A method and system for real-time measurement of dynamic pressure in seawater

Abstract

The application relates to a seawater dynamic pressure real-time measurement method and system and relates to the technical field of marine pressure measurement.The application comprises the following steps: improving a measurement probe based on a Michelson interferometer, collecting sample signals under different temperature pressures, analyzing the frequency difference between an interference peak and an absorption peak, combining temperature pressure data, and constructing a data set.The application combines a neural network architecture and physical guidance to establish a Gaussian process regression model as a pressure calculation model, adopts Gaussian process regression of a composite kernel function, trains the model by using the data set, makes the model generate a frequency difference response curve caused only by pressure under the condition of a fixed temperature reference value, combines partial derivative analysis to quantize pressure sensitivity, and ensures the accuracy and reliability of model decoupling results.In actual application, total pressure signals and static pressure signals of a sea area collected by a dynamic pressure measurement probe are subjected to small continuous dynamic pressure identification through the pressure calculation model.

Description

Technical Field

[0001] This application relates to the field of marine pressure measurement technology, and in particular to a method and system for real-time measurement of dynamic pressure in seawater. Background Technology

[0002] In harsh marine environments, seawater pressure measurements typically still rely on traditional dynamic pressure measurement techniques, such as pitot tubes and towed force methods. These methods inherently have low frequencies and insufficient frequency response, making it impossible to accurately capture the minute, transient, continuous dynamic pressure signals in seawater, thus failing to meet the demands of high-precision measurements. Furthermore, in the face of harsh conditions such as high salinity, high pressure, and strong currents in the marine environment, traditional sensors are susceptible to physical impacts and electromagnetic interference, leading to significant decreases in measurement accuracy, data drift, and even equipment failure.

[0003] It is evident that current technologies are not yet capable of achieving high-precision measurement of minute, continuous dynamic pressures in seawater. Summary of the Invention

[0004] This application provides a method and system for real-time dynamic pressure measurement in seawater, and establishes a high-precision measurement scheme for pressure that is small, continuous and dynamically changing in seawater. This includes, but is not limited to, introducing a high-precision pressure measurement structure based on the Michelson interferometry and combining it with a neural network model to achieve high-precision pressure measurement based on temperature correction.

[0005] In a first aspect, this application provides a method for real-time measurement of dynamic pressure in seawater, including: Under preset temperature and pressure, the laser beam generated by the light source module is split, and the measurement sample signal output by the dynamic pressure measurement probe and the standard sample signal output by the calibration optical path are collected. The dynamic pressure measurement probe is a measurement probe based on Michelson interferometry. The frequency difference between interference peaks and absorption peaks was analyzed based on standard sample signals and measured sample signals, and a dataset was constructed by combining temperature and pressure data. A physical-guided Gaussian process regression model based on a neural network architecture is constructed as a stress calculation model. Gaussian process regression and hyperparameter optimization are introduced, and a stress calculation model is trained using a dataset until the model training is complete. Frequency difference response curves and stress sensitivity are generated in the model. Acquire measurement signals from the target sea area collected by the dynamic pressure measurement probe. The measurement signals include total pressure signals and static pressure signals. Based on the total pressure signal and static pressure signal, a pressure calculation model is used to identify small, continuous dynamic pressure and obtain dynamic pressure measurement results.

[0006] Optionally, under preset temperature and pressure conditions, the laser beam generated by the light source module is split to collect the measurement sample signal output by the dynamic pressure measurement probe and the standard sample signal output by the calibration optical path, including: For marine environments, under preset temperature and pressure conditions, the laser beam generated by the beam splitter module is divided into a first beam and a second beam. The first beam is output to the dynamic pressure measurement probe through the measurement optical path, and the total pressure sample signal and static pressure sample signal output by the dynamic pressure measurement probe are collected as measurement sample signals. The second beam is output to the calibration optical path to collect standard sample signals.

[0007] Optionally, based on the analysis of the frequency difference between the interference peak and the absorption peak using standard sample signals and measured sample signals, and combined with temperature and pressure data, a dataset is constructed, including: Interference peaks are calculated based on measured sample signals, and absorption peaks are analyzed by compensating for laser frequency fluctuations based on standard sample signals. Based on interference peaks and absorption peaks, combined with measurement parameters, the frequency difference is analyzed; By using pressure samples from the reflective membrane in the acquired dynamic pressure measurement probe, the frequency difference is updated, and a dataset is constructed by combining temperature and pressure data.

[0008] Optionally, based on interference and absorption peaks, and in conjunction with measurement parameters, frequency differences can be analyzed, including: Based on the time interval between the recorded interference peaks and absorption peaks, the time delay is determined, and the frequency tuning range of the laser is obtained. Analyze the initial frequency difference based on interference peaks and absorption peaks; Using time delay and tuning range as measurement parameters, and combining them with the initial frequency difference, the frequency difference is calculated within the corresponding scanning period.

[0009] Optionally, the frequency difference is updated using pressure samples from the reflective membrane in the acquired dynamic pressure measurement probe, and combined with temperature and pressure data to construct a dataset, including: The first pressure of the first reflective film and the second pressure of the second reflective film in the dynamic pressure measurement probe are collected respectively as pressure samples. The second pressure is the pressure obtained after applying dynamic pressure on the basis of the first pressure. Under pressure samples, the update frequency differs; Integrate temperature and pressure data with frequency differences to construct a dataset.

[0010] Optionally, Gaussian process regression and hyperparameter optimization are introduced to train a stress calculation model using the dataset until the model training is complete. Frequency difference response curves and stress sensitivity are then generated within the model, including: Feature engineering and standardization preprocessing are performed on the dataset to obtain feature samples, which are then input into the stress calculation model. In the stress calculation model, Gaussian process regression with composite kernel function and initial parameters are used for training. The performance of the model on different datasets is evaluated, and the performance data is output. Based on the analysis of residual distribution and prediction uncertainty using performance data, hyperparameter optimization is performed using marginal likelihood maximization to obtain a well-trained stress calculation model. Based on the trained pressure calculation model, a frequency difference response curve caused by pressure is generated under preset temperature conditions, and the pressure sensitivity is quantified by partial derivative analysis. Based on the frequency difference response curve and pressure sensitivity, the decoupling results are determined through three-dimensional response surface visualization and physical consistency verification. The decoupling results are used to update the pressure calculation model.

[0011] Optionally, before acquiring the measurement sample signal output by the dynamic pressure measurement probe, the following steps are also included: In the multiphysics simulation software, a simulated dynamic pressure measurement probe is constructed. The simulated dynamic pressure measurement probe includes two simulated reflective films corresponding to the optical paths, and the film parameters of each simulated reflective film are the initial parameters. Physical simulations are performed within a preset temperature and pressure variation range. Based on the acquired simulation signals, the simulated optical path difference and simulated interference peaks between optical paths are analyzed, and the diaphragm displacement is calculated and adjusted. The initial parameters are updated based on the diaphragm displacement, and the simulated dynamic pressure measurement probe is adjusted synchronously.

[0012] Optionally, based on the acquired analog signal analysis of the simulated optical path difference and simulated interference peaks between optical paths, the diaphragm displacement is calculated and adjusted, including: Based on the acquired analog signal, the diaphragm displacement is analyzed from the initial parameters; Based on the change in diaphragm displacement, the initial optical path difference between the two optical paths is analyzed; Based on the initial optical path difference and the refractive index of air, analyze the target optical path difference; Based on the target optical path difference and the simulated incident wavelength, the simulated interference peak is calculated; The diaphragm displacement is adjusted based on the simulated interference peak until the diaphragm displacement is the same as the incident wavelength or the error between the diaphragm displacement and the incident wavelength is lower than the preset value.

[0013] Secondly, this application provides a real-time dynamic pressure measurement system in seawater, including a light source module, a measurement module, a data acquisition module, and a data processing module; The light source module is used to generate a laser beam; The measurement module is used to split the generated laser beam and output the measurement sample signal from the dynamic pressure measurement probe and the standard sample signal from the calibration optical path. The dynamic pressure measurement probe is a measurement probe based on Michelson interferometry. The data acquisition module is used to acquire measurement sample signals and standard sample signals; as well as measurement signals from the target sea area, including total pressure signals and hydrostatic pressure signals. The data processing module is used to analyze the frequency difference between interference peaks and absorption peaks based on standard sample signals and measured sample signals, and to construct a dataset by combining temperature and pressure data. It adopts a neural network architecture to construct a physics-guided Gaussian process regression model as a pressure calculation model. It introduces Gaussian process regression and hyperparameter optimization, and uses the dataset to train the pressure calculation model until the model training is complete, generating frequency difference response curves and pressure sensitivity in the model. It is also used to identify small, continuous dynamic pressures based on total pressure signals and static pressure signals using the pressure calculation model, and to obtain dynamic pressure measurement results.

[0014] Optionally, in the dynamic pressure measurement probe, the two optical paths are arranged in a vertical structure to generate the total pressure signal and the static pressure signal, respectively.

[0015] In summary, this application's embodiments are based on an improved Michelson interferometer measurement probe. The probe collects measurement sample signals and standard sample signals from seawater under different temperatures and pressures. It analyzes the frequency difference between interference peaks and absorption peaks using these sample signals, and constructs a frequency difference-temperature-pressure dataset based on this frequency difference and temperature and pressure data. Then, guided by physics and combined with a neural network, a Gaussian process regression model is established as the pressure calculation model. The model is trained using the dataset, employing Gaussian process regression with a composite kernel function during training. Hyperparameter optimization is used to optimize the model parameters, enabling it to generate a frequency difference response curve solely caused by pressure under a fixed temperature reference value. Partial derivative analysis is used to quantify pressure sensitivity, ensuring the accuracy and reliability of the model's decoupling results. Finally, in practical applications, the two optical paths of the dynamic pressure measurement probe are arranged perpendicularly. When an object passes the probe, the total pressure changes slightly, while the static pressure remains constant. Based on this principle, the total pressure and static pressure signals of the sea area are collected, and the pressure calculation model is used to identify minute, continuous dynamic pressure. As can be seen, this application addresses the high-precision measurement of minute, continuous, and dynamically changing pressure in seawater. By designing a high-precision pressure measurement structure based on the Michelson interferometry and combining it with a neural network model, a high-precision pressure measurement scheme based on temperature correction is realized, enabling high-precision measurement of minute, continuous, and dynamically changing pressure in seawater. Attached Figure Description

[0016] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0017] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 A flowchart illustrating a method for real-time measurement of dynamic pressure in seawater, provided as an embodiment of this application; Figure 2 This is a flowchart illustrating the process of establishing a pressure calculation model, as provided in this application. Figure 3 This is a schematic diagram of a measurement probe structure based on Michelson interferometry provided as an example in this application; Figure 4 This is a schematic diagram illustrating an application scenario provided by this application; Figure 5 This is a structural diagram of a real-time dynamic pressure measurement system in seawater provided in an embodiment of this application. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0020] To facilitate understanding of the embodiments of this application, further explanations and descriptions will be provided below in conjunction with the accompanying drawings and specific embodiments. These embodiments do not constitute a limitation on the embodiments of this application.

[0021] Figure 1 The flowchart illustrates a method for real-time measurement of dynamic pressure in seawater, as provided in this application embodiment. The method may specifically include the following steps: Step 110: Under preset temperature and pressure, split the laser beam generated by the light source module, and collect the measurement sample signal output by the dynamic pressure measurement probe and the standard sample signal output by the calibration optical path.

[0022] Among them, the dynamic pressure measurement probe is a measurement probe based on Michelson interferometry.

[0023] In this embodiment, a dynamic pressure measurement probe (hereinafter referred to as the measurement probe) is designed based on an improved Michelson interferometer probe. This embodiment proposes a method for real-time measurement of dynamic pressure in seawater by combining a neural network model. Before applying the measurement method to the actual marine environment, samples can be collected and the model trained.

[0024] Reference Figure 2 As shown, specifically, during sample collection, the system can be placed in a seawater sample for signal sample acquisition. First, different temperature and pressure environments are established. At each temperature and pressure, the light source module generates a laser beam. By splitting the laser beam, two laser beams are obtained, which can be output to the measurement probe and the calibration optical path respectively. One beam is processed by the two optical paths in the measurement probe, outputting the corresponding static pressure sample signal and total pressure sample signal at different pressures, forming the measurement sample signal; the other beam passes through the calibration optical path and serves as the wavelength standard.

[0025] Optionally, this embodiment can split the laser beam generated by the light source module under preset temperature and pressure conditions to collect the measurement sample signal output by the dynamic pressure measurement probe and the standard sample signal output by the calibration optical path. Specifically, it can include: for marine environments, splitting the laser beam generated by the light source module under preset temperature and pressure conditions to obtain a first beam and a second beam; outputting the first beam to the dynamic pressure measurement probe through the measurement optical path to collect the total pressure sample signal and static pressure sample signal output by the dynamic pressure measurement probe as measurement sample signals; and outputting the second beam to the calibration optical path to collect the standard sample signal.

[0026] In this embodiment, the light beam generated by the light source module can be split by a beam splitter. The resulting first light beam is output to the measurement probe through the measurement optical path. The two optical paths of the measurement probe split the first light beam into two paths. The reflective films of the two optical paths reflect the light beams of the corresponding optical paths, generating total pressure sample signals and static pressure sample signals.

[0027] In the calibration optical path, the second beam passes through an adjustable attenuator and then through an optical fiber coupled hydrogen fluoride (HF) gas absorption cell (absorption peak wavelength 1260.7417 nm) as a wavelength standard to obtain a standard signal sample. This wavelength standard can compensate for laser frequency fluctuations caused by system drift and environmental noise in real time.

[0028] In practical implementation, one of the innovations of this application lies in improving the measurement probe based on Michelson interferometry. Specifically, before constructing the actual device, improvement tests can be conducted using simulation software to adjust the parameters of each structure in the measurement probe. Optionally, before acquiring the measurement sample signal output by the dynamic pressure measurement probe, the following steps are also included: constructing a simulated dynamic pressure measurement probe in multiphysics simulation software. The simulated dynamic pressure measurement probe includes simulated reflective films corresponding to two optical paths, with the film parameters of each simulated reflective film being the initial parameters; performing physical simulations under preset temperature and pressure variation ranges; analyzing the simulated optical path difference and simulated interference peaks between the optical paths based on the acquired simulated signals; calculating and adjusting the film displacement; updating the initial parameters according to the film displacement; and synchronously adjusting the simulated dynamic pressure measurement probe.

[0029] Reference Figure 3 The measurement probe structure shown has two perpendicular optical paths for measuring static pressure and total pressure, respectively. The beam is split by a beam splitter and output to reflective film 1 and reflective film 2. After being reflected by the reflective films, the measurement signal is acquired by the system's data acquisition module.

[0030] In practical implementation, a measurement probe, including a stainless steel diaphragm, can be designed in multiphysics simulation software (COMSO). Its initial parameters are a diameter of D mm and a thickness of C mm. The changes of the diaphragm under pressures of 0-35℃ and 0-100MPa can be calculated respectively.

[0031] In the simulation environment, the acquired simulated signals also consist of total pressure and static pressure signals. These simulated signals are used to calculate the simulated optical path difference. The simulated interference peak is then calculated using this simulated optical path difference, which is used to adjust the diaphragm's displacement. When the diaphragm's displacement meets certain conditions, the diaphragm's diameter and thickness are determined. The simulated measurement probe undergoes simultaneous structural adjustments, thus determining the various structures and parameters within the actual measurement probe system, and designing a Michelson interferometer-based measurement probe.

[0032] In one optional embodiment, the simulated optical path difference and simulated interference peak between the optical paths are analyzed based on the acquired simulated signal, and the diaphragm displacement is calculated and adjusted. Specifically, this may include: using the acquired simulated signal as a reference, analyzing the diaphragm displacement from the initial parameters; analyzing the initial optical path difference between the two optical paths based on the diaphragm displacement; analyzing the target optical path difference based on the initial optical path difference and the refractive index of air; calculating the simulated interference peak based on the target optical path difference and the simulated incident wavelength; and adjusting the diaphragm displacement based on the simulated interference peak until the diaphragm displacement is the same as the incident wavelength or the error between the diaphragm displacement and the incident wavelength is lower than a preset value.

[0033] In practical implementation, to accurately obtain the optical path difference and interference peak, this embodiment introduces the following formulas for calculation: For optical path difference : Determine the optical path length in the air for each of the two paths, i.e. Figure 3 shown and The refractive index of air. That is, ①.

[0034] For interference peaks : optical path difference Based on this, the wavelength of the incident light is introduced. ,have ②.

[0035] When the diaphragm displacement is within the set temperature and pressure variation range, When left and right, determine the diameter and thickness of the diaphragm.

[0036] Furthermore, during sample collection, the measuring probe can be placed in a constant-temperature water bath, with the external temperature controlled by the bath. The stainless steel reflective film in the reference optical path maintains the same temperature as the external environment through thermal equilibrium of its outer wall, thus eliminating the additional effects of temperature on the two films. When the external seawater is subjected to changes in temperature and pressure, the interference peak of the measuring optical path shifts. By calculating the change in frequency difference between the reference and measuring optical paths, the sample can be obtained and used to subsequently establish calculation models for temperature, frequency difference, and pressure.

[0037] Step 120: Analyze the frequency difference between the interference peak and the absorption peak based on the standard sample signal and the measured sample signal, and construct a dataset by combining temperature and pressure data.

[0038] In this embodiment, during the sample acquisition stage, the interference peak is calculated using the measured sample signal according to the aforementioned formula ②; the absorption peak is calculated using the standard sample signal. Then, the frequency difference between the interference peak and the absorption peak is analyzed. Since the samples are collected under different temperatures and pressures, a dataset can be established based on the frequency difference and the corresponding temperature and pressure data.

[0039] Optionally, the above-mentioned analysis of the frequency difference between interference peaks and absorption peaks based on standard sample signals and measured sample signals, combined with temperature and pressure data, to construct a dataset may include the following sub-steps: Sub-step 1201 involves calculating the interference peak based on the measured sample signal, and performing laser frequency fluctuation compensation based on the standard sample signal to analyze the absorption peak.

[0040] Sub-step 1202: Based on the interference peak and absorption peak, and combined with the measurement parameters, analyze the frequency difference.

[0041] A unified explanation is provided for sub-steps 1202-1203: The measurement parameters include time delay and laser frequency tuning range.

[0042] In practical applications, when measuring changes in seawater temperature or pressure, the diaphragm deforms, causing a shift in the measured interference peaks. Therefore, calculating the seawater pressure at different temperatures by measuring the frequency difference and accurately determining the frequency difference between the interference peaks in the measurement optical path and the gas absorption peaks is crucial for demodulating the signal.

[0043] The time delay determines the frequency difference between the measured interference peak and the HF absorption peak. In this embodiment, the time delay and other measurement parameters are used as one of the inputs. First, the initial frequency difference between the interference peak and the absorption peak is calculated. Then, combined with the measurement parameters, the frequency difference is calculated within the scanning period.

[0044] In practical implementation, the above-mentioned analysis of frequency difference based on interference peaks and absorption peaks, combined with measurement parameters, may include: determining the time delay based on the recorded time interval between interference peaks and absorption peaks, and obtaining the laser frequency tuning range; analyzing the initial frequency difference based on interference peaks and absorption peaks; and using the time delay and tuning range as measurement parameters, combined with the initial frequency difference, to calculate the frequency difference in the corresponding scanning cycle.

[0045] In practice, the system can be controlled by LabVIEW software on a host computer, using a current driver to adjust the laser's drive current for frequency sweeping. By processing the signals collected from the seawater sample in the measurement optical path, the software can record various measurement parameters and periods for calculating the frequency difference.

[0046] For frequency difference ,have: During the scan cycle In the middle, the initial frequency difference between the interference peak and the absorption peak Laser frequency tuning range Delay For input, according to Calculate the frequency difference.

[0047] Sub-step 1203: Using the pressure sample of the reflective membrane in the collected dynamic pressure measurement probe, update the frequency difference, and combine it with temperature and pressure data to construct a dataset.

[0048] Reference Figure 3 As shown, in this embodiment, piston pressure gauges are connected to reflective film 2 and reflective film 1 respectively, and pressure is applied to reflective film 1. ( (Range range is 0-100MPa), applied at 2 points on the reflective film. ( The range is 0-1 MPa). At different static pressures... Next, the frequency difference is calculated, and by combining the temperature data and pressure data, a temperature-pressure-frequency difference dataset can be constructed for subsequent model training.

[0049] In practical implementation, the above-mentioned use of pressure samples from the reflective membrane in the dynamic pressure measurement probe to update the frequency difference, combined with temperature and pressure data, to construct a dataset can specifically include: collecting the first pressure of the first reflective membrane and the second pressure of the second reflective membrane in the dynamic pressure measurement probe as pressure samples, wherein the second pressure is the pressure obtained after applying dynamic pressure on the basis of the first pressure; updating the frequency difference under the pressure samples; and integrating the temperature and pressure data and the frequency difference to construct a dataset.

[0050] Understandable, refer to Figure 3 As shown, assuming the first reflective film is reflective film 1 and the second reflective film is reflective film 2, then the first pressure is the total pressure and the second pressure is the static pressure.

[0051] Step 130: Using a neural network architecture, a physical-guided Gaussian process regression model is constructed as the stress calculation model.

[0052] Reference Figure 2 As shown, this embodiment constructs a pressure calculation model based on the Gaussian process regression (GPR) model using a physics-guided neural network. This pressure calculation model is a pressure-temperature-frequency difference response model established using the Gaussian process regression method, hereinafter referred to as the model.

[0053] Step 140 introduces Gaussian process regression and hyperparameter optimization, and trains the stress calculation model using the dataset until the model training is complete, generating frequency difference response curves and stress sensitivity in the model.

[0054] Reference Figure 2 As shown, before training, the dataset can be preprocessed uniformly, such as feature processing and dividing the dataset into training and validation sets. The model is then trained using this dataset. During training, Gaussian process regression with a composite kernel function is used, and hyperparameter optimization is employed to improve model accuracy until training is complete. After training, a frequency difference response curve caused solely by pressure can be generated under a fixed temperature reference value, and partial derivative analysis can be used to quantify pressure sensitivity.

[0055] In one optional embodiment, Gaussian process regression and hyperparameter optimization are introduced to train a stress calculation model using a dataset until the model training is complete. Frequency difference response curves and stress sensitivity are generated within the model. Specifically, this may include: performing feature engineering and standardization preprocessing on the dataset to obtain feature samples, which are then input into the stress calculation model; training the stress calculation model using Gaussian process regression with a composite kernel function and initial parameters, evaluating the model's performance on different datasets, and outputting performance data; analyzing residual distribution and prediction uncertainty based on the performance data, and using marginal likelihood maximization for hyperparameter optimization to obtain a trained stress calculation model; generating a frequency difference response curve caused by pressure under preset temperature conditions based on the trained stress calculation model, and quantifying stress sensitivity using partial derivative analysis; and determining the decoupling result based on the frequency difference response curve and stress sensitivity through three-dimensional response surface visualization and physical consistency verification, with the decoupling result used to update the stress calculation model.

[0056] In this embodiment, the dataset is systematically processed, integrating temperature-pressure data with measurement frequency differences, and the data is preprocessed using feature engineering and standardization. Feature engineering is used to extract feature samples, and standardization is used to integrate data from different dimensions.

[0057] The data is input into the model. For model training, a composite kernel function is used to train the basic GPR model with default hyperparameters. The model's performance on the training and validation sets is evaluated, the residual distribution and prediction uncertainty are analyzed, and optimization is performed based on the predicted data output by the model, such as using marginal likelihood maximization for hyperparameter optimization, until the model training is complete.

[0058] Based on the trained model, a frequency difference response curve induced solely by pressure is generated under a fixed temperature reference value. Partial derivative analysis is then used to quantify pressure sensitivity. Finally, the accuracy and reliability of the decoupling results are ensured through 3D response surface visualization and physical consistency verification. The model can be further updated and optimized based on the decoupling results.

[0059] Therefore, this embodiment first designs the measurement probe structure and conducts testing under different pressures and temperatures. Then, based on the change in optical path difference at different temperatures, it studies the measurement data under different pressure conditions and establishes an inversion model of seawater pressure, selecting an appropriate number of measurement points to ensure resolution. An innovative dynamic pressure correction algorithm capable of operating at different temperatures is proposed and applied for the first time to a high-precision dynamic pressure measurement method in seawater.

[0060] Step 150: Acquire measurement signals from the target sea area collected by the dynamic pressure measurement probe. The measurement signals include total pressure signal and static pressure signal.

[0061] Step 160: Based on the total pressure signal and static pressure signal, use the pressure calculation model to identify the small, continuous dynamic pressure and obtain the dynamic pressure measurement results.

[0062] A unified explanation of steps 150-160 is provided below: Reference Figure 4 As shown, taking a submarine as an example, the dynamic pressure measurement probe is placed below the target sea area (e.g., underwater). One reflective membrane faces the object to be measured, used to measure the total pressure and collect the total pressure signal; the other reflective membrane is perpendicular to the object to be measured, used to measure the static pressure and collect the static pressure signal. When the submarine passes the measurement probe, the pressure at the total pressure measuring end changes slightly, while the pressure at the static pressure measuring end remains unchanged. The collected signals are input into the model, and the model calculates the minute dynamic pressure, which is taken as the dynamic pressure measurement result.

[0063] As can be seen, this application's embodiment addresses the high-precision measurement of minute, continuous, and dynamically changing pressure in seawater. First, a high-precision pressure measurement structure based on the Michelson interferometry is designed, with two perpendicular optical paths in the measurement probe to collect total pressure and static pressure signals respectively. Based on the improved probe structure, a neural network model is introduced. Specifically, the measurement probe collects measurement sample signals and standard sample signals from seawater under different temperatures and pressures. The frequency difference between interference peaks and absorption peaks is analyzed using these sample signals. Based on this frequency difference and combined with temperature and pressure data, a frequency difference-temperature-pressure dataset is constructed. Subsequently, a Gaussian process regression model is established using a neural network under physical guidance as the pressure calculation model. The model is trained using the dataset, employing Gaussian process regression with a composite kernel function during training. Hyperparameter optimization is used to optimize the model parameters, enabling the model to generate a frequency difference response curve caused solely by pressure under a fixed temperature reference value. Partial derivative analysis is then used to quantify pressure sensitivity, ensuring the accuracy and reliability of the model's decoupling results. Finally, in practical applications, the two optical paths of the dynamic pressure measurement probe are arranged in a perpendicular structure. When an object passes through the probe, the total pressure changes slightly, while the static pressure remains constant. Based on this principle, the total pressure and static pressure signals of the sea area are collected, and minute, continuous dynamic pressure is identified through a pressure calculation model. Thus, this embodiment achieves high-precision measurement of minute, continuous dynamic pressure in seawater, solving the technical problems of low inherent frequency, insufficient frequency response, and susceptibility to physical shocks and electromagnetic interference in existing seawater pressure measurement technologies.

[0064] It should be noted that, for the sake of simplicity, the method embodiments are described as a series of actions. However, those skilled in the art should know that the embodiments of this application are not limited to the described order of actions, because according to the embodiments of this application, some steps may be performed in other orders or simultaneously.

[0065] like Figure 5 As shown in the illustration, this application also provides a real-time dynamic pressure measurement system in seawater, comprising: It includes a light source module, a measurement module, a data acquisition module, and a data processing module; The light source module is used to generate a laser beam; The measurement module is used to split the generated laser beam and output the measurement sample signal from the dynamic pressure measurement probe and the standard sample signal from the calibration optical path. The dynamic pressure measurement probe is a measurement probe based on Michelson interferometry. The data acquisition module is used to acquire measurement sample signals and standard sample signals; as well as measurement signals from the target sea area, including total pressure signals and hydrostatic pressure signals. The data processing module is used to analyze the frequency difference between interference peaks and absorption peaks based on standard sample signals and measured sample signals, and to construct a dataset by combining temperature and pressure data. It adopts a neural network architecture to construct a physics-guided Gaussian process regression model as a pressure calculation model. It introduces Gaussian process regression and hyperparameter optimization, and uses the dataset to train the pressure calculation model until the model training is complete, generating frequency difference response curves and pressure sensitivity in the model. It is also used to identify small, continuous dynamic pressures based on total pressure signals and static pressure signals using the pressure calculation model, and to obtain dynamic pressure measurement results.

[0066] The high-precision dynamic pressure measurement system based on optical principles designed in this embodiment mainly includes a light source module, a measurement optical path module, a data acquisition module, and a data processing module. The design method of the measurement optical path is studied to achieve both large measurement range and high resolution; the method of receiving optical signals and the control method of the host computer over the received signals are studied to achieve real-time data transmission and measurement.

[0067] In the above system, the light source module consists of a laser and a circuit board. The laser, which serves as the light source, can be a narrow-linewidth DFB laser of 1260±1nm. The laser circuit board controls the laser current and temperature. The measurement module consists of a beam splitter, a measurement optical path, and a calibration optical path.

[0068] The laser beam is split into two paths after passing through a beam splitter. One beam passes through a calibration optical path, which consists of an adjustable attenuator and a gas chamber. After passing through the adjustable attenuator, the beam passes through an optical fiber-coupled hydrogen fluoride (HF) gas absorption cell (absorption peak wavelength 1260.7417 nm) in the gas chamber and is then output to the data acquisition module. This can serve as a wavelength standard and can compensate for laser frequency fluctuations caused by system drift and environmental noise in real time. The other beam passes through a measurement optical path, which consists of a circulator and a dynamic pressure measurement probe based on Michelson interferometry. After passing through the circulator, the beam reaches the measurement probe, which reflects optical signals, including total pressure and static pressure signals. These optical signals are then output to the data acquisition module via the circulator.

[0069] The data acquisition module consists of a photodetector, a signal input module, and a voltage output module. The voltage output module, controlled by LabVIEW software on a host computer, outputs voltage to a current driver to adjust the laser's drive current for frequency sweeping and outputting the laser beam. The photodetector converts the received optical signals from the calibration and measurement optical paths into electrical signals, enabling the acquisition, transmission, and processing of opto-electrical information. The electrical signals received by the photodetector are output to the data processing module through the signal input module.

[0070] The data processing module performs data processing based on the received signals, including processing the signals collected from the seawater sample in the measurement optical path, establishing a calculation model, and, in practical applications, performing dynamic pressure measurement based on the total pressure signal and static pressure signal through the model, and outputting dynamic pressure values.

[0071] Optionally, in the dynamic pressure measurement probe, the two optical paths are arranged in a vertical structure to generate the total pressure signal and the static pressure signal, respectively.

[0072] It should be noted that the real-time dynamic pressure measurement system in seawater provided in this application embodiment can execute the real-time dynamic pressure measurement method in seawater provided in any embodiment of this application, and has the corresponding functions and beneficial effects of the method.

[0073] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0074] The above description is merely a specific embodiment of this application, enabling those skilled in the art to understand or implement this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features claimed herein.

Claims

1. A method for real-time measurement of dynamic pressure in seawater, characterized in that, include: Under preset temperature and pressure, the laser beam generated by the light source module is split, and the measurement sample signal output by the dynamic pressure measurement probe and the standard sample signal output by the calibration optical path are collected. The dynamic pressure measurement probe is a measurement probe based on Michelson interferometry. The frequency difference between interference peaks and absorption peaks was analyzed based on standard sample signals and measured sample signals, and a dataset was constructed by combining temperature and pressure data. A physical-guided Gaussian process regression model based on a neural network architecture is constructed as a stress calculation model. Gaussian process regression and hyperparameter optimization are introduced, and a stress calculation model is trained using a dataset until the model training is complete. Frequency difference response curves and stress sensitivity are generated in the model. Acquire measurement signals from the target sea area collected by the dynamic pressure measurement probe. The measurement signals include total pressure signals and static pressure signals. Based on the total pressure signal and static pressure signal, a pressure calculation model is used to identify small, continuous dynamic pressure and obtain dynamic pressure measurement results.

2. The method according to claim 1, characterized in that, Under preset temperature and pressure conditions, the laser beam generated by the light source module is split, and the measurement sample signal output by the dynamic pressure measurement probe and the standard sample signal output by the calibration optical path are collected, including: For marine environments, under preset temperature and pressure conditions, the laser beam generated by the beam splitter module is divided into a first beam and a second beam. The first beam is output to the dynamic pressure measurement probe through the measurement optical path, and the total pressure sample signal and static pressure sample signal output by the dynamic pressure measurement probe are collected as measurement sample signals. The second beam is output to the calibration optical path to collect standard sample signals.

3. The method according to claim 1, characterized in that, Based on the analysis of the frequency difference between interference peaks and absorption peaks using standard sample signals and measured sample signals, and combined with temperature and pressure data, a dataset is constructed, including: Interference peaks are calculated based on measured sample signals, and absorption peaks are analyzed by compensating for laser frequency fluctuations based on standard sample signals. Based on interference peaks and absorption peaks, combined with measurement parameters, the frequency difference is analyzed; By using pressure samples from the reflective membrane in the acquired dynamic pressure measurement probe, the frequency difference is updated, and a dataset is constructed by combining temperature and pressure data.

4. The method according to claim 3, characterized in that, Based on interference peaks and absorption peaks, combined with measurement parameters, the frequency difference is analyzed, including: Based on the time interval between the recorded interference peaks and absorption peaks, the time delay is determined, and the frequency tuning range of the laser is obtained. Analyze the initial frequency difference based on interference peaks and absorption peaks; Using time delay and tuning range as measurement parameters, and combining them with the initial frequency difference, the frequency difference is calculated within the corresponding scanning period.

5. The method according to claim 3, characterized in that, Using pressure samples from the reflective membrane in the acquired dynamic pressure measurement probe, the frequency difference is updated, and combined with temperature and pressure data, a dataset is constructed, including: The first pressure of the first reflective film and the second pressure of the second reflective film in the dynamic pressure measurement probe are collected respectively as pressure samples. The second pressure is the pressure obtained after applying dynamic pressure on the basis of the first pressure. Under pressure samples, the update frequency differs; Integrate temperature and pressure data with frequency differences to construct a dataset.

6. The method according to claim 1, characterized in that, Gaussian process regression and hyperparameter optimization are introduced to train a stress calculation model using a dataset until the model training is complete. Frequency difference response curves and stress sensitivity are then generated within the model, including: Feature engineering and standardization preprocessing are performed on the dataset to obtain feature samples, which are then input into the stress calculation model. In the stress calculation model, Gaussian process regression with composite kernel function and initial parameters are used for training. The performance of the model on different datasets is evaluated, and the performance data is output. Based on the analysis of residual distribution and prediction uncertainty using performance data, hyperparameter optimization is performed using marginal likelihood maximization to obtain a well-trained stress calculation model. Based on the trained pressure calculation model, a frequency difference response curve caused by pressure is generated under preset temperature conditions, and the pressure sensitivity is quantified by partial derivative analysis. Based on the frequency difference response curve and pressure sensitivity, the decoupling results are determined through three-dimensional response surface visualization and physical consistency verification. The decoupling results are used to update the pressure calculation model.

7. The method according to any one of claims 1-6, characterized in that, Before acquiring the measurement sample signal output by the dynamic pressure measurement probe, the following steps are also included: In the multiphysics simulation software, a simulated dynamic pressure measurement probe is constructed. The simulated dynamic pressure measurement probe includes two simulated reflective films corresponding to the optical paths, and the film parameters of each simulated reflective film are the initial parameters. Physical simulations are performed within a preset temperature and pressure variation range. Based on the acquired simulation signals, the simulated optical path difference and simulated interference peaks between optical paths are analyzed, and the diaphragm displacement is calculated and adjusted. The initial parameters are updated based on the diaphragm displacement, and the simulated dynamic pressure measurement probe is adjusted synchronously.

8. The method according to claim 7, characterized in that, Based on the acquired analog signal analysis of the simulated optical path difference and simulated interference peaks between optical paths, the diaphragm displacement is calculated and adjusted, including: Based on the acquired analog signal, the diaphragm displacement is analyzed from the initial parameters; Based on the change in diaphragm displacement, the initial optical path difference between the two optical paths is analyzed; Based on the initial optical path difference and the refractive index of air, analyze the target optical path difference; Based on the target optical path difference and the simulated incident wavelength, the simulated interference peak is calculated; The diaphragm displacement is adjusted based on the simulated interference peak until the diaphragm displacement is the same as the incident wavelength or the error between the diaphragm displacement and the incident wavelength is lower than the preset value.

9. A real-time dynamic pressure measurement system in seawater, characterized in that, It includes a light source module, a measurement module, a data acquisition module, and a data processing module; The light source module is used to generate a laser beam; The measurement module is used to split the generated laser beam and output the measurement sample signal from the dynamic pressure measurement probe and the standard sample signal from the calibration optical path. The dynamic pressure measurement probe is a measurement probe based on Michelson interferometry. The data acquisition module is used to acquire measurement sample signals and standard sample signals; as well as measurement signals from the target sea area, including total pressure signals and hydrostatic pressure signals. The data processing module is used to analyze the frequency difference between interference peaks and absorption peaks based on standard sample signals and measured sample signals, and to construct a dataset by combining temperature and pressure data. It adopts a neural network architecture to construct a physics-guided Gaussian process regression model as a pressure calculation model. It introduces Gaussian process regression and hyperparameter optimization, and uses the dataset to train the pressure calculation model until the model training is complete, generating frequency difference response curves and pressure sensitivity in the model. It is also used to identify small, continuous dynamic pressures based on total pressure signals and static pressure signals using the pressure calculation model, and to obtain dynamic pressure measurement results.

10. The system according to claim 9, characterized in that, In the dynamic pressure measurement probe, two optical paths are arranged in a vertical structure to generate total pressure and static pressure signals, respectively.

Images