In-situ multimodal detection method and system for deep-sea high-pressure stress corrosion simulation

Abstract

The application provides a deep-sea high-pressure stress corrosion simulation method and system for in-situ multi-modal detection, and relates to the technical field of corrosion detection. The method comprises the following steps: constructing a deep-sea high-pressure stress corrosion environment simulation unit to collect simulation data; controlling a multi-dimensional motion platform to alternately switch detection based on in-situ optical data sets; coupling and analyzing the in-situ optical data sets and multi-modal data sets, constructing a stress-chemical correlation array to predict the corrosion risk of deep-sea high pressure; synchronizing the corrosion risk assessment report to the multi-dimensional motion platform for detection optimization, updating the multi-modal data set for closed-loop analysis, and formulating an automatic optimization detection strategy. The application solves the technical problem of low detection efficiency in the prior art due to the fact that the in-situ detection method under deep-sea high-pressure environment cannot simultaneously satisfy in-situ Raman and in-situ DIC testing, and improves the material corrosion detection efficiency in deep-sea environment by adopting a multi-dimensional motion platform to alternately switch detection.

Description

Technical Field

[0001] This application relates to the field of corrosion detection technology, and in particular to a deep-sea high-pressure stress corrosion simulation method and system for in-situ multimodal detection. Background Technology

[0002] In-situ testing technology is becoming increasingly widely used, especially in corrosion detection and material performance evaluation. However, most existing in-situ testing technologies are limited to ambient temperature and pressure or low-pressure environments, making it difficult to meet the monitoring needs of corrosion behavior under the high pressure environment of the deep sea. The extreme conditions of the deep-sea environment, such as high pressure, low temperature, and darkness, pose severe challenges to the material selection, structural design, and performance stability of in-situ testing equipment. This results in numerous bottlenecks in the in-situ testing process simulating the high pressure environment of the deep sea, making it difficult to effectively achieve long-term and accurate monitoring under such conditions.

[0003] In particular, the combination of in-situ Raman spectroscopy and in-situ digital image correlation (DIC) technologies faces challenges such as poor equipment coordination and low detection efficiency. In-situ Raman spectroscopy is an analytical instrument that incorporates an in-situ reaction system, enabling micron-level micro-area detection of samples. In-situ DIC, or in-situ digital image correlation, is a technology system that combines experimental loading environment with optical measurement methods to acquire full-field displacement and strain distribution in real-time, dynamically, and non-contactly during the deformation of materials or structures under stress. Existing in-situ detection methods cannot meet the testing requirements of simultaneous in-situ Raman spectroscopy and in-situ DIC in the high-pressure environment of the deep sea, thus affecting detection efficiency and increasing the complexity of equipment switching and operation.

[0004] In summary, existing technologies suffer from the technical problem that in-situ detection methods under high-pressure deep-sea environments cannot simultaneously perform in-situ Raman and in-situ DIC testing, resulting in low detection efficiency. Summary of the Invention

[0005] The purpose of this application is to provide a deep-sea high-pressure stress corrosion simulation method and system for in-situ multimodal detection, in order to solve the technical problem that the in-situ detection method in the deep-sea high-pressure environment cannot meet the requirements of simultaneous in-situ Raman and in-situ DIC testing, resulting in low detection efficiency.

[0006] In view of the above problems, this application provides a method and system for simulating deep-sea high-pressure stress corrosion for in-situ multimodal detection.

[0007] Firstly, this application provides a deep-sea high-pressure stress corrosion simulation method for in-situ multimodal detection. This method is implemented through a deep-sea high-pressure stress corrosion simulation system. The method includes: constructing a deep-sea high-pressure stress corrosion environment simulation unit to collect simulation data and obtain an in-situ optical dataset; controlling a multi-dimensional motion platform to alternately switch detection based on the in-situ optical dataset to obtain a multimodal dataset; coupling the in-situ optical dataset with the multimodal dataset to construct a stress-chemical correlation array for predicting corrosion risks under deep-sea high-pressure stress and constructing a corrosion risk assessment report; synchronizing the corrosion risk assessment report to the multi-dimensional motion platform for detection optimization, updating the multimodal dataset for closed-loop analysis, and formulating an automatic optimization detection strategy.

[0008] Optionally, an in-situ external observation of the high-pressure reactor is performed through a sapphire optical observation window to generate an in-situ external optical observation dataset; a retractable Raman probe is controlled to perform in-situ internal acquisition of the high-pressure reactor to obtain an in-situ internal Raman spectrum; the in-situ external optical observation dataset and the in-situ internal Raman spectrum are time-synchronized to plan a detection sequence; and the in-situ external optical observation dataset and the in-situ internal Raman spectrum are correlated according to the detection sequence to construct the in-situ optical dataset.

[0009] Optionally, the in-situ optical observation dataset outside the vessel is decomposed to obtain sample surface morphology observation data and dynamic change observation data; the sample surface morphology observation data is enhanced according to the dynamic change observation data to obtain optical morphology features; the in-situ Raman spectrum inside the vessel is baseline corrected, and peak position is identified based on the correction results to determine the optochemical features; the optical morphology features and the optochemical features are correlated and analyzed, and the in-situ optical dataset is constructed based on the correlation.

[0010] Optionally, the DIC detection module and the Raman detection module are installed in parallel on a multi-dimensional motion platform; surface feature analysis is performed based on the in-situ optical dataset to determine surface feature distribution data; key points are identified according to the surface feature distribution data, and multiple key regions are divided according to the identification results; focusing analysis is performed based on the Raman detection module to determine Raman focusing parameters; field of view analysis is performed based on the DIC detection module to determine DIC imaging parameters; based on the Raman focusing parameters and the DIC imaging parameters, the DIC detection module and the Raman detection module are controlled to alternately switch detection across the multiple key regions to generate the multimodal dataset.

[0011] Optionally, the coordinates of the multiple key regions are read to determine the key region coordinate map; based on the Raman focusing parameters and the DIC imaging parameters, detection analysis is performed to determine the target detection parameters; the target detection parameters are mapped to the key region coordinate map for motion planning to generate a detection movement path; the DIC detection module and the Raman detection module are switched and analyzed according to the detection movement path to determine an alternating detection scheme; the DIC detection module and the Raman detection module are controlled to perform detection according to the alternating detection scheme to generate the multimodal dataset.

[0012] Optionally, the in-situ optical dataset and the multimodal dataset are time-synchronized to generate first synchronized data; the in-situ optical dataset and the multimodal dataset are spatially transformed to generate second synchronized data; the first synchronized data and the second synchronized data are registered and aligned to construct a spatiotemporal sequence feature database; the spatiotemporal sequence feature database is dimensionality-reduced to extract key feature parameters; the parameter correlation strength between stress parameters and chemical parameters is calculated based on the key feature parameters, and multivariate combination analysis is performed according to the parameter correlation strength to construct a stress-chemical correlation array.

[0013] Optionally, based on the key feature parameters, key stress feature parameters and key chemical feature parameters are obtained through analysis; linear correlation calculation is performed on the key stress feature parameters and the key chemical feature parameters to obtain the linear correlation strength; nonlinear correlation calculation is performed on the key stress feature parameters and the key chemical feature parameters to obtain the nonlinear correlation strength; and grey relational analysis is performed on the linear correlation strength and the nonlinear correlation strength to obtain the parameter correlation strength.

[0014] Optionally, the corrosion risk assessment report is read for risk distribution to identify the coordinates of high-risk areas, where the coordinates of high-risk areas contain risk values; the corrosion risk assessment report is mapped to a multi-dimensional motion platform to increase the density of detection points at the coordinates of the high-risk areas according to the risk values, generating an optimized detection movement path; the DIC detection module and the Raman detection module are controlled to perform mobile detection according to the optimized detection movement path, generating a multimodal update dataset; a real-time comparison is performed between the multimodal update dataset and the multimodal dataset to generate detection change values; when the detection change values ​​are higher than a preset change threshold, the multimodal dataset is updated.

[0015] Optionally, parameter change identification is performed on the multimodal dataset to obtain parameter change trends; detection performance analysis is performed on the multimodal updated dataset according to the parameter change trends to generate data detection gain information; the multimodal dataset is evaluated and optimized based on the data detection gain information to formulate a detection optimization knowledge base; and closed-loop feedback control is performed based on the detection optimization knowledge base to formulate the automatic optimization detection strategy.

[0016] Secondly, this application also provides an in-situ multimodal detection deep-sea high-pressure stress corrosion simulation system for performing the in-situ multimodal detection deep-sea high-pressure stress corrosion simulation method as described in the first aspect. The in-situ multimodal detection deep-sea high-pressure stress corrosion simulation system includes: a data acquisition module for constructing a deep-sea high-pressure stress corrosion environment simulation unit to acquire simulation data and obtain an in-situ optical dataset; a switching detection module for controlling a multi-dimensional motion platform to alternately switch detection based on the in-situ optical dataset to obtain a multimodal dataset; a corrosion risk prediction module for coupling the in-situ optical dataset with the multimodal dataset to construct a stress-chemical correlation array for predicting corrosion risk under deep-sea high-pressure stress and constructing a corrosion risk assessment report; and a closed-loop analysis module for synchronizing the corrosion risk assessment report to the multi-dimensional motion platform for detection optimization, updating the multimodal dataset for closed-loop analysis, and formulating an automatic optimization detection strategy.

[0017] One or more technical solutions provided in this application have at least the following beneficial effects: By constructing a deep-sea high-pressure stress corrosion environment simulation unit to collect simulated data, an in-situ optical dataset is obtained; based on the in-situ optical dataset, a multi-dimensional motion platform is controlled to alternately switch detection to obtain a multi-modal dataset; the in-situ optical dataset and the multi-modal dataset are coupled and analyzed to construct a stress-chemical correlation array for predicting corrosion risks under deep-sea high-pressure stress, and a corrosion risk assessment report is constructed; the corrosion risk assessment report is synchronized to the multi-dimensional motion platform for detection optimization, the multi-modal dataset is updated for closed-loop analysis, and an automatic optimization detection strategy is formulated. In other words, by constructing a deep-sea high-pressure stress corrosion environment simulation unit, combining the coupled analysis of the in-situ optical dataset and the multi-modal dataset, employing alternating detection by a multi-dimensional motion platform, and determining an automatic optimization detection strategy based on the real-time corrosion risk assessment report, the accuracy and efficiency of material corrosion detection in the deep-sea environment are improved.

[0018] The above description is merely an overview of the technical solution of this application. To better understand the technical means of this application and to facilitate its implementation according to the description, and to make the above and other objects, features, and advantages of this application more apparent, specific embodiments of this application are described below. It should be understood that the content described in this section is not intended to identify key or important features of the embodiments of this application, nor is it intended to limit the scope of this application. Other features of this application will become readily apparent through the following description. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in 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, the drawings described below are merely exemplary. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0020] Figure 1 This is a schematic diagram of the process for the deep-sea high-pressure stress corrosion simulation method for in-situ multimodal detection in this application.

[0021] Figure 2 This is a schematic diagram of the deep-sea high-pressure stress corrosion simulation system for in-situ multimodal detection in this application.

[0022] Figure labeling: Data acquisition module 11, switching detection module 12, corrosion risk prediction module 13, closed-loop analysis module 14. Detailed Implementation

[0023] This application provides a deep-sea high-pressure stress corrosion simulation method and system for in-situ multimodal detection, solving the technical problem of low detection efficiency in existing technologies due to the inability of in-situ detection methods to simultaneously perform in-situ Raman and in-situ DIC tests under deep-sea high-pressure environments. By constructing a deep-sea high-pressure stress corrosion environment simulation unit, combining the coupled analysis of in-situ optical datasets and multimodal datasets, and employing a multidimensional motion platform for alternating detection, the application determines an automatically optimized detection strategy based on real-time corrosion risk assessment reports, thereby improving the accuracy and efficiency of material corrosion detection in deep-sea environments.

[0024] The technical solutions of this application will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. It should be understood that this application is not limited to the exemplary embodiments described herein. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application. It should also be noted that, for ease of description, only the parts related to this application are shown in the accompanying drawings, not all of them.

[0025] Example 1, please refer to the appendix. Figure 1 This application provides a deep-sea high-pressure stress corrosion simulation method for in-situ multimodal detection, wherein the in-situ multimodal detection deep-sea high-pressure stress corrosion simulation method is executed by an in-situ multimodal detection deep-sea high-pressure stress corrosion simulation system, and the in-situ multimodal detection deep-sea high-pressure stress corrosion simulation method specifically includes the following steps: A deep-sea high-pressure stress corrosion environment simulation unit was constructed to collect simulation data and obtain in-situ optical datasets.

[0026] Furthermore, this application also includes the following steps: performing in-situ observations of the exterior of the high-pressure reactor through a sapphire optical observation window to generate an in-situ optical observation dataset; controlling a retractable Raman probe to perform in-situ acquisitions inside the high-pressure reactor to obtain in-situ Raman spectra; synchronizing the in-situ optical observation dataset outside the reactor with the in-situ Raman spectra inside the reactor in time to plan a detection sequence; and performing correlation analysis between the in-situ optical observation dataset outside the reactor and the in-situ Raman spectra inside the reactor according to the detection sequence to construct the in-situ optical dataset.

[0027] Furthermore, this application also includes the following steps: decomposing the in-situ optical observation dataset outside the vessel to obtain sample surface morphology observation data and dynamic change observation data; performing enhancement analysis on the sample surface morphology observation data according to the dynamic change observation data to obtain optical morphology features; performing baseline correction on the in-situ Raman spectrum inside the vessel, and determining the optochemical features by peak position identification based on the correction results; performing correlation change analysis on the optical morphology features and the optochemical features, and constructing the in-situ optical dataset based on the correlation relationship.

[0028] Specifically, achieving in-situ detection inside a high-pressure reactor is quite difficult because the equipment used for in-situ detection needs to withstand not only high pressure but also corrosion. Therefore, in-situ detection is achieved externally to the high-pressure reactor by creating a sapphire optical observation window. Sapphire is a material with excellent optical properties and high-pressure resistance, commonly used for optical observation windows in deep-sea, high-temperature, and high-pressure environments. Through the sapphire optical observation window, external optical equipment, such as a high-resolution camera or imaging system, monitors the state of the sample inside the high-pressure reactor in real time, obtaining an in-situ optical observation dataset that records the surface state, morphology, and changes of the experimental sample inside the high-pressure reactor. In-situ external observation refers to observing the internal sample through a sapphire optical observation window from outside the high-pressure reactor.

[0029] The high-pressure reactor includes a vessel body, a cover, and a testing platform. One end of the vessel body is open, and the cover is placed over the opening. The testing platform is located inside the vessel body for placing samples. The side of the vessel body has a first mounting hole and a second mounting hole, with the ends of the first and second mounting holes facing the samples on the testing platform. A first optical mirror covers the end of the vessel body with the first mounting hole facing the testing platform. A Raman mount has a second optical mirror encapsulated at one end, and the Raman mount has a first channel leading to the second optical mirror. The Raman mount passes through the second mounting hole, and the second optical mirror on the Raman mount extends into the interior of the vessel body, facing the testing platform. A detection device includes a retractable Raman probe and an image lens. The image lens passes through the first mounting hole, and the retractable Raman probe passes through the first channel. During in-situ DIC testing, the image lens can be inserted into the first mounting hole, and the sample on the testing platform can be detected using the first optical mirror. During in-situ Raman testing, the retractable Raman probe can be inserted into the first channel of the Raman mount, and the sample on the testing platform can be detected using the second optical mirror.

[0030] A retractable Raman probe is used to perform in-situ Raman spectral data acquisition inside a high-pressure reactor. The probe is inserted into the reactor to collect Raman spectral data of the sample under high pressure. The probe can be extended or retracted as needed to adjust its depth within the reactor. For optimal Raman signal acquisition, the closer the probe is to the sample, the stronger the signal. However, in high-pressure reactors, it is difficult to maintain a close distance between the sample inside the reactor and the external Raman spectrometer. In such cases, the probe is inserted into the reactor at a distance of 1mm to 2mm from the sample. One end of the Raman mount, equipped with a second optical mirror, extends into the reactor, bringing the second mirror close to the sample on the testing platform to meet the requirements of in-situ Raman testing. The Raman probe illuminates the sample with a laser and collects the scattered light caused by molecular vibrations to obtain the in-situ Raman spectrum inside the reactor.

[0031] The in-situ optical observation dataset outside the vessel and the in-situ Raman spectrum inside the vessel are synchronized in time to ensure that the two datasets correspond at the same point in time. Data from different time points are matched to ensure that the acquired information from the optical data outside the vessel and the Raman data inside the vessel can be correlated, thus ensuring the synchronicity of the two types of detection data at the same moment. Planning the detection sequence refers to arranging the data in a pre-set time sequence according to the purpose of the experiment and the order of data acquisition, ensuring the effective coordination of different data points during the experiment.

[0032] Based on the monitoring sequence, the in-situ optical dataset outside the vessel was combined with the in-situ Raman spectrum inside the vessel through correlation analysis. The in-situ optical observation dataset outside the vessel was decomposed to extract sample surface morphology observation data and dynamic change observation data. The sample surface morphology observation data records the static morphological features of the sample, such as microcracks, corrosion spots, and surface roughness; the dynamic change observation data records the changes of these features over time, such as corrosion process and deformation. For example, assuming that image data of the sample under high pressure was obtained through a high-resolution camera, after decomposition, two datasets were obtained: one is the sample surface morphology observation data, which includes the length of the surface crack: 5.2 mm and the depth: 0.3 mm; the other is the dynamic change observation data, which records the crack propagation rate, such as 0.1 mm per hour.

[0033] Enhancement analysis of sample surface morphology observation data is performed based on dynamic change observation data. Image processing and signal processing methods are used to further analyze the original data, enhancing its features and making the surface morphology data clearer and more accurate during analysis. Dynamic change observation data provides information on the changes in the sample surface at different time points, revealing subtle changes that occur in the material during the experiment. Sample surface morphology observation data records the state of the sample surface at a specific moment. Time-series comparisons can be used to observe changes in the material's surface morphology over time, guiding the enhancement analysis of the surface morphology data. During the enhancement analysis, image processing algorithms are applied to the sample surface morphology data to remove noise from the images, making features such as surface cracks and corrosion spots more clearly visible. After enhancement analysis, key information about the surface morphology in the images is extracted, resulting in optical morphology features, including crack depth, width, and propagation rate, or the distribution and size of corrosion spots.

[0034] Baseline correction is performed on the in-situ Raman spectra within the vessel. By fitting the background signal and subtracting interfering components, a more accurate Raman spectrum is obtained. Polynomial fitting is used to remove background noise from the spectrum, or rolling averaging is used to smooth the spectral data, eliminating interference caused by environmental factors or instrumentation and improving signal accuracy. The corrected Raman spectrum will display multiple Raman scattering peaks, each representing a specific chemical substance in the sample. By analyzing the position, shape, and intensity of multiple Raman scattering peaks, the optical chemical characteristics of the sample can be identified. For example, the oxide characteristics of certain materials will be visible at 1000 cm⁻¹. -1 A peak may appear nearby, or certain molecular structures may appear at 3000 cm⁻¹. -1 Specific peak positions are formed in the left and right regions. By identifying the peak positions in the Raman spectrum, the optical chemical characteristics of the sample can be extracted, such as molecular vibrational modes, chemical bonding types, or the redox state of the sample under high pressure.

[0035] Optical morphological features and photochemical features were analyzed in combination according to the detection sequence. Since changes in sample surface morphology are often accompanied by changes in chemical composition, correlation analysis can reveal the intrinsic relationship between the two. For example, the appearance of certain corrosion spots is related to oxide formation, or crack propagation is related to the chemical degradation of the material. By analyzing the correlation changes of optical morphological features and photochemical features, an in-situ optical dataset was generated, including crack information, corrosion characteristics, and corresponding chemical information on the material surface, such as oxide formation and the concentration of corrosion products. Correlation analysis revealed a significant incremental relationship between iron oxide formation and crack propagation. Based on this correlation, the interaction between crack propagation and oxidation reaction can be inferred and incorporated into the in-situ optical dataset. For example, assuming data was collected every 30 minutes during the experiment, image processing yielded the optical morphological features of crack propagation: the crack extended from 1.0 mm to 2.5 mm. Simultaneously, Raman spectroscopy analysis yielded the photochemical features of oxide formation: the peak position of iron oxide was at 550 cm⁻¹. -1 The peak intensity gradually increases.

[0036] By synchronizing and correlating in-situ optical observation data outside the vessel with in-situ Raman spectroscopy data inside the vessel, the constructed in-situ optical dataset provides a multi-dimensional record of sample changes, which can not only reveal changes in sample surface morphology, but also provide detailed information on chemical composition, corrosion process and aging mechanism.

[0037] Based on the in-situ optical dataset, a multi-dimensional motion platform is controlled to alternately switch detection to obtain a multi-modal dataset.

[0038] Furthermore, this application also includes the following steps: installing the DIC detection module and the Raman detection module in parallel on a multi-dimensional motion platform; performing surface feature analysis based on the in-situ optical dataset to determine surface feature distribution data; marking key points according to the surface feature distribution data and dividing multiple key regions based on the marking results; performing focusing analysis based on the Raman detection module to determine Raman focusing parameters; performing field-of-view analysis based on the DIC detection module to determine DIC imaging parameters; and controlling the DIC detection module and the Raman detection module to alternately switch detection across the multiple key regions based on the Raman focusing parameters and the DIC imaging parameters to generate the multimodal dataset.

[0039] Furthermore, this application also includes the following steps: traversing the multiple key regions to read coordinates and determine the key region coordinate map; performing detection analysis based on the Raman focusing parameters and the DIC imaging parameters to determine the target detection parameters; mapping the target detection parameters to the key region coordinate map for motion planning to generate a detection movement path; performing switching analysis on the DIC detection module and the Raman detection module according to the detection movement path to determine an alternating detection scheme; controlling the DIC detection module and the Raman detection module to perform detection according to the alternating detection scheme to generate the multimodal dataset.

[0040] Specifically, the DIC (Displacement and Induction) testing module is a non-destructive testing method based on optical image acquisition and image processing technology, used to monitor the deformation, displacement, and strain of materials under external forces in real time. The DIC module acquires full-field displacement and strain distribution data by capturing images of the material surface and comparing image changes at different time points. During in-situ DIC testing, an image lens can be inserted into the first mounting hole, and the sample on the test platform can be inspected using a first optical mirror.

[0041] The Raman detection module is a device module used for in-situ Raman spectroscopy detection. Raman spectroscopy technology obtains information such as the chemical composition and molecular structure of a sample by illuminating the sample surface with light of a specific wavelength and analyzing the reflected or scattered spectra. The DIC detection module and the Raman detection module are installed in parallel on a multi-dimensional motion platform, ensuring that both modules can be used simultaneously in the same experiment. The multi-dimensional motion platform provides precise control over the DIC and Raman detection modules, allowing for flexible spatial switching between them.

[0042] Surface feature analysis is performed using in-situ optical datasets. This involves analyzing the surface features of a sample to obtain surface feature distribution data, including the number, location, size, and shape of the surface features. For example, in a corrosion experiment, the in-situ optical dataset showed several cracks and corrosion spots on the sample surface. Analysis revealed that the crack length was 2-5 mm and the corrosion spot diameter was 1-3 mm.

[0043] Key points are identified based on surface feature distribution data. Key features on the surface are marked, resulting in identification results, such as crack initiation coordinates (5.5, 2.2) and corrosion spot coordinates (7.3, 4.1). Key points are significant feature points on the sample surface, such as crack initiation points and corrosion spot locations. Based on the identification results, i.e., the location and distribution of the identified key points, the sample surface is divided into several key regions, typically the areas with the most significant surface damage or changes. These regions are then the focus of attention to understand the damage evolution process of the material under extreme conditions.

[0044] Focusing analysis is performed using the Raman detection module. By adjusting the focusing parameters of the optical system, appropriate focal length, laser power, and incident angle are selected to ensure that the laser beam can accurately illuminate the sample surface and obtain the optimal Raman signal. After focusing analysis, the Raman focusing parameters, i.e., the focusing settings in Raman spectroscopy analysis, are determined, including focal length, probe incident angle, and laser power. For example, the Raman probe has a laser power of 10mW, a focal length of 50μm, and an incident angle of 45°. DIC detection acquires deformation and strain data by capturing images of the sample surface. To ensure accurate capture of minute changes on the sample surface, field-of-view analysis must be performed to determine the optimal imaging area. Field-of-view analysis determines the required imaging range and establishes the DIC imaging parameters, i.e., the various technical settings affecting DIC image quality, including imaging resolution, acquisition frequency, and exposure time. For example, the imaging resolution of the DIC detection module is set to 2μm, and the exposure time to 50ms to ensure that minute changes in crack propagation can be accurately captured.

[0045] Multiple key areas, including crack initiation points and corrosion spots, were traversed. Coordinates of each key area were read, and using optical imaging systems and other technologies, the spatial coordinates of each area were precisely determined, resulting in a key area coordinate map. This key area coordinate map integrates the coordinate data of the key areas into a graphic representation, showing the distribution of each key area on the sample surface. Based on the pre-set Raman focusing parameters and DIC imaging parameters, detection analysis was performed to determine the target detection parameters, ensuring that specific detection requirements, such as resolution, sensitivity, and dynamic range, are met during Raman spectroscopy analysis and DIC deformation measurement.

[0046] Based on the target detection parameters and the coordinate map of the key areas, motion planning is performed to determine the optimal detection paths for the DIC and Raman probes, ensuring that the DIC and Raman detection modules can cover all key areas and maximize detection efficiency. For example, the detection movement path first aligns with the key area at coordinates (10,20), then moves to the key area at coordinates (30,50), and so on, until all key areas are traversed.

[0047] During the detection process, to avoid excessively long detection times or wasted module resources, it is necessary to rationally arrange the alternating use of the DIC and Raman modules, considering detection efficiency, data quality, and experimental requirements. Switching should be performed based on the detection movement path to avoid overlapping coverage of areas or missing important regions. For example, first use the Raman detection module to acquire spectra in the critical region at coordinates (10,20), then move to the critical region at coordinates (30,50) and use the DIC detection module for deformation measurement. Then move again to the critical region at coordinates (60,70) and use the Raman detection module for spectra acquisition, and so on, until all critical regions have been traversed. By switching analysis, an alternating detection scheme is determined, that is, to determine which key areas to use the Raman detection module and which key areas to use the DIC detection module. For example, the Raman detection module is used in the key areas with coordinates (10,20), (60,70) and (90,60), while the DIC detection module is used in the key areas with coordinates (30,50), (20,90), (50,10), (70,30), (40,80), (80,40) and (25,25).

[0048] According to the determined alternating detection scheme, the DIC detection module and the Raman detection module are controlled to work alternately to acquire sample surface deformation data and chemical composition data in real time, generating a multimodal dataset containing multidimensional data such as optical morphology, chemical composition, strain, and displacement. By alternating the use of Raman and DIC modules, combined with precise target detection parameters, motion planning, and alternating detection scheme, experimental efficiency and data quality are effectively improved. Through precise positioning and detection of key areas on the sample surface, multidimensional information such as sample morphology, chemical composition, displacement, and strain can be obtained simultaneously.

[0049] The in-situ optical dataset and the multimodal dataset are coupled and analyzed to construct a stress-chemical correlation array for predicting corrosion risk under deep-sea high-pressure stress, and a corrosion risk assessment report is constructed.

[0050] Furthermore, this application also includes the following steps: synchronizing the in-situ optical dataset with the multimodal dataset in time to generate first synchronized data; transforming the in-situ optical dataset with the multimodal dataset in spatial coordinates to generate second synchronized data; registering and aligning the first synchronized data with the second synchronized data to construct a spatiotemporal sequence feature database; performing dimensionality reduction processing on the spatiotemporal sequence feature database to extract key feature parameters; calculating the parameter correlation strength between stress parameters and chemical parameters based on the key feature parameters; and performing multivariate combination analysis according to the parameter correlation strength to construct a stress-chemical correlation array.

[0051] Furthermore, this application also includes the following steps: analyzing the key feature parameters to obtain key stress feature parameters and key chemical feature parameters; performing linear correlation calculations based on the key stress feature parameters and the key chemical feature parameters to obtain the linear correlation strength; performing nonlinear correlation calculations based on the key stress feature parameters and the key chemical feature parameters to obtain the nonlinear correlation strength; and performing grey relational analysis based on the linear correlation strength and the nonlinear correlation strength to obtain the parameter correlation strength.

[0052] Specifically, since optical data and other multimodal data are typically acquired at different times, it is necessary to synchronize the in-situ optical dataset with the multimodal dataset so that they correspond to the same time point. For example, suppose the in-situ optical dataset was acquired from T1=10:00AM to T2=10:05AM, while the multimodal dataset was acquired from T3=10:01AM to T4=10:06AM. Interpolation methods can be used to align the two datasets to a common time, such as 10:01AM to 10:04AM, thus generating the first synchronized data.

[0053] Since in-situ optical datasets and multimodal datasets use different coordinate systems, spatial coordinate transformation ensures that they can be compared and analyzed within a unified spatial coordinate system. For example, suppose the in-situ optical dataset is based on a camera coordinate system, such as pixel coordinates, while the multimodal dataset is based on a laser probe coordinate system. Through spatial coordinate transformation algorithms, such as coordinate transformation matrices, the spatial coordinates of both datasets can be mapped to a common coordinate system. Transforming the spatial coordinates of the in-situ optical dataset and the multimodal dataset allows data from different devices to be compared and analyzed within a unified spatial coordinate system.

[0054] The first and second synchronized data are registered and aligned. Using a registration and alignment algorithm, the in-situ optical dataset and the multimodal dataset are aligned temporally and spatially, resulting in a spatiotemporal sequence feature database containing various data features in both time and space. Since the spatiotemporal sequence feature database contains a large amount of redundant data, dimensionality reduction is used to extract the most representative and relevant feature parameters, reducing computational burden and obtaining key feature parameters. Principal component analysis is then used to map the high-dimensional data to a low-dimensional space, retaining the most important feature parameters.

[0055] Key characteristic parameters are analyzed to obtain key stress characteristic parameters and key chemical characteristic parameters. Key stress characteristic parameters include strain concentration factor, stress amplitude, and loading frequency. The strain concentration factor describes the degree of strain generated in a localized area of ​​the material due to stress concentration; it typically reflects the risk of fracture or fatigue in the stress concentration zone through a local increase in strain. The stress amplitude is the maximum stress value experienced by the material or structure during loading, reflecting the intensity of the applied external force and its impact on the material. The loading frequency is the frequency of applied external force, i.e., the number of times the force is applied per unit time, affecting the fatigue life and stress distribution of the material.

[0056] Key chemical characteristic parameters include characteristic Raman peak intensity, peak shift, and half-maximum width (HWHM). Characteristic Raman peak intensity is the peak intensity in a Raman spectrum, reflecting the molecular response of a material under a specific vibrational mode; higher intensity indicates a higher concentration of the chemical component. Peak shift is the amount of displacement of a peak in a Raman spectrum, typically used to analyze structural changes and stress states of molecules or crystals. HWHM is the width of the peak, representing the extent of peak expansion. A wider peak may indicate impurities or structural defects in the material. For example, a characteristic Raman peak intensity of 500 units might have a peak shift of 10 cm⁻¹. -1 The half-peak width is 15cm -1 The stress amplitude is 200 MPa, the strain concentration factor is 3, and the loading frequency is 1 Hz.

[0057] Given key stress and chemical characteristic parameters, a linear regression model is used to calculate the linear relationship between them. For example, the linear relationship between stress amplitude and characteristic Raman peak intensity can be used to quantify their correlation strength. Linear correlation calculations, performed using methods such as linear regression, determine the linear relationship between two variables to reveal the direct relationship between stress and chemical characteristic parameters and to provide the correlation strength. For instance, a linear regression model might yield the linear relationship between stress amplitude and characteristic Raman peak intensity as: peak intensity = 10 * stress amplitude - 100.

[0058] For complex relationships that linear regression cannot adequately describe, nonlinear regression is used to calculate the nonlinear relationship between key stress characteristic parameters and key chemical characteristic parameters. For example, the relationship obtained through nonlinear regression model analysis is: peak intensity = 5 * (stress amplitude). 2 +50. At a stress amplitude of 200 MPa, the peak intensity will be 5000 units.

[0059] Grey relational analysis, combining linear and nonlinear correlation strengths, assesses the overall correlation strength between stress and chemical characteristics, extracts the most important influencing factors, derives parametric correlation strength, and identifies key relationships between stress and chemical characteristics. For example, assuming a linear correlation strength of 0.8 and a nonlinear correlation strength of 0.6, the final correlation strength obtained from grey relational analysis is 0.75, indicating a strong correlation between stress and chemical characteristics. Grey relational analysis is a method based on grey system theory used to evaluate the degree of correlation between multiple variables. It is suitable for situations where data is incomplete or uncertain and can extract the most important information from a series of data. Linear and nonlinear correlation strengths provide two different perspectives on stress and chemical characteristics; grey relational analysis combines these two perspectives, calculating the final correlation strength by similarity. Grey relational analysis methods often use the grey scale of data to assess the degree of correlation.

[0060] Multivariate combination analysis aims to construct a comprehensive stress-chemical correlation array by weighting and combining multiple key stress and chemical characteristic parameters using statistical methods. This array reflects the combined influence of various factors and provides a comprehensive analysis of material properties. Principal component analysis integrates multiple key features, identifies the most important variables and relationships between stress and chemical characteristics, assigns them appropriate weights, and derives the stress-chemical correlation array, demonstrating the response of chemical characteristics under different stress conditions. For example, the stress-chemical correlation array constructed in a certain experiment is shown in Table 1. Table 1 Stress-Chemical Correlation Array

[0061] By revealing the correlation between stress and chemical reactions, especially its changes under complex environments, linear and nonlinear correlation calculation methods can provide a comprehensive quantitative description of the relationship between the two, while grey relational analysis can further uncover the key links between stress and chemical characteristics. The resulting stress-chemical correlation array effectively optimizes material design and improves the corrosion resistance of materials.

[0062] Corrosion risk prediction is performed using an established stress-chemical correlation array, combined with known experimental data and corrosion models. The changes in chemical reactions of the material under different stress levels are calculated through numerical simulations or experimental data, and then the risk areas and corrosion rates are predicted based on these changes. For example, assuming a material is subjected to a high-pressure environment in the deep sea, the characteristic peak shift in its Raman spectrum changes by 10 cm⁻¹. -1The corresponding stress amplitude is 200 MPa. Using a correlation array, the corrosion rate of the material under this stress condition is predicted to be 0.2 mm / year, from which the corrosion area and potential risks can be calculated. Based on the corrosion risk prediction results and considering the characteristics of the deep-sea high-pressure environment, a corrosion risk assessment report is constructed, including corrosion risk distribution, corrosion rate prediction, material life assessment, and recommended measures. The corrosion risk assessment report, based on the corrosion risk prediction results, combined with experimental data and mathematical models, provides a detailed report describing the material's corrosion risk status and possible future corrosion trends.

[0063] The corrosion risk assessment report is synchronized to the multi-dimensional motion platform for detection optimization, the multi-modal dataset is updated for closed-loop analysis, and an automatic optimization detection strategy is formulated.

[0064] Furthermore, this application also includes the following steps: reading the risk distribution of the corrosion risk assessment report, identifying the coordinates of high-risk areas, wherein the coordinates of the high-risk areas contain risk values; mapping the corrosion risk assessment report to a multi-dimensional motion platform to increase the density of detection points at the coordinates of the high-risk areas according to the risk values, generating an optimized detection movement path; controlling the DIC detection module and the Raman detection module to perform mobile detection according to the optimized detection movement path, generating a multimodal update dataset; performing real-time comparison between the multimodal update dataset and the multimodal dataset to generate detection change values; updating the multimodal dataset when the detection change value is higher than a preset change threshold.

[0065] Furthermore, this application also includes the following steps: identifying parameter changes based on the multimodal dataset to obtain parameter change trends; performing detection performance analysis on the multimodal updated dataset according to the parameter change trends to generate data detection gain information; evaluating and optimizing the multimodal dataset based on the data detection gain information to formulate a detection optimization knowledge base; and performing closed-loop feedback control based on the detection optimization knowledge base to formulate the automatic optimization detection strategy.

[0066] Specifically, the corrosion risk assessment report is mapped to a multi-dimensional motion platform. Risk values ​​for each area are extracted from the report, and a risk distribution map is drawn to identify areas with higher corrosion risk, ensuring focused attention on potentially high-risk areas. The density of detection points in high-risk areas is increased based on the risk values. For example, if the area of ​​a high-risk region is 10 cm²... 2 The original detection point density was 1 point / cm². 2 After optimization, the number of points / cm increased to 2. 2This improves detection accuracy. The optimized detection path refers to optimizing the movement routes of the DIC and Raman detection modules based on the coordinates of high-risk areas in the corrosion risk assessment report, making the detection of high-risk areas more efficient and accurate.

[0067] The DIC and Raman detection modules are switched and moved according to the optimized detection path to perform more detailed detection in new high-risk areas, generating a multimodal updated dataset. The multimodal updated dataset is a new dataset generated by re-collecting data. The multimodal updated dataset is compared with the original multimodal dataset in real time to obtain the detection change value, i.e., the difference, representing the degree of deviation between the new and old data. The magnitude of the change value helps determine whether the data needs to be updated. If the change value is greater than a set threshold, it indicates that the previous data may have a large error and needs to be updated. The preset change threshold is a tolerance range set when comparing data. When the detection change value exceeds the preset change threshold, the detection result is considered to have changed significantly, and the dataset needs to be updated. At this point, there is a large deviation between the multimodal updated dataset and the original multimodal dataset. The multimodal dataset is updated based on the multimodal updated dataset. Essentially, if the analysis started with only one data collection, this is equivalent to backtracking and collecting data again to improve accuracy.

[0068] By identifying and analyzing various parameters in the multimodal dataset, key trends in the corrosion process are determined, and the direction of corrosion development is predicted. The multimodal dataset collects data such as Raman spectroscopy and DIC imaging, and identifies trends in parameters such as chemical reaction intensity and surface strain. For example, assuming the intensity of a characteristic peak in the Raman spectrum gradually increases, and the surface strain gradually increases, it indicates that the material is experiencing more corrosive stress. If the surface strain in a high-risk area increases from 0.01% to 0.03% within one week, while the intensity of the characteristic peak of the chemical reaction intensity increases from 1000 to 1500, it indicates that corrosion is intensifying.

[0069] Based on parameter change trends, detection performance analysis is performed on the multimodal update dataset to improve detection performance. Detection strategies are optimized with the goal of improving detection accuracy. Data detection gain information refers to relevant information about improved detection performance obtained after performance analysis of the multimodal dataset, including the improvement in detection capability brought about by increasing detection point density or improving certain detection parameters.

[0070] Based on the data detection gain information, the multimodal dataset is evaluated and optimized, and a detection optimization knowledge base is established. This knowledge base includes how to adjust detection strategies and equipment configurations according to different corrosion states, aiding in the formulation and adjustment of detection strategies. Closed-loop feedback control is implemented based on the detection optimization knowledge base to formulate automatic optimization detection strategies and dynamically adjust the detection process. This involves automatically adjusting parameters such as the detection flow, equipment settings, or sampling density to achieve optimal detection results. For example, when the corrosion change value in a high-risk area reaches a threshold, the parameters of the detection module or the sampling frequency are automatically switched, optimizing the detection path and detection point density. When the multimodal dataset shows accelerated corrosion progress in a certain detection area, the detection path is automatically adjusted, increasing the detection frequency and achieving more efficient detection.

[0071] By feeding the corrosion risk assessment report back to the multi-dimensional motion platform, the detection path is dynamically adjusted based on high-risk areas, thereby avoiding omission of important areas and improving overall detection accuracy. The automatic optimization detection strategy can automatically adjust the detection parameters and path according to the real-time changes in corrosion status, avoiding redundant detection or omission of high-risk areas, thereby improving detection efficiency.

[0072] In summary, the in-situ multimodal detection method for deep-sea high-pressure stress corrosion simulation provided in this application has the following beneficial effects: In-situ optical datasets are obtained by constructing a deep-sea high-pressure stress corrosion environment simulation unit for simulation data acquisition; multimodal datasets are obtained by controlling a multidimensional motion platform to alternately switch detection based on the in-situ optical datasets; the in-situ optical datasets and the multimodal datasets are coupled and analyzed to construct a stress-chemical correlation array for predicting corrosion risks under deep-sea high-pressure stress, and a corrosion risk assessment report is constructed; the corrosion risk assessment report is synchronized to the multidimensional motion platform for detection optimization, the multimodal datasets are updated for closed-loop analysis, and an automatic optimization detection strategy is formulated. In other words, by constructing a deep-sea high-pressure stress corrosion environment simulation unit, combining the coupled analysis of in-situ optical datasets and multimodal datasets, employing alternating detection by a multidimensional motion platform, and determining an automatic optimization detection strategy based on the real-time corrosion risk assessment report, the accuracy and efficiency of material corrosion detection in deep-sea environments are improved.

[0073] Example 2: Based on the same inventive concept as the deep-sea high-pressure stress corrosion simulation method for in-situ multimodal detection in Example 1, this application also provides a deep-sea high-pressure stress corrosion simulation system for in-situ multimodal detection. Please refer to the appendix. Figure 2 The in-situ multimodal detection deep-sea high-pressure stress corrosion simulation system includes: The data acquisition module 11 is used to construct a deep-sea high-pressure stress corrosion environment simulation unit to collect simulated data and obtain an in-situ optical dataset; the switching detection module 12 is used to control the multi-dimensional motion platform to alternately switch detection based on the in-situ optical dataset to obtain a multi-modal dataset; the corrosion risk prediction module 13 is used to couple the in-situ optical dataset with the multi-modal dataset for analysis, construct a stress-chemical correlation array to predict the corrosion risk of deep-sea high-pressure stress, and construct a corrosion risk assessment report; the closed-loop analysis module 14 is used to synchronize the corrosion risk assessment report to the multi-dimensional motion platform for detection optimization, update the multi-modal dataset for closed-loop analysis, and formulate an automatic optimization detection strategy.

[0074] Furthermore, the data acquisition module 11 in the deep-sea high-pressure stress corrosion simulation system for in-situ multimodal detection is also used for: performing in-situ observations of the exterior of the high-pressure reactor through a sapphire optical observation window to generate an in-situ optical observation dataset; controlling a retractable Raman probe to perform in-situ acquisitions inside the high-pressure reactor to obtain in-situ Raman spectra; synchronizing the in-situ optical observation dataset outside the reactor with the in-situ Raman spectra inside the reactor in time to plan a detection sequence; and performing correlation analysis between the in-situ optical observation dataset outside the reactor and the in-situ Raman spectra inside the reactor according to the detection sequence to construct the in-situ optical dataset.

[0075] Furthermore, the data acquisition module 11 in the deep-sea high-pressure stress corrosion simulation system for in-situ multimodal detection is also used for: decomposing the in-situ optical observation dataset outside the vessel to obtain sample surface morphology observation data and dynamic change observation data; performing enhancement analysis on the sample surface morphology observation data according to the dynamic change observation data to obtain optical morphology features; performing baseline correction on the in-situ Raman spectrum inside the vessel, and determining the optochemical features by peak position identification based on the correction results; performing correlation change analysis on the optical morphology features and the optochemical features, and constructing the in-situ optical dataset based on the correlation relationship.

[0076] Furthermore, the switching detection module 12 in the in-situ multimodal detection deep-sea high-pressure stress corrosion simulation system is also used for: installing the DIC detection module and the Raman detection module in parallel on the multi-dimensional motion platform; performing surface feature analysis based on the in-situ optical dataset to determine surface feature distribution data; marking key points according to the surface feature distribution data and dividing multiple key regions according to the marking results; performing focusing analysis based on the Raman detection module to determine Raman focusing parameters; performing field-of-view analysis based on the DIC detection module to determine DIC imaging parameters; and controlling the DIC detection module and the Raman detection module to alternately switch detection across the multiple key regions based on the Raman focusing parameters and the DIC imaging parameters to generate the multimodal dataset.

[0077] Furthermore, the switching detection module 12 in the in-situ multimodal detection deep-sea high-pressure stress corrosion simulation system is also used for: traversing the multiple key regions to read coordinates and determine the key region coordinate map; performing detection analysis based on the Raman focusing parameters and the DIC imaging parameters to determine the target detection parameters; mapping the target detection parameters to the key region coordinate map for motion planning to generate a detection movement path; performing switching analysis on the DIC detection module and the Raman detection module according to the detection movement path to determine an alternating detection scheme; and controlling the DIC detection module and the Raman detection module to perform detection according to the alternating detection scheme to generate the multimodal dataset.

[0078] Furthermore, the corrosion risk prediction module 13 in the deep-sea high-pressure stress corrosion simulation system for in-situ multimodal detection is also used for: synchronizing the in-situ optical dataset with the multimodal dataset in time to generate first synchronized data; transforming the in-situ optical dataset with the multimodal dataset in spatial coordinates to generate second synchronized data; registering and aligning the first synchronized data with the second synchronized data to construct a spatiotemporal sequence feature database; performing dimensionality reduction processing on the spatiotemporal sequence feature database to extract key feature parameters; calculating the parameter correlation strength between stress parameters and chemical parameters based on the key feature parameters; and performing multivariate combination analysis according to the parameter correlation strength to construct a stress-chemical correlation array.

[0079] Furthermore, the corrosion risk prediction module 13 in the in-situ multimodal detection deep-sea high-pressure stress corrosion simulation system is also used for: analyzing the key feature parameters to obtain key stress feature parameters and key chemical feature parameters; performing linear correlation calculations based on the key stress feature parameters and the key chemical feature parameters to obtain the linear correlation strength; performing nonlinear correlation calculations based on the key stress feature parameters and the key chemical feature parameters to obtain the nonlinear correlation strength; and performing grey relational analysis based on the linear correlation strength and the nonlinear correlation strength to obtain the parameter correlation strength.

[0080] Furthermore, the closed-loop analysis module 14 in the in-situ multimodal detection deep-sea high-pressure stress corrosion simulation system is also used for: reading the risk distribution of the corrosion risk assessment report, identifying the coordinates of high-risk areas, wherein the coordinates of the high-risk areas contain risk values; mapping the corrosion risk assessment report to a multidimensional motion platform to increase the density of detection points at the coordinates of the high-risk areas according to the risk values, generating an optimized detection movement path; controlling the DIC detection module and the Raman detection module to perform mobile detection according to the optimized detection movement path, generating a multimodal update dataset; performing real-time comparison between the multimodal update dataset and the multimodal dataset to generate detection change values; and updating the multimodal dataset when the detection change value is higher than a preset change threshold.

[0081] Furthermore, the closed-loop analysis module 14 in the in-situ multimodal detection deep-sea high-pressure stress corrosion simulation system is also used for: identifying parameter changes based on the multimodal dataset to obtain parameter change trends; performing detection performance analysis on the multimodal updated dataset according to the parameter change trends to generate data detection gain information; evaluating and optimizing the multimodal dataset based on the data detection gain information to formulate a detection optimization knowledge base; and performing closed-loop feedback control based on the detection optimization knowledge base to formulate the automatic optimization detection strategy.

[0082] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Figure 1 The in-situ multimodal detection deep-sea high-pressure stress corrosion simulation method and specific examples in Example 1 are also applicable to the in-situ multimodal detection deep-sea high-pressure stress corrosion simulation system in this embodiment. Through the foregoing detailed description of the in-situ multimodal detection deep-sea high-pressure stress corrosion simulation method, those skilled in the art can clearly understand the in-situ multimodal detection deep-sea high-pressure stress corrosion simulation system in this embodiment. Therefore, for the sake of brevity, it will not be described in detail here.

[0083] The above description of the disclosed embodiments enables those skilled in the art to make or use 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 disclosed herein.

[0084] Obviously, those skilled in the art can make several improvements and modifications to this application without departing from the principles of this application, and these improvements and modifications also fall within the protection scope of this application.

Claims

1. An in-situ multimodal detection method for simulating deep-sea high-pressure stress corrosion, characterized in that, include: A deep-sea high-pressure stress corrosion environment simulation unit was constructed to collect simulation data and obtain in-situ optical datasets; Based on the in-situ optical dataset, a multi-dimensional motion platform is controlled to alternately switch detection to obtain a multi-modal dataset; The in-situ optical dataset and the multimodal dataset are coupled and analyzed to construct a stress-chemical correlation array for corrosion risk prediction under deep-sea high-pressure stress, and a corrosion risk assessment report is constructed. The corrosion risk assessment report is synchronized to the multi-dimensional motion platform for detection optimization, the multi-modal dataset is updated for closed-loop analysis, and an automatic optimization detection strategy is formulated. In-situ observations outside the high-pressure reactor are conducted through a sapphire optical observation window, generating an in-situ optical observation dataset. In-situ observations outside the reactor refer to observing internal samples through a sapphire optical observation window from outside the high-pressure reactor. The retractable Raman probe was controlled to perform in-situ data acquisition inside the high-pressure reactor, obtaining the in-situ Raman spectrum inside the reactor; Synchronize the in-situ optical observation dataset outside the vessel with the in-situ Raman spectrum inside the vessel in time, and plan the detection sequence; The in-situ optical observation dataset outside the vessel and the in-situ Raman spectrum inside the vessel are correlated and analyzed according to the detection sequence to construct the in-situ optical dataset; The DIC detection module and the Raman detection module are installed in parallel on the multi-dimensional motion platform; Surface feature analysis is performed based on the in-situ optical dataset to determine the surface feature distribution data. Key points are identified based on the surface feature distribution data, and multiple key regions are divided based on the identification results. The Raman focusing parameters are determined by performing focusing analysis based on the Raman detection module. The DIC imaging parameters are determined by performing field-of-view analysis based on the DIC detection module. Based on the Raman focusing parameters and the DIC imaging parameters, the DIC detection module and the Raman detection module are controlled to alternately switch detection across the multiple key regions to generate the multimodal dataset.

2. The deep-sea high-pressure stress corrosion simulation method for in-situ multimodal detection as described in claim 1, characterized in that, The in-situ optical dataset is constructed by correlating the external in-situ optical observation dataset with the internal in-situ Raman spectrum according to the detection sequence, using the following method: The in-situ optical observation dataset outside the vessel is decomposed to obtain sample surface morphology observation data and dynamic change observation data. The surface morphology observation data of the sample is enhanced by the dynamic change observation data to obtain optical morphology characteristics. Baseline correction was performed on the in-situ Raman spectrum inside the vessel, and the optical chemical characteristics were determined by peak position identification based on the correction results. The optical morphological features and the optical chemical features are correlated and changed, and the in-situ optical dataset is constructed based on the correlation.

3. The deep-sea high-pressure stress corrosion simulation method for in-situ multimodal detection as described in claim 1, characterized in that, Based on the Raman focusing parameters and the DIC imaging parameters, the method controls the DIC detection module and the Raman detection module to alternately switch detection across multiple key regions to generate the multimodal dataset. The coordinates of the multiple key regions are read and the coordinate map of the key regions is determined. Based on the Raman focusing parameters and the DIC imaging parameters, the target detection parameters are determined through detection and analysis. The target detection parameters are mapped to the key area coordinate map for motion planning, generating a detection movement path; The switching analysis between the DIC detection module and the Raman detection module is performed according to the detection movement path to determine the alternating detection scheme; The DIC detection module and the Raman detection module are controlled to perform detection according to the alternating detection scheme to generate the multimodal dataset.

4. The deep-sea high-pressure stress corrosion simulation method for in-situ multimodal detection as described in claim 1, characterized in that, The in-situ optical dataset and the multimodal dataset are coupled for analysis to construct a stress-chemical correlation array. The method includes: The in-situ optical dataset and the multimodal dataset are synchronized in time to generate the first synchronization data; The in-situ optical dataset and the multimodal dataset are transformed in spatial coordinates to generate second synchronization data; The first synchronization data and the second synchronization data are registered and aligned to construct a spatiotemporal sequence feature database; The spatiotemporal sequence feature database is subjected to dimensionality reduction processing to extract key feature parameters; Based on the key characteristic parameters, the correlation strength between stress parameters and chemical parameters is calculated. Multivariate combination analysis is then performed according to the correlation strength to construct a stress-chemical correlation array.

5. The deep-sea high-pressure stress corrosion simulation method for in-situ multimodal detection as described in claim 4, characterized in that, The method for calculating the parametric correlation strength between stress parameters and chemical parameters based on the aforementioned key characteristic parameters includes: Based on the analysis of the key characteristic parameters, key stress characteristic parameters and key chemical characteristic parameters are obtained. The linear correlation strength is obtained by performing a linear correlation calculation based on the key stress characteristic parameters and the key chemical characteristic parameters. The nonlinear correlation strength is obtained by performing nonlinear correlation calculations based on the key stress characteristic parameters and the key chemical characteristic parameters. Grey relational analysis is performed based on the linear correlation strength and the nonlinear correlation strength to obtain the parameter correlation strength.

6. The deep-sea high-pressure stress corrosion simulation method for in-situ multimodal detection as described in claim 1, characterized in that, The corrosion risk assessment report is synchronized to the multidimensional motion platform for detection optimization, the multimodal dataset is updated for closed-loop analysis, and an automatic optimization detection strategy is formulated. The method includes: The corrosion risk assessment report is analyzed to determine the risk distribution and identify the coordinates of high-risk areas, where the coordinates of high-risk areas include risk values. The corrosion risk assessment report is mapped to a multi-dimensional motion platform, and the density of detection points is increased according to the risk value to the coordinates of the high-risk area, thereby generating an optimized detection movement path. The DIC detection module and the Raman detection module are controlled to perform motion detection according to the detection motion optimization path to generate a multimodal update dataset; Based on the real-time comparison between the aforementioned multimodal update dataset and the multimodal dataset, a detected change value is generated; When the detected change value is higher than the preset change threshold, the multimodal dataset is updated.

7. The deep-sea high-pressure stress corrosion simulation method for in-situ multimodal detection as described in claim 6, characterized in that, The multimodal dataset is updated for loop closure analysis, and an automatic optimization detection strategy is developed. The method includes: Based on the multimodal dataset, parameter changes are identified to obtain parameter change trends; The detection performance of the multimodal updated dataset is analyzed according to the parameter change trend to generate data detection gain information; The multimodal dataset is evaluated and optimized based on the data detection gain information, and a detection optimization knowledge base is established. Based on the aforementioned detection optimization knowledge base, closed-loop feedback control is implemented to formulate the automatic optimization detection strategy.

8. A deep-sea high-pressure stress corrosion simulation system for in-situ multimodal detection, characterized in that, The step of implementing the deep-sea high-pressure stress corrosion simulation method for in-situ multimodal detection according to any one of claims 1 to 7, wherein the deep-sea high-pressure stress corrosion simulation system for in-situ multimodal detection comprises: The data acquisition module is used to construct a deep-sea high-pressure stress corrosion environment simulation unit to collect simulation data and obtain in-situ optical datasets. The switching detection module is used to control the multi-dimensional motion platform to alternately switch detection based on the in-situ optical dataset to obtain a multimodal dataset; The corrosion risk prediction module is used to couple and analyze the in-situ optical dataset with the multimodal dataset, construct a stress-chemical correlation array to predict the corrosion risk of deep-sea high-pressure stress, and construct a corrosion risk assessment report. The closed-loop analysis module is used to synchronize the corrosion risk assessment report to the multi-dimensional motion platform for detection optimization, update the multi-modal dataset for closed-loop analysis, and formulate an automatic optimization detection strategy.

Images