperature change and salinity through the solubility of salinity. The formula is: ,in, represents the solubility of salinity under the influence of water temperature, Indicates standard water temperature The salinity value at is the temperature influence factor, which describes the attenuation rate of water temperature on the salinity dissolving capacity; S423: Analyze the effect of ocean current velocity v on salinity mixing and distribution. Use the salinity diffusion model in fluid dynamics to describe the effect of ocean current velocity changes on salinity gradients. The expression is: ,in, represents the rate of change of salinity distribution caused by ocean current velocity, is the ocean current influence factor, is the depth gradient of salinity; S424: The effects of water temperature and ocean current speed are comprehensively modeled to construct a comprehensive relationship expression of environmental data on salinity changes: ,in, It represents the salinity value under the combined effect of water temperature and ocean current velocity at a specified depth d. By analyzing in detail the effects of water temperature on salinity solubility and the effects of ocean current velocity on salinity distribution, an accurate relationship expression between environmental data and the depth-salinity model is constructed, providing a clear theoretical basis and practical application solution for dynamically correcting model parameters, ensuring that the model adapts to real-time changes in the marine environment.

[0029] S43 specifically includes: S431: Based on the comprehensive relationship expression of environmental data on salinity changes, extract the impact of environmental factors on the salinity model, including water temperature factors and flow rate factor , which are used to describe the influence of water temperature and ocean current speed on salinity; S432: Defining the kernel function of the Gaussian process regression model , and the temperature factor and flow rate factor As the hyperparameter of the Gaussian process kernel function, the expression is: , where the parameters of the kernel function are set to initial values ​​according to environmental conditions; S433: Dynamically adjust the hyperparameters in the kernel function based on the comprehensive relationship expression of environmental data to salinity changes; specifically adjust the temperature factor through the optimization algorithm that maximizes the log-likelihood function and flow rate factor , to minimize the prediction error, the formula is: ,in, is the log-likelihood function, y is the observed value of salinity change, K is the kernel matrix, is the sample size; S434: Obtain the optimal temperature factor by maximizing the optimization of the log-likelihood function and flow rate factor , and feed the optimized parameters back to the Gaussian process regression model to dynamically correct the depth-salinity mathematical model; the above steps can significantly improve the adaptability and prediction accuracy of the salinity depth model to different environmental conditions by dynamically adjusting the temperature factor and velocity factor in the Gaussian process regression model and optimizing the kernel function hyperparameters of the model, providing a scientific basis for deep salinity prediction in complex marine environments.

[0030] S5 specifically includes: S51: input the pre-processed salinity data in S2 into a dynamically corrected depth-salinity mathematical model, which is constructed and dynamically adjusted by a Gaussian process regression algorithm and can describe the nonlinear relationship between depth and salinity; S52: Using the inverse function of the depth-salinity mathematical model, the corresponding seawater depth is calculated according to the salinity value; S53: Perform confidence interval analysis on the calculation results, quantify the uncertainty of the calculation results through the covariance matrix of the model, and generate predicted values ​​and confidence intervals for each depth to ensure the credibility of the results; S54: All predicted depth results are organized into depth sets, matched with actual salinity data, and the final salinity-depth relationship table is output; by inputting the preprocessed salinity data into the dynamically corrected depth-salinity mathematical model, combined with the model's inverse calculation and confidence interval analysis, the corresponding seawater depth information can be accurately derived, providing precise data support for marine environment monitoring and salinity change analysis.

[0031] The present invention covers any substitution, modification, equivalent method and scheme made on the essence and scope of the present invention. In order to make the public have a thorough understanding of the present invention, specific details are described in detail in the following preferred embodiments of the present invention, but those skilled in the art can fully understand the present invention without the description of these details. In addition, in order to avoid unnecessary confusion about the essence of the present invention, well-known methods, processes, procedures, components and circuits are not described in detail.

[0032] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A method for acquiring data at different depths of seawater based on a salinity sensor, characterized in that: The following steps are involved: S1: Deploy salinity sensors at different depths in the target ocean to collect salinity data of seawater at different depths in real time; S2: preprocessing the collected salinity data, including noise removal, outlier removal and missing data interpolation; S3: Based on long-term historical salinity data, a depth-salinity mathematical model was constructed using the Gaussian process regression algorithm; S4: Real-time collection of environmental data, including water temperature and ocean current speed, and dynamic correction of the parameters of the depth-salinity model based on the environmental data, adjusting the kernel function and related hyperparameters in the mathematical model of depth and salinity to adapt to the real-time changing ocean environment; S5: Input the salinity data pre-processed in S2 into the dynamically corrected depth-salinity mathematical model to calculate the corresponding seawater depth information.

2. The method for acquiring data of different depths of seawater based on a salinity sensor according to claim 1, characterized in that: The S1 specifically includes: S11: Determine the sampling depth location according to the water depth distribution of the target sea area, specifically select several depth points in the surface layer of 0-50 meters, the middle layer of 50-200 meters and the deep layer below 200 meters; S12: at each selected depth point, deploy a salinity sensor, each of which is fixed at a corresponding depth by a mooring cable or an anchoring device; S13: Use the depth meter of the salinity sensor to monitor and record its depth information in real time to ensure that the collected salinity data matches the corresponding depth one by one.

3. The method for acquiring data of different depths of seawater based on a salinity sensor according to claim 1, characterized in that: The S2 specifically includes: S21: noise removal processing is performed on the collected salinity data, and the salinity signal is filtered using a bandpass filter to remove high-frequency noise and low-frequency interference; S22: Remove outliers from salinity data, and use the Z-Score algorithm to detect the salinity value of each data point. If the Z value of a data point exceeds a set threshold, the data point is considered an outlier and is removed; S23: Interpolate and complete missing data. Linear interpolation or spline interpolation is used to fill in missing salinity data in the time series to ensure data continuity and consistency.

4. The method for acquiring data of different depths of seawater based on a salinity sensor according to claim 1, characterized in that: The S3 specifically includes: S31: Obtain long-term historical salinity data of the target sea area, including salinity observations at different periods and depths, covering a period of at least two years; S32: Normalize the historical salinity data and adjust the salinity data at different depths to the standardized range. Specifically, the minimum-maximum normalization method is used to map the salinity values ​​to the interval [0, 1] to eliminate the magnitude differences of data at different depths and observation periods. S33: Based on the historical salinity data normalized in S32, the depth of each data record is used as an input variable and the salinity is used as an output variable to construct an input-output pair of depth-salinity to form a training data set; S34: Based on the training data set formed by S33, the Gaussian process regression algorithm is used to train the depth-salinity mathematical model. Specifically, the radial basis function is selected as the kernel function, and the hyperparameters of the kernel function are optimized by the maximum likelihood estimation method, so that the depth-salinity mathematical model can capture the nonlinear relationship between depth and salinity.

5. The method for acquiring data of different depths of seawater based on a salinity sensor according to claim 4, characterized in that: The S33 specifically includes: S331: Extract each data record from the historical salinity data normalized by S32, including the depth and normalized salinity , where i represents the sequence number of the data; S332: Depth As the input variable of the model, construct the input vector X, which is expressed as: , where n is the total number of data records, Indicates the depth value of each record; S333: Normalize salinity value As the output variable of the model, construct the output vector Y, which is expressed as: ,in, Indicates the salinity value corresponding to each record; S334: Pair the input vector X and the output vector Y to form a training data set T, which is expressed as: ; S335: In order to improve the coverage of the training data set, the training data is enhanced; specifically, a linear interpolation method is used to fill in the missing data within the continuous depth range.

6. The method for acquiring data of different depths of seawater based on a salinity sensor according to claim 5, characterized in that: The S34 specifically includes: S341: Initialize the Gaussian process regression model and select the radial basis function as the kernel function to measure the similarity between different depths in the training data; S342: Based on the training data set, calculate the similarity of each pair of depth data to generate a covariance matrix, where each element in the covariance matrix represents the correlation between two depth positions; S343: Calculate the predicted salinity value at any test depth according to the covariance matrix and the salinity value in the training data, and obtain the predicted value corresponding to the test depth through the similarity between the training depth and the test depth and the observation noise parameter; S344: Optimizing the similarity parameters and observation noise parameters of the kernel function. The optimization process is based on maximizing the probability value of the training data under the model to ensure that the model can best fit the training data, thereby obtaining a depth-salinity mathematical model.

7. The method for acquiring data of different depths of seawater based on a salinity sensor according to claim 1, characterized in that: The S4 specifically includes: S41: The water temperature and ocean current speed data of the target area are collected in real time through the deployed water temperature sensors and ocean current speed sensors; the water temperature sensors are used to accurately monitor the temperature changes of seawater; the ocean current speed sensors are used to measure the flow speed of the ocean surface or at different depths; at the same time, the collected data are pre-processed to remove noise and complete the missing data to ensure the integrity of the data; S42: Based on the pre-processed environmental data, analyze the effects of water temperature and ocean current speed on salinity changes, and determine the relationship between environmental data and the depth-salinity model; water temperature affects the solubility of salinity, and ocean current speed affects the mixing and distribution of salinity; S43: Based on the analysis results of S42, the kernel function and related hyperparameters in the depth-salinity model are adjusted; the temperature factor in the kernel function is adjusted specifically according to the change in water temperature to optimize the response of the model under different water temperatures; and the velocity factor in the kernel function is dynamically adjusted according to the change in ocean current speed.

8. The method for acquiring data of different depths of seawater based on a salinity sensor according to claim 1, characterized in that: The S42 specifically includes: S421: extracting time series data of water temperature T and ocean current velocity v based on the preprocessed environmental data; S422: Analyze the effect of water temperature T on salinity solubility, and describe the relationship between water temperature change and salinity through the solubility of salinity. The formula is: ,in, It indicates the solubility of salinity under the influence of water temperature. Indicates standard water temperature The salinity value at is the temperature influence factor; S423: Analyze the effect of ocean current velocity v on salinity mixing and distribution. Use the salinity diffusion model in fluid dynamics to describe the effect of ocean current velocity changes on salinity gradients. The expression is: ,in, represents the rate of change of salinity distribution caused by ocean current velocity, is the ocean current influence factor, is the depth gradient of salinity; S424: The effects of water temperature and ocean current speed are comprehensively modeled to construct a comprehensive relationship expression of environmental data on salinity changes: ,in, It represents the salinity value under the combined effect of water temperature and ocean current speed at a specified depth d.

9. The method for acquiring data of different depths of seawater based on a salinity sensor according to claim 8, characterized in that: The S43 specifically includes: S431: Based on the comprehensive relationship expression of environmental data on salinity changes, extract the impact of environmental factors on the salinity model, including water temperature factors and flow rate factor ; S432: Define the kernel function of the Gaussian process regression model and set the temperature factor and flow rate factor As a hyperparameter of the Gaussian process kernel function; S433: Dynamically adjust the hyperparameters in the kernel function based on the comprehensive relationship expression of environmental data to salinity changes; specifically adjust the temperature factor through the optimization algorithm that maximizes the log-likelihood function and flow rate factor , to minimize the prediction error; S434: Obtain the optimal temperature factor by maximizing the optimization of the log-likelihood function and flow rate factor The optimized parameters are fed back into the Gaussian process regression model to dynamically correct the depth-salinity mathematical model.

10. The method for acquiring data of different depths of seawater based on a salinity sensor according to claim 1, characterized in that: The S5 specifically includes: S51: inputting the salinity data pre-processed in S2 into the dynamically corrected depth-salinity mathematical model; S52: Using the inverse function of the depth-salinity mathematical model, the corresponding seawater depth is calculated according to the salinity value; S53: Perform confidence interval analysis on the calculation results, quantify the uncertainty of the calculation results through the covariance matrix of the model, and generate predicted values ​​and confidence intervals for each depth to ensure the credibility of the results; S54: All predicted depth results are sorted into a depth set, matched with actual salinity data, and a final salinity-depth relationship table is output.

Images