Multi-method coordinated seawater temperature and salinity pressure multi-parameter high-precision demodulation method

Abstract

The application provides a multi-method coordinated seawater temperature-salinity pressure multi-parameter high-precision demodulation method, and belongs to the technical field of sensing signal demodulation based on computer data processing; first, a local encapsulated micro-optical fiber coaxial Mach-Zehnder interferometer sensor is used for sensing measurement of seawater refractive index and temperature-salinity pressure, a database of seawater refractive index and temperature-salinity pressure and a detection wavelength is established, and the correction of the seawater refractive index in the infrared band is completed; meanwhile, a machine learning algorithm is used for training and prediction of the temperature-salinity pressure and characteristic wavelength value, so that the demodulation of any single parameter in the three parameters of the temperature-salinity pressure is realized; then, a double-parameter variable element sensitivity matrix containing changes of external environmental parameters is established; finally, based on the demodulation result of the machine learning single parameter, the influence of the parameter change on the peak shift is eliminated, and the demodulation of the remaining two parameters is realized by using the double-parameter variable element matrix. The application avoids the demodulation error caused by the complex cross-sensitivity problem among the three parameters, so that the demodulation result is more accurate.

Description

Technical Field

[0001] This invention belongs to the field of sensor signal demodulation technology based on computer data processing, and particularly relates to a high-precision demodulation method for multiple parameters of seawater temperature, salinity, and pressure using a multi-method collaborative approach. Background Technology

[0002] Seawater temperature and salinity are fundamental elements of marine hydrological observation and important parameters of the marine ecological environment. The variation of seawater temperature and salinity with depth (pressure) is called the temperature-salinity-depth (pressure) profile. Accurate measurement of temperature-salinity-pressure data is the basis for studying ocean dynamics processes and obtaining information such as seawater density and sound velocity distribution in marine surveys. With the development of fiber optic sensing technology, various types of fiber optic sensors for simultaneous measurement of multiple parameters of seawater temperature, salinity, and pressure have been proposed. However, after the sensor completes the parameter measurement, the demodulation method of the temperature-salinity-pressure sensing signal determines the accuracy of the measured temperature-salinity-pressure data.

[0003] Currently, demodulation of temperature, salinity, and pressure (TWP) sensor signals mainly employs constant sensitivity matrices and machine learning demodulation methods. However, compared to two-parameter demodulation using a second-order sensitivity matrix, demodulation using a third-order constant sensitivity matrix amplifies the cross-sensitivity between parameters, ultimately leading to an order-of-magnitude increase in demodulation error. Secondly, the sensitivity matrix elements used to construct the constant sensitivity matrix are measured under specific environmental parameters and cannot fully reflect the changes in temperature / salinity / pressure sensitivity with variations in the test environment. Demodulation using machine learning algorithms requires an extremely large amount of data for model training; training and prediction on small sample datasets makes it difficult to achieve high-precision demodulation of all three TWP parameters simultaneously.

[0004] Furthermore, obtaining the temperature / salinity / pressure sensitivity of micro-fiber sensors as a function of the ambient temperature, salinity, and pressure requires relying on the expression for the seawater refractive index with respect to seawater temperature, salinity, pressure, and the detection wavelength; however, existing expressions for the seawater refractive index... n ( T , S , P The infrared band (λ) is not applicable to fiber optic sensors. Therefore, before obtaining the sensitivity of temperature / salinity / pressure changes with the test environment and establishing the variable element sensitivity matrix, the seawater refractive index formula needs to be corrected for the infrared band. Summary of the Invention

[0005] To address the above problems, this invention proposes a multi-method collaborative high-precision demodulation method for seawater temperature, salinity, and pressure parameters, comprising the following steps:

[0006] S1. The refractive index of seawater is sensed and measured using a partially encapsulated micro-fiber coaxial Mach-Zehnder interferometer sensor. The refractive index of seawater in the infrared band is calculated based on the seawater dispersion relation, and the seawater refractive index sensitivity of the sensor is obtained by fitting it with the wavelength.

[0007] S2 utilizes a partially encapsulated micro-fiber coaxial Mach-Zehnder interferometer sensor to sense and measure seawater temperature, salinity, and pressure. Based on the seawater refractive index sensitivity obtained in S1, the refractive index values ​​under different temperature, salinity, and pressure conditions are calculated, and a database of seawater refractive index, temperature, salinity, pressure, and detection wavelength is established.

[0008] S3, based on the database established by S2, uses a machine learning fitting algorithm to obtain a corrected expression for the refractive index of seawater applicable to the infrared band.

[0009] S4. Establish a dataset of temperature, salinity, pressure, and detection wavelength, train and predict the machine learning regression prediction model, and demodulate any single parameter among the three parameters of temperature, salinity, and pressure.

[0010] S5. Using the modified seawater refractive index expression from S3 and the measured seawater refractive index sensing sensitivity, a dual-parameter variable element sensitivity matrix equation is established, which includes the variation of external environmental parameters.

[0011] S6, based on the demodulation results of the single parameter of machine learning in S4, combined with the measured sensitivity evaluation, and the influence of the parameter change on the peak shift is eliminated;

[0012] S7 uses the dual-parameter variable element sensitivity matrix equation to demodulate the remaining two parameters based on the change in trough wavelength.

[0013] Preferably, the specific implementation process of S1 is as follows:

[0014] A testing system was constructed, comprising a broadband light source, a spectrometer, and a micro-fiber sensor. The refractive index of seawater was altered by gradually diluting it with distilled water. Transmission spectra of a locally encapsulated micro-fiber coaxial Mach-Zehnder interferometer sensor at different seawater refractive indices were collected using the broadband light source and spectrometer, and the corresponding refractive indices were measured using an Abbe refractometer. The collected spectral signals were smoothed, and the wavelength values ​​of multiple troughs were extracted. The refractive index of seawater at different wavelengths was measured using an infrared refractometer, and the seawater dispersion relation was obtained by fitting the data.

[0015] ;

[0016] in n The refractive index of seawater; λ The wavelength is measured in nm; the refractive index of seawater measured by the Abbe refractometer is used to represent the refractive index of seawater. λ =589.0nm) corresponds to the infrared band, and the sensitivity of the sensor to seawater refractive index is calculated.

[0017] Preferably, the specific implementation process of S2 is as follows:

[0018] A testing system was constructed, comprising a broadband light source, a spectrometer, a micro-fiber sensor, and a pressure vessel. A partially encapsulated micro-fiber coaxial Mach-Zehnder interferometer temperature-salinity-pressure (TSP) sensor was placed in a pressure vessel filled with seawater. An external heating device was used to change the seawater temperature, and distilled water or a high-concentration sodium chloride solution was added to alter the salinity. An external pressure pump was used to pressurize or depressurize the seawater. A broadband light source and spectrometer collected spectral signals under different TSP conditions. Simultaneously, a temperature-salinity-depth (TSD) meter was placed inside the pressure vessel to record the TSP values ​​of the test environment. The refractive index value of the seawater corresponding to the random TSP spectral signal was calculated based on the refractive index sensor sensitivity. This refractive index value was then combined with the TSP values ​​measured by the TSD meter and the interference trough wavelengths to form a data set, which was then used to construct a database.

[0019] Preferably, the specific implementation process of S3 is as follows:

[0020] The database data established in S2 was fitted using a machine learning fitting algorithm (nonlinear least squares method) with five parameters, including seawater refractive index, temperature, salinity, pressure, and detection wavelength, to derive a seawater refractive index formula suitable for the infrared band:

[0021] ;

[0022] in, n Represents the refractive index of seawater. T Represents seawater temperature, measured in °C; S Represents seawater salinity, measured in ‰; P This represents seawater pressure, measured in MPa. λ This represents the detection wavelength, measured in nm.

[0023] Preferably, a random temperature, salinity, and pressure (THP) sensing experiment is conducted again on the partially encapsulated micro-fiber coaxial Mach-Zehnder interferometer. Multiple transmission spectra under random THP environments are collected, and multiple trough wavelength values ​​are extracted to obtain a dataset of THP and trough wavelength values. Using the THP database, 80% of the data is used for machine learning model training, and 20% is used for model prediction verification. Three machine learning algorithms, namely support vector machine, random forest, and neural network, are used to train and predict on the same data. The temperature demodulation result of the support vector machine is selected for subsequent salinity and pressure demodulation.

[0024] Preferably, the specific implementation process of S5 is as follows:

[0025] By fitting the corrected infrared band seawater refractive index expression using S3, the partial derivatives of the refractive index with respect to salinity and pressure are obtained. Using the measured refractive index sensitivity at different wavelengths, data fitting is performed on the center wavelength and refractive index sensitivity to obtain the functional relationship of refractive index sensitivity with respect to wavelength. Multiplying the partial derivatives with the refractive index sensitivity function, the salinity sensitivity function and pressure sensitivity function under different temperature, salinity, and pressure conditions are obtained as the four variable matrix elements of the second-order sensitivity matrix, and then the dual-parameter variable element sensitivity matrix and matrix equation are established.

[0026] Preferably, the specific implementation process of S6 is as follows:

[0027] Based on the temperature demodulation results, the trough wavelength shift caused by temperature changes is eliminated using the measured temperature sensitivity to obtain the new trough wavelength value:

[0028] ;

[0029] in, λ 0 represents the original wavelength of the trough, in nm. S T Temperature sensitivity of the trough, in nm / ℃; This represents the change in temperature.

[0030] Preferably, the specific implementation process of S7 is as follows:

[0031] By selecting any two troughs in the spectrum and using the established sensitivity matrix containing variable elements, the salinity and pressure values ​​of seawater can be obtained by solving the matrix equation.

[0032] Compared with the prior art, the present invention has the following beneficial effects:

[0033] 1. Compared with existing fiber optic temperature, salinity, and pressure sensor signal demodulation methods, the method of this invention combines machine learning with variable matrix, which not only compensates for the influence of environmental parameters on temperature / salinity / pressure sensitivity, but also avoids demodulation errors caused by the complex cross-sensitivity problem among the three parameters;

[0034] 2. Compared with the formula applicable to the relationship between seawater refractive index and temperature, salinity, pressure and wavelength in the 400-700nm wavelength range, the modified formula for the relationship between seawater refractive index and temperature, salinity, pressure and wavelength in the infrared band used in the method of this invention makes the demodulation results more accurate. Attached Figure Description

[0035] To more clearly illustrate the technical solutions of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the following description is only one embodiment of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0036] Figure 1 This is a schematic diagram of the overall process flow of the present invention.

[0037] Figure 2 This is a diagram of the seawater refractive index sensing test system in an embodiment of the present invention.

[0038] Figure 3 The images show transmission spectra of seawater at different refractive indices in this embodiment of the invention.

[0039] Figure 4 This is a dispersion trend diagram of seawater and pure water in an embodiment of the present invention.

[0040] Figure 5 This is a fitting diagram of the seawater refractive index sensitivity in the infrared band according to an embodiment of the present invention.

[0041] Figure 6 This is a diagram of the seawater temperature, salinity, and pressure three-parameter sensing and testing system in an embodiment of the present invention.

[0042] Figure 7 The images show transmission spectra under different seawater temperature, salinity, and pressure conditions in embodiments of the present invention. Detailed Implementation

[0043] The present invention will be further described below with reference to embodiments. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0044] The overall process of this invention is as follows Figure 1As shown, firstly, a partially encapsulated micro-fiber-coaxial Mach-Zehnder interferometer sensor is used to measure the refractive index and temperature-salinity-pressure (TSP) of seawater, establishing a database of seawater refractive index in relation to temperature, salinity, pressure, and detection wavelength, thus correcting the seawater refractive index formula in the infrared band. Simultaneously, machine learning algorithms are used to train and predict TSP and characteristic wavelength values, achieving high-precision demodulation of any single parameter among the three parameters. Then, using the corrected seawater refractive index formula and the measured seawater refractive index sensing sensitivity, a dual-parameter variable element sensitivity matrix is ​​established, incorporating variations in external environmental parameters. Finally, based on the demodulation results of the single parameter obtained through machine learning, combined with the measured sensitivity, the influence of parameter changes on peak shift is evaluated and eliminated. Afterward, the dual-parameter variable element matrix is ​​used to achieve accurate demodulation of the remaining two parameters.

[0045] In this embodiment, a refractive index sensing experiment was first conducted on a partially encapsulated micro-fiber coaxial Mach-Zehnder interferometer sensor. Distilled water was gradually added dropwise to dilute an existing seawater sample to change its refractive index. Figure 2 As can be seen, the transmission spectra of the locally packaged micro-fiber coaxial Mach-Zehnder interferometer sensor at different seawater refractive indices were collected using a broadband light source and a spectrometer, and the corresponding seawater refractive indices were measured using an Abbe refractometer. The collected transmission spectra at different seawater refractive indices and the corresponding Abbe refractometer measurements are shown below. Figure 3 As can be seen, the collected spectral signals were then smoothed using Origin software with a window size of 199, and the wavelength values ​​at each trough position were extracted; for example... Figure 4 It is evident that the dispersion trend of seawater differs slightly from that of pure water. Therefore, to obtain accurate refractive index data for seawater, the dispersion relation of seawater was used to calculate the Abbe refractometer measurements to the infrared band. Then, the refractive index was fitted to multiple wavelength troughs to obtain the refractive index sensing sensitivity of the sensor at each trough. Figure 5 visible.

[0046] Next, a series of temperature, salinity, and pressure sensing experiments were conducted on the partially encapsulated micro-fiber-coaxial Mach-Zehnder interferometer sensor. The partially encapsulated micro-fiber-coaxial Mach-Zehnder interferometer sensor was placed in a pressure tank filled with seawater. An external heating device was used to change the temperature of the seawater in the pressure tank, and distilled water or a high-concentration sodium chloride solution was added to change the salinity. An external pressure pump was used to pressurize / depressurize the seawater inside the tank. Figure 6 As can be seen, simultaneously, broadband light sources and spectrometers were used to collect spectral signals under different temperature, salinity, and pressure (TSP) environments. A commercially available temperature, salinity, and depth (CTD) meter was placed inside the pressure tank to synchronously record the TSP data of the seawater environment within the tank. A total of 539 spectra under different TSP conditions and their corresponding TSP values ​​were obtained, of which 8 sets of spectra and TSP values ​​are shown below. Figure 7 visible.

[0047] Subsequently, based on the refractive index sensing sensitivity and the trough wavelength values ​​of the random temperature, salinity, and pressure (TDP) sensing spectral signals, the seawater refractive index corresponding to each spectral signal was calculated. The seawater refractive index, the TDP-measured temperature, salinity, and pressure corresponding to the spectrum, and the trough wavelength were combined to form a data set, thereby constructing a database containing 3395 sets of seawater refractive index, temperature, salinity, pressure, and wavelength. Using the existing formula for seawater refractive index relationships, the coefficients were set as unknowns, and various algorithms were used to fit the data to obtain correlation coefficients. Based on the fitting parameters... R 2 By selecting parameter values ​​fitted using the nonlinear least squares method, a formula for the refractive index of seawater applicable to the infrared band (1200nm-1650nm) is derived. R 2 =0.9044), the formula for the seawater refractive index is shown in [reference needed]. Figure 4 The formula for the refractive index of seawater applicable to the infrared band is derived as follows:

[0048] ;

[0049] in, n Represents the refractive index of seawater. T Represents seawater temperature, measured in °C; S Represents seawater salinity, measured in ‰; P Represents seawater pressure, measured in MPa. λ The wavelength represents the detection wavelength, measured in nm. Next, a random temperature, salinity, and pressure (THP) sensing experiment was conducted again on the locally encapsulated micro-fiber coaxial Mach-Zehnder interferometer. Multiple transmission spectra under random THP environments were collected, and multiple trough wavelength values ​​were extracted to obtain a dataset of THP and trough wavelength values. Using this THP dataset, 80% of the data was used for machine learning model training, and 20% was used for model prediction validation. Support vector machines (SVM), random forests, and neural networks were used to train and predict on the same data. The demodulation results were evaluated using three errors: mean absolute error (MAE), root mean square error (RMSE), and mean absolute percentage error relative to full scale (MAPE). The temperature demodulation result from the SVM was selected for subsequent salinity demodulation. The THP demodulation errors of the three algorithms are shown in Table 1.

[0050] Table 1. Temperature-salinity-pressure demodulation errors of the three algorithms.

[0051]

[0052] Next, salinity and pressure are demodulated using a variable matrix. First, the partial derivatives of the refractive index with respect to salinity and pressure are obtained using the modified formula for the refractive index of seawater in the infrared band. Then, using measured refractive index sensitivities at different wavelengths, data fitting is performed on the center wavelength and refractive index sensitivity parameters to obtain a functional relationship between the refractive index sensitivity and wavelength. Multiplying the partial derivatives with the refractive index sensitivity function yields the salinity sensitivity function and the pressure sensitivity function, which vary with external environmental temperature, salinity, and pressure conditions, respectively. Specifically:

[0053] ;

[0054] ;

[0055] ;

[0056] ;

[0057] in, S S1 and S S2 These represent the salinity sensitivity of the two selected troughs, respectively. λ 1 and λ 2 represents the wavelength values ​​of the two selected troughs; S P1 and S P2 These represent the pressure sensitivity of the two selected troughs.

[0058] The above four sensitivity functions are used as the four variable elements of the second-order sensitivity matrix to establish a second-order variable element sensitivity matrix. S Variable Specifically:

[0059] ;

[0060] Then, a matrix equation containing the sensitivity matrix of the aforementioned variables is established for subsequent demodulation of the salinity and pressure parameters. The matrix equation is as follows:

[0061] ;

[0062] in, S 0, P 0 represents the initial values ​​for salinity and pressure, respectively. λ 10 , λ 20 These are the initial wavelength values ​​for the selected troughs.

[0063] Finally, using the temperature demodulation results from machine learning, the wavelength shift of the valley caused by temperature changes in the spectrum to be demodulated was removed by using the measured temperature sensitivity, and the new valley wavelength value was obtained. Two valleys were randomly selected, and the seawater salinity and pressure values ​​were demodulated using the sensitivity matrix and matrix equation established above. A total of 53 sets of data were demodulated, and some of the values ​​are shown in Table 2.

[0064] Table 2. Partial values ​​of seawater salinity and pressure obtained from demodulation.

[0065]

[0066] To evaluate the demodulation accuracy and precision, the temperature, salinity, and pressure values ​​demodulated by the proposed method were compared with the CTD measurements. It can be seen that the MAPE values ​​for temperature, salinity, and pressure demodulated using a single machine learning algorithm were 5.64%, 5.53%, and 28.89%, respectively, while the MAPE values ​​for temperature, salinity, and pressure demodulated using the proposed method were 5.64%, 3.66%, and 9.25%, respectively. Among these, the salinity error was reduced by 33.8%, and the pressure error was reduced by 68.0%.

[0067] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

[0068] While the specific embodiments of the present invention have been described above, they are not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art without creative effort based on the technical solutions of the present invention are still within the scope of protection of the present invention.

Claims

1. A multi-method collaborative high-precision demodulation method for seawater temperature, salinity, and pressure parameters, characterized in that, The process includes the following: S1. The refractive index of seawater is sensed and measured using a partially encapsulated micro-fiber coaxial Mach-Zehnder interferometer sensor. The refractive index of seawater in the infrared band is calculated based on the seawater dispersion relation, and the seawater refractive index sensitivity of the sensor is obtained by fitting it with the wavelength. S2 utilizes a partially encapsulated micro-fiber coaxial Mach-Zehnder interferometer sensor to sense and measure seawater temperature, salinity, and pressure. Based on the seawater refractive index sensitivity obtained in S1, the refractive index values ​​under different temperature, salinity, and pressure conditions are calculated, and a database of seawater refractive index, temperature, salinity, pressure, and detection wavelength is established. S3, based on the database established by S2, uses a machine learning fitting algorithm to obtain a corrected expression for the refractive index of seawater applicable to the infrared band. S4. Establish a dataset of temperature, salinity, pressure and detection wavelength, train and predict the machine learning regression prediction model, and achieve demodulation of any single parameter among the three parameters of temperature, salinity and pressure. S5, using the corrected seawater refractive index expression from S3 and the measured seawater refractive index sensing sensitivity, establishes a dual-parameter variable element sensitivity matrix equation that varies with external environmental parameters; the specific implementation process is as follows: By fitting the corrected infrared band seawater refractive index expression using S3, the partial derivative functions of the refractive index with respect to salinity and pressure are obtained. Using the measured refractive index sensitivity at different wavelengths, data fitting is performed on the center wavelength and refractive index sensitivity to obtain the functional relationship of refractive index sensitivity with respect to wavelength. Multiplying the partial derivative function with the refractive index sensitivity function, the salinity sensitivity function and pressure sensitivity function under different temperature, salinity, and pressure conditions are obtained as the four variable matrix elements of the second-order sensitivity matrix, and then a dual-parameter variable element sensitivity matrix and matrix equation are established. S6, based on the demodulation results of the single parameter in machine learning in S4, combines the measured sensitivity to evaluate and eliminate the influence of the parameter change on the peak shift; the specific implementation process is as follows: Based on the temperature demodulation results, the peak shift caused by temperature changes is eliminated using the measured temperature sensitivity to obtain the new trough wavelength value: ； in, λ 0 represents the original wavelength of the trough, in nm. S T Temperature sensitivity of the trough, in nm / ℃. This refers to the change in temperature. S7 uses the dual-parameter variable element sensitivity matrix equation to demodulate the remaining two parameters based on the change in trough wavelength.

2. The multi-method collaborative high-precision demodulation method for seawater temperature, salinity, and pressure parameters as described in claim 1, characterized in that, The specific implementation process of S1 is as follows: A testing system was constructed, comprising a broadband light source, a spectrometer, and a micro-fiber sensor. The refractive index of seawater was altered by gradually diluting it with distilled water. Transmission spectra of a locally encapsulated micro-fiber coaxial Mach-Zehnder interferometer sensor at different seawater refractive indices were collected using the broadband light source and spectrometer, and the corresponding refractive indices were measured using an Abbe refractometer. The collected spectral signals were smoothed, and the wavelength values ​​of multiple troughs were extracted. The refractive index of seawater at different wavelengths was measured using an infrared refractometer, and the seawater dispersion relation was obtained by fitting the data. ； in n The refractive index of seawater, λ The wavelength is measured in nm. The refractive index of seawater measured by the Abbe refractometer is calculated to correspond to the infrared band using the seawater dispersion relation, and the seawater refractive index sensitivity of the sensor is calculated.

3. The multi-method collaborative high-precision demodulation method for seawater temperature, salinity, and pressure parameters as described in claim 1, characterized in that, The specific implementation process of S2 is as follows: A testing system was constructed, comprising a broadband light source, a spectrometer, a micro-fiber sensor, and a pressure vessel. A partially encapsulated micro-fiber coaxial Mach-Zehnder interferometer temperature-salinity-pressure (TSP) sensor was placed in a pressure vessel filled with seawater. An external heating device was used to change the seawater temperature, and distilled water or a high-concentration sodium chloride solution was added to change the salinity. An external pressure pump was used to pressurize or depressurize the seawater. Broadband light source and spectrometer collected spectral signals under different TSP environments. Simultaneously, a temperature-salinity-depth (TSD) meter was placed inside the pressure vessel to record the TSP values ​​of the test environment. The seawater refractive index value corresponding to the random TSP spectral signal was calculated based on the seawater refractive index sensor sensitivity. This seawater refractive index value was combined with the TSP values ​​measured by the TSD meter and the trough wavelength values ​​to form a data set, which was then used to construct a database.

4. The multi-method collaborative high-precision demodulation method for seawater temperature, salinity, and pressure parameters as described in claim 1, characterized in that, The specific implementation process of S3 is as follows: The database data established in S2 was fitted using the nonlinear least squares method of machine learning with five parameters, including seawater refractive index, temperature, salinity, pressure, and detection wavelength, to derive a seawater refractive index formula suitable for the infrared band: ； in, n Represents the refractive index of seawater. T Represents seawater temperature, measured in °C; S Represents seawater salinity, measured in ‰; P This represents seawater pressure, measured in MPa. λ This represents the detection wavelength, measured in nm.

5. The multi-method collaborative high-precision demodulation method for seawater temperature, salinity, and pressure parameters as described in claim 1, characterized in that, In step S4, a support vector machine is used as the regression prediction model, and the temperature parameter is selected as the demodulation result.

6. The multi-method collaborative high-precision demodulation method for seawater temperature, salinity, and pressure parameters as described in claim 1, characterized in that, The specific implementation process of S7 is as follows: By selecting any two troughs in the spectrum and using the established sensitivity matrix containing variable elements, the salinity and pressure values ​​of seawater can be obtained by solving the matrix equation.

Images