Marine cable life prediction and health management method and system based on multi-source data fusion

Abstract

The present application relates to the technical field of marine cable monitoring, and particularly relates to a marine cable life prediction and health management method and system based on multi-source data fusion, which constructs an electrochemical impedance spectrum and a fiber-optic humidity sensor array to synchronously collect multi-source signals under a coupled environment; combines marine environment and working condition parameters to dynamically compensate the signals, eliminate interference and generate a state feature set; uses an electro-chemical coupling field tomographic reconstruction algorithm to construct a spatial corrosion-aging distribution map; determines real defects through spatio-temporal adaptive segmentation and defect dynamics screening; and finally uses a hyperbolic tangent-power law attenuation joint model to calculate corrosion depth and aging rate to realize risk assessment. The present application improves the accuracy of marine cable defect identification and life prediction under complex environments.

Description

Technical Field

[0001] This invention relates to the field of marine cable monitoring technology, and in particular to a method and system for marine cable life prediction and health management based on multi-source data fusion. Background Technology

[0002] With the continuous improvement of ship electrification and intelligence, marine power cables, as a key component of ship power and control systems, directly affect the safety and stability of the entire ship's electrical system. However, during long-term voyages, cables are often subjected to the coupled effects of harsh marine environments such as high salt spray, high humidity, high temperature, and complex electromagnetic interference, leading to multiple degradation behaviors in their insulation materials, including electrochemical corrosion and aging of computer-readable storage media. Traditional cable condition monitoring methods mostly rely on single-parameter detection, such as partial discharge measurement, insulation resistance testing, or temperature monitoring, which are insufficient to comprehensively reflect the integrated degradation mechanism of cables under multi-physical field coupling environments, especially lacking sufficient sensitivity and spatial resolution for early weak corrosion and progressive insulation aging.

[0003] In recent years, multi-source data fusion technology has been gradually applied in the field of equipment health monitoring, improving the comprehensiveness and accuracy of condition perception by integrating multiple sensing modalities. However, existing technologies still have many limitations in the life prediction and health management of marine cables: on the one hand, conventional sensors are difficult to simultaneously respond to the electrochemical characteristics of corrosion processes and the evolution of dielectric properties of insulating materials; on the other hand, marine environmental parameters (such as salt spray deposition and local humidity fluctuations) and operating conditions (such as load current changes) significantly interfere with monitoring signals, leading to distorted feature extraction and a high false alarm rate. In addition, existing methods generally lack the ability to finely reconstruct the spatial distribution characteristics of corrosion and aging defects, making it difficult to accurately locate defects and predict their evolution trends, thus limiting their practical application in life assessment and operation and maintenance decisions.

[0004] The above content is only used to help understand the technical solution of the present invention and does not represent an admission that the above content is prior art. Summary of the Invention

[0005] The main objective of this invention is to provide a method and system for predicting the lifespan and managing the health of marine cables based on multi-source data fusion. This aims to solve the technical problems of existing marine cable health monitoring technologies, which are unable to achieve multi-source collaborative perception, environmental interference suppression, spatial distribution reconstruction, and accurate prediction of degradation trends for corrosion and aging defects under high salt spray, high humidity, and high temperature coupling environments, resulting in low sensitivity for early fault identification and insufficient accuracy for condition assessment.

[0006] To achieve the above objectives, this invention provides a method for predicting the lifespan and managing the health of marine cables based on multi-source data fusion, the method comprising:

[0007] S1. Using an electrochemical impedance spectroscopy sensor and a distributed optical fiber humidity sensor deployed on marine cables, multi-source time-domain signals characterizing the corrosion and aging behavior of cable insulation are collected simultaneously.

[0008] S2. Based on the ship's marine environment parameters and cable load conditions, the multi-source time-domain signals are dynamically compensated to generate a set of compensated state characteristic parameters.

[0009] S3. Based on the state characteristic parameter set, a spatial corrosion-aging distribution map of the cable insulation layer is constructed using an electro-chemical coupled field tomography reconstruction algorithm. The electro-chemical coupled field tomography reconstruction algorithm is an algorithm that integrates electrical and chemical field response data and reconstructs the spatial distribution of corrosion-aging inside the cable insulation layer through inversion calculation.

[0010] S4. Extract defect areas and determine authenticity of the spatial corrosion-aging distribution map to identify the real corrosion defect areas and the real aging defect areas.

[0011] S5. Based on the actual corrosion defect area and the actual aging defect area, a joint degradation model is used to calculate the corrosion depth and insulation aging rate, and a risk assessment is performed in conjunction with a preset safety threshold. The electro-chemical coupled field tomography reconstruction algorithm is an algorithm that integrates electrical and chemical field response data and reconstructs the spatial distribution of corrosion-aging inside the cable insulation layer through inversion calculation.

[0012] Optionally, step S2 specifically includes:

[0013] S21. Real-time collection of salt spray deposition density and micro-environmental humidity data on the surface of marine cables through a distributed environmental sensing network deployed on the cable surface; querying a pre-built environmental parameter-electrochemical disturbance mapping table based on the salt spray deposition density and micro-environmental humidity data to obtain the electrochemical interference compensation coefficient corresponding to the salt spray concentration and environmental humidity value at each cable segment.

[0014] S22. For the time-domain signals collected by each sensor channel, the electrochemical interference compensation coefficient is applied to the amplitude normalization process of the electrochemical impedance signal, and the compensation amount of the effective ion diffusion path is calculated in combination with the design wall thickness parameters of the marine cable insulation layer.

[0015] S23. The normalized time-domain signal is fused with the effective ion diffusion path compensation amount at the data level to generate a compensated state characteristic parameter set including the electrochemical impedance attenuation coefficient, ion migration time constant and humidity diffusion characteristic parameters.

[0016] Optionally, step S3 specifically includes:

[0017] S31. The compensated set of state characteristic parameters is used as an input variable and transmitted to the pre-constructed electrochemical coupling iterative tomographic reconstruction model. The electrochemical coupling iterative tomographic reconstruction model has built-in prior physical constraints based on finite element corrosion simulation analysis.

[0018] S32. By constructing an objective function to minimize the weighted mean square error between the actual measured signal and the model simulation signal, the spatial distribution matrix of the corrosion field and aging field inside the insulation layer of the marine cable is iteratively inverted to solve the spatial corrosion-aging distribution map with high resolution.

[0019] S33. Perform spatiotemporal joint filtering on the spatial corrosion-aging distribution map to effectively suppress artifacts caused by noise interference during the tomographic reconstruction process while preserving the sharpness of the corrosion defect edge features.

[0020] Optionally, step S4 specifically includes:

[0021] S41. Based on the statistical distribution characteristics of the healthy substrate area of ​​marine cables in the spatial corrosion-aging distribution map, a dynamic segmentation threshold is determined. The dynamic segmentation threshold is adjusted spatiotemporally according to the local area standard deviation and the cumulative operating time of the ship.

[0022] S42. Based on the dynamic segmentation threshold, perform a multidimensional connected region labeling algorithm on the spatial corrosion-aging distribution map to obtain an initial set of candidate corrosion defect regions and an initial set of candidate aging defect regions.

[0023] S43. For each candidate region in the initial set of candidate corrosion defect regions, extract its impedance spectrum phase entropy characteristic parameters and time-domain relaxation characteristic parameters; for each candidate region in the initial set of candidate aging defect regions, extract its dielectric loss growth characteristic parameters and humidity hysteresis loop characteristic parameters; and match and verify the extracted characteristic parameters with the pre-constructed defect dynamics characteristic database. When the impedance spectrum phase entropy characteristic parameter is within the preset corrosion characteristic range and the time-domain relaxation characteristic parameter is below the preset threshold, the candidate region is confirmed as a real corrosion defect region; when the dielectric loss growth characteristic parameter shows a continuous growth trend and the humidity hysteresis loop characteristic parameter exceeds the preset creep critical value, the candidate region is confirmed as a real aging defect region.

[0024] Optionally, the process of constructing the defect dynamics feature database includes:

[0025] B1. Collect standard-sized marine cable samples, prepare defect samples with known corrosion depth and insulation aging rate, and accurately measure the impedance spectrum phase entropy characteristic parameters, time-domain relaxation characteristic parameters, dielectric loss growth characteristic parameters, and humidity hysteresis loop characteristic parameters corresponding to each defect sample in a simulated marine environment.

[0026] B2. Based on the corrosion depth and insulation aging rate, establish a hierarchical mapping table of defect dynamic characteristic parameters, wherein micro corrosion defects correspond to the first phase entropy characteristic range, deep corrosion defects correspond to the second phase entropy characteristic range, initial insulation aging corresponds to the first dielectric loss growth range, and accelerated insulation aging corresponds to the second dielectric loss growth range.

[0027] B3. Establish a correlation between the defect dynamic characteristic parameters in the hierarchical mapping table and the critical influence threshold of the remaining service life of the marine cable to form a defect dynamic characteristic database for defect authenticity and severity assessment.

[0028] Optionally, the following steps, performed after step S5, may also be included:

[0029] S6. Based on the spatial location and distribution density of the actual corrosion defect area and the actual aging defect area, generate a heat map of the equipment health status, and evaluate the remaining life coefficient of the marine cable based on the heat map of the equipment health status and the Arrhenius model.

[0030] S7. When the remaining life coefficient is lower than the preset safe operation threshold, an equipment health analysis report is automatically generated, which includes defect spatial distribution characteristic parameters, severity level and corresponding ship operation condition optimization suggestions.

[0031] Optionally, step S1 specifically includes:

[0032] S11. Obtain the insulation material type parameters and design environmental resistance level parameters of the marine cable to be monitored. Based on the insulation material type parameters and design environmental resistance level parameters, query the pre-built material-sensor configuration mapping table to determine the excitation frequency range of the electrochemical impedance spectroscopy sensor and the spatial sampling frequency of the fiber optic humidity sensor suitable for this type of marine cable.

[0033] S12. Under the excitation frequency range and spatial sampling frequency, control the multi-modal sensor array to scan the cable to be monitored synchronously, and capture the electrochemical transient response characteristics caused by corrosion and the dynamic change sequence of insulation humidity caused by environmental fluctuations in real time through the high-speed data acquisition system.

[0034] Optionally, the sampling rate of the electrochemical impedance spectroscopy sensor is configured to match the typical response timescale of the electrochemical reaction of metal corrosion in marine cable insulation; the humidity sensitivity of the distributed fiber optic humidity sensor is configured to match the moisture diffusion rate in the early stages of insulation material aging.

[0035] Furthermore, to achieve the above objectives, the present invention also provides a marine cable life prediction and health management system based on multi-source data fusion, the system comprising:

[0036] The sensor array module is used to simultaneously acquire multi-source time-domain signals characterizing the corrosion and aging behavior of the cable insulation layer by utilizing an electrochemical impedance spectroscopy sensor and a distributed fiber optic humidity sensor deployed on marine cables.

[0037] The signal compensation module is used to dynamically compensate the multi-source time-domain signal based on the ship's marine environment parameters and cable load conditions, and generate a set of compensated state feature parameters.

[0038] The tomographic reconstruction module is used to construct a spatial corrosion-aging distribution map of the cable insulation layer based on the state feature parameter set and through an electro-chemical coupled field tomographic reconstruction algorithm. The electro-chemical coupled field tomographic reconstruction algorithm is an algorithm that integrates electrical and chemical field response data and reconstructs the spatial distribution of corrosion-aging inside the cable insulation layer through inversion calculation.

[0039] The defect identification module is used to extract defect areas and distinguish between genuine and fake data from the spatial corrosion-aging distribution map, and to determine the real corrosion defect areas and the real aging defect areas.

[0040] The risk assessment module is used to calculate the corrosion depth and insulation aging rate based on the actual corrosion defect area and the actual aging defect area using a joint degradation model, and to conduct risk assessment in conjunction with a preset safety threshold. The joint degradation model is a composite mathematical model that combines the hyperbolic tangent function to describe the initial rapid degradation stage and the power law function to describe the long-term gradual degradation stage, and is used to perform unified quantitative modeling of corrosion depth and insulation aging rate.

[0041] This invention provides a method for predicting the lifespan and managing the health of marine cables based on multi-source data fusion. The method constructs a multi-modal sensing array comprising an electrochemical impedance spectroscopy sensor and a distributed fiber optic humidity sensor. This enables multi-physics collaborative sensing of the insulation corrosion process and dielectric aging behavior of marine cables under high salt spray, high humidity, and high temperature coupling environments. It can simultaneously capture the time-frequency evolution characteristics of electrochemical impedance and the gradient changes in dielectric properties, significantly improving the detection sensitivity of early degradation characteristics. By introducing marine environmental parameters and real-time cable load parameters, dynamic compensation processing of multi-source signals effectively eliminates the interference of complex environmental disturbances on monitoring data, improving... The accuracy and robustness of state feature extraction are improved. Furthermore, an electro-chemical coupled field tomography reconstruction algorithm is used to generate a spatial corrosion-aging distribution map, achieving high-resolution visual reconstruction of the internal defect distribution of the cable insulation layer, overcoming the shortcomings of traditional methods in spatial positioning capabilities. Combining spatiotemporal adaptive threshold segmentation and a true / false discrimination mechanism based on defect dynamics characteristics, the true corrosion and aging defect areas can be accurately identified, reducing the false detection rate. Finally, a hyperbolic tangent-power-law decay joint model is used to quantitatively analyze key degradation parameters, achieving joint modeling and risk criterion fusion of corrosion depth and insulation aging rate, improving the scientific rigor of cable remaining life prediction and the intelligence level of operation and maintenance decisions. In summary, this invention solves the problems of weak multi-source information fusion capability, poor environmental adaptability, low defect identification accuracy, and insufficient reliability of life assessment in existing technologies, significantly enhancing the full life-cycle health management level of marine cables in harsh marine environments. Attached Figure Description

[0042] Figure 1 This is a flowchart illustrating an embodiment of the marine cable life prediction and health management method based on multi-source data fusion of the present invention.

[0043] Figure 2 This is a schematic diagram of the specific process of step S2 in one embodiment of the marine cable life prediction and health management method based on multi-source data fusion of the present invention.

[0044] Figure 3 This is a schematic diagram of the specific process of step S3 in one embodiment of the marine cable life prediction and health management method based on multi-source data fusion of the present invention.

[0045] Figure 4 This is a structural block diagram of an embodiment of the marine cable life prediction and health management system based on multi-source data fusion of the present invention.

[0046] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0047] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0048] Reference Figure 1 , Figure 1 This is a flowchart illustrating an embodiment of the marine cable life prediction and health management method based on multi-source data fusion according to the present invention.

[0049] In one embodiment, a method for predicting the lifespan and health management of marine cables based on multi-source data fusion is provided, the method comprising:

[0050] Step S1: Construct a multimodal sensor array for the marine cable to be monitored. The multimodal sensor array includes an electrochemical impedance spectroscopy sensor and a distributed fiber optic humidity sensor. The multimodal sensor array is used to synchronously acquire multi-source time-domain signal sequences. These sequences characterize the time-frequency evolution of electrochemical impedance caused by insulation corrosion in marine cables operating under high salt spray, high humidity, and high temperature coupled environments, as well as the gradient changes in dielectric properties of the insulation material caused by water molecule penetration.

[0051] The multimodal sensor array can be an integrated sensing system composed of various types of sensors, used to synchronously acquire time-series response signals of different physical quantities. It can be used to achieve multi-physics collaborative sensing of the corrosion and aging behavior of marine cables' insulation layers under high salt spray, high humidity, and high temperature coupling environments. The electrochemical impedance spectroscopy (EIS) sensor can be a sensing device that characterizes the electrochemical state of material interfaces by applying broadband AC excitation and measuring the system impedance response. It can be used to capture the time-frequency evolution characteristics of the electrochemical impedance of cable insulation layers caused by corrosion, reflecting early, weak electrochemical degradation processes. In one embodiment, the EIS sensor can calculate the complex impedance spectrum by applying a small-amplitude AC voltage with a frequency scan between the cable conductor and the shielding layer and measuring the current response. The distributed fiber optic humidity sensor can be a continuous humidity sensing device based on the sensitivity of optical signals in optical fibers to local environmental humidity. It can be used to monitor in real time the changes in dielectric properties caused by water molecule penetration along the cable length, providing spatially resolved humidity distribution information. For example, a distributed fiber optic humidity sensor can invert local humidity by detecting the intensity or wavelength shift of backscattered light using fiber Bragg gratings (FBGs) or Raman / Brillouin scattering principles.

[0052] Multi-source time-domain signal sequences can be a collection of raw data of multidimensional physical quantities that vary over time and are synchronously acquired by a multi-modal sensor array. These sequences can be used to carry raw sensing information about the corrosion and aging behavior of cable insulation layers in complex marine environments. In one specific embodiment, the multi-source time-domain signal sequence can periodically sample electrochemical impedance spectroscopy (EIS) sensors and distributed fiber optic humidity sensors through a synchronous triggering mechanism, forming a time-aligned data stream. The time-frequency evolution characteristics of electrochemical impedance spectroscopy can be the dynamic pattern of the complex impedance of cable insulation materials changing with frequency and time during corrosion. These characteristics can be used to characterize the stage-specific features of the electrochemical corrosion process, such as the rupture of the interfacial passivation film and the formation of ion channels. The dielectric property gradient change characteristics can be the spatial gradient distribution of the local dielectric constant or loss tangent caused by the non-uniform penetration of water molecules along the radial or axial direction of the cable. These characteristics can reflect the degree of moisture absorption aging of the insulation material and its spatial non-uniformity.

[0053] Constructing a multimodal sensor array for monitoring marine cables can be achieved by deploying electrochemical impedance spectroscopy sensors and distributed fiber optic humidity sensors within the cable body according to a predetermined topology. Furthermore, this array can be constructed by embedding miniature impedance probes inside the cable's outer sheath and simultaneously laying moisture-sensitive optical fibers along the cable core, thus establishing the hardware foundation for synchronously sensing electrochemical and humidity responses. The synchronous acquisition of multi-source time-domain signal sequences using the multimodal sensor array can be achieved by triggering periodic data acquisition from both types of sensors using a unified clock. For example, the synchronous acquisition of multi-source time-domain signal sequences using the multimodal sensor array can be implemented using nanosecond-level synchronous sampling control via an FPGA, thereby obtaining time-aligned electrochemical and humidity response data and supporting multiphysics correlation analysis.

[0054] Step S2: Obtain marine environmental parameters for ship navigation and real-time load condition parameters for cables. Based on these parameters, dynamically compensate the multi-source time-domain signal sequence to eliminate measurement interference caused by fluctuations in salt spray concentration and changes in ambient temperature and humidity, and generate a compensated set of state characteristic parameters. Marine environmental parameters for ship navigation can be a set of external variables describing the marine environment in which the ship is located. These parameters can be used to compensate for interference from environmental factors such as salt spray, temperature, and humidity on sensor signals. Further, marine environmental parameters may include, but are not limited to, one or more of the following: salt spray deposition rate, cabin relative humidity, and ambient temperature fluctuations. Real-time load condition parameters for cables can be internal variables reflecting the current electrical operating state of the cable. These parameters can be used to correct the effects of thermal effects and electromagnetic interference caused by changes in load current on sensor signals. For example, real-time load condition parameters may include, but are not limited to, one or more of the following: effective value of load current, harmonic distortion rate, and instantaneous power fluctuations.

[0055] Dynamic compensation processing is a signal preprocessing operation that corrects the original sensor signal in real time based on environmental and operating parameters. It can eliminate signal drift caused by non-degradation factors, improving the accuracy and robustness of feature extraction. The compensated state feature parameter set can be a set of multi-dimensional feature vectors that retain true degradation information after dynamic compensation processing. This compensated state feature parameter set can serve as reliable input data for subsequent tomographic reconstruction and defect identification.

[0056] Obtaining marine environmental parameters and real-time load condition parameters of cables during ship navigation can be achieved by reading real-time data from the ship's environmental monitoring system and power monitoring system. Furthermore, this can be accomplished by acquiring cabin temperature and humidity and salt spray concentration data from the ship's automation system via CAN bus, or by reading cable load current and harmonic data from intelligent circuit breakers or power quality analyzers, thus providing an external disturbance reference for dynamic compensation. Dynamic compensation processing of multi-source time-domain signal sequences based on marine environmental parameters and real-time operating condition parameters can be achieved by establishing a mapping model of environment-operating condition-signal interference and performing inverse correction on the original signal. For example, dynamic compensation processing of multi-source time-domain signal sequences based on marine environmental parameters and real-time operating condition parameters can be achieved by constructing a compensation coefficient matrix using a multiple linear regression model, or by using a neural network to learn the nonlinear relationship between environmental disturbances and signal offset, thereby suppressing signal distortion caused by non-degradation factors and improving feature authenticity. Generating a set of compensated state feature parameters can be achieved by extracting time-frequency domain features, gradient features, etc., from the compensated signal to form a feature vector. Furthermore, the generated set of compensated state feature parameters can form standardized input data suitable for tomographic reconstruction.

[0057] Step S3: Based on the compensated state feature parameter set, the spatial corrosion-aging distribution map of the marine cable insulation layer is constructed using the electro-chemical coupled field tomography reconstruction algorithm. The spatial corrosion-aging distribution map reflects the response characteristics of the cable insulation material under the coupling effect of multiple physical fields.

[0058] The electro-chemical coupled-field tomography reconstruction algorithm can be a mathematical model that integrates electrical and chemical field response data to reconstruct the spatial distribution of corrosion-aging within the cable insulation layer through inversion calculations. This algorithm can be used to achieve high-resolution three-dimensional visualization of defect distribution. In one specific embodiment, the algorithm can map impedance and humidity data from boundary measurements to an internal spatial grid based on a finite element forward model and an iterative optimization inversion strategy. The spatial corrosion-aging distribution map can be a two-dimensional or three-dimensional image that quantifies the degree of corrosion and aging level within the cable insulation layer, using spatial coordinates as a reference. This map can visually display the spatial location, extent, and severity of defects.

[0059] The response characteristics under multi-physics coupling can represent the comprehensive degradation behavior of cable insulation materials under the combined effects of electric, chemical, thermal, and wet fields. These response characteristics can serve as physical constraints for tomographic reconstruction algorithms, ensuring that the reconstruction results conform to actual physical mechanisms. Constructing a spatial corrosion-aging distribution map of marine cable insulation using an electro-chemical coupled-field tomographic reconstruction algorithm can be achieved by inputting compensated feature parameters into an inversion model to solve for the internal spatial degradation distribution. Furthermore, constructing this spatial corrosion-aging distribution map using the electro-chemical coupled-field tomographic reconstruction algorithm can employ a finite element forward modeling + Landweber iterative inversion process, or by training a convolutional neural network to directly map boundary data to the internal image, thereby achieving high-resolution spatial visualization of defects.

[0060] Step S4: Perform spatiotemporal adaptive threshold segmentation on the spatial corrosion-aging distribution map to obtain a set of candidate regions for corrosion defects and aging defects. Then, based on the preset defect dynamics feature screening rules, perform a true / false judgment on the set of candidate defect regions to determine the true corrosion defect region and the true aging defect region.

[0061] Spatiotemporally adaptive thresholding segmentation is an image processing method that dynamically adjusts the segmentation threshold based on temporal and spatial dimensions to extract potential defect regions. It can be used to effectively separate true degradation regions from random fluctuations in noisy backgrounds. The candidate region set for corrosion and aging defects can be a preliminary set of spatial regions that may contain true degradation, identified after thresholding segmentation. This candidate region set can be used to provide targets for subsequent authenticity determination.

[0062] The preset defect kinetic feature screening rules can be a set of discrimination criteria based on the physical laws of corrosion and aging processes. These rules can be used to distinguish between real defects and transient interference or measurement noise. Real corrosion defect areas can be regions of insulation layer damage confirmed by kinetic feature verification as caused by electrochemical corrosion. These areas can serve as the spatial basis for calculating corrosion depth in life assessment. Real aging defect areas can be regions of performance degradation confirmed by kinetic feature verification as caused by hygroscopic aging of the medium. These areas can serve as the spatial basis for calculating aging rate in life assessment.

[0063] Spatiotemporally adaptive thresholding segmentation of the spatial corrosion-aging distribution map can be performed by dynamically setting the segmentation threshold based on local pixel intensity and its temporal evolution. Furthermore, this segmentation can be achieved using the Otsu method combined with a sliding time window to calculate the locally optimal threshold, or by using a Markov random field model for spatial context-adaptive segmentation. This allows for the extraction of potential defect regions and avoids over-segmentation or under-segmentation caused by a fixed threshold. Obtaining a candidate region set for corrosion and aging defects can be achieved by marking connected regions in the segmentation results as candidate defects. Further, this candidate region set can form a list of defect regions to be verified.

[0064] The set of candidate defect regions is evaluated for authenticity based on pre-defined defect kinetic feature screening rules. This can involve checking whether each candidate region conforms to the physical evolution laws of corrosion or aging. For example, this evaluation can verify that the region area monotonically increases over time at a rate conforming to the Arrhenius relation, or check whether the direction of the median elevation points towards the water molecule penetration path, thus eliminating false regions caused by noise or transient interference. Determining the true corrosion defect region and the true aging defect region can involve retaining the candidate regions that pass kinetic verification as the final defect result. Furthermore, determining the true corrosion defect region and the true aging defect region can output a high-confidence defect spatial identifier.

[0065] Step S5: For the actual corrosion defect area and the actual aging defect area, the hyperbolic tangent-power law attenuation joint model is used to calculate the corrosion depth parameter, insulation aging rate parameter and spatial distribution density parameter corresponding to each defect. When the corrosion depth parameter of any defect exceeds the preset safety threshold or the insulation aging rate parameter exceeds the preset aging critical threshold, it is determined that the marine cable has a corrosion or insulation aging risk that affects the safe operation of the ship's power system.

[0066] The hyperbolic tangent-power-law decay joint model can be a composite mathematical model that combines the hyperbolic tangent function to describe the initial rapid degradation stage with the power-law function to describe the long-term gradual degradation stage. This model can be used for unified quantitative modeling of corrosion depth and insulation aging rate. In a specific embodiment, the hyperbolic tangent-power-law decay joint model can be achieved by fitting measured degradation parameters to a form such as... The function form, where, The health degradation index of the cable corresponds to the corrosion depth parameter or insulation aging rate parameter in the specific calculation; This represents the service life of marine cables; Corresponding hyperbolic tangent part The saturation limit value is used to describe the maximum rapid degradation rate of a cable caused by sudden environmental changes in the early stages of degradation. The shape control factor corresponding to the hyperbolic tangent is used to control the rate of change in the initial degradation stage and reflects the severity of degradation in the initial stage. Corresponding power law part The proportionality coefficient is used to describe the basic degradation rate constant caused by the cumulative effects of material aging and continuous environmental influences during long-term service. The attenuation index, corresponding to the power-law part, is used to describe the nonlinear evolution characteristics of cable aging during long-term service, that is, the trend of degradation accelerating or slowing down over time.

[0067] For example, assuming the model parameters of a marine cable are set as a = 0.15, b = 0.8, c = 0.01, and d = 0.5, the corrosion depth characteristic value after 4 years of service is calculated. First, the hyperbolic tangent part is calculated by substituting time t into 0.15 multiplied by tanh (0.8 multiplied by 4), i.e., 0.15 multiplied by tanh (3.2), resulting in approximately 0.1495, which represents the contribution of the initial rapid degradation stage. Next, the power-law part is calculated by multiplying 0.01 by 4 to the power of 0.5, resulting in 0.02, reflecting the long-term cumulative aging degree. The two parts are added together to obtain the total corrosion depth characteristic value of 0.1695. If the preset safety threshold is set to 0.18, since the calculated result 0.1695 is less than the threshold, the cable is determined to be in a safe operating state. Conversely, if the calculated result exceeds this value, the system will determine that there is a risk of insulation corrosion and issue a warning.

[0068] Corrosion depth parameters can be quantitative indicators characterizing the radial penetration depth of corrosion defects along the insulation layer. Corrosion depth parameters can be used to determine whether the insulation is approaching a critical breakdown state. Insulation aging rate parameters can be the rate of change of dielectric properties deterioration per unit time. Insulation aging rate parameters can be used to reflect the speed of the aging process and support the estimation of remaining life. Spatial distribution density parameters can be statistical measures of the number or severity of defects per unit length or unit area. Spatial distribution density parameters are used to assess the concentration and systemic risks of degradation. Preset safety thresholds can be upper limits of corrosion depth set based on the cable insulation design margin. Preset safety thresholds can be used as criteria to determine whether corrosion risk endangers power system safety. Preset aging critical thresholds can be warning values ​​for insulation aging rates determined based on material failure experiments. Preset aging critical thresholds can be used as criteria to determine whether aging has entered an irreversible accelerated stage. Corrosion or insulation aging risk can be a safety warning state triggered when any critical degradation parameter exceeds the corresponding threshold. Corrosion or insulation aging risk can be used to drive operation and maintenance decisions, such as maintenance, replacement, or load reduction.

[0069] The corrosion depth, insulation aging rate, and spatial distribution density parameters corresponding to each defect are calculated using a hyperbolic tangent-power-law attenuation joint model. This allows for the fitting of quantitative indicators of the actual defect area to the joint model parameters. Furthermore, the calculation of these parameters using the hyperbolic tangent-power-law attenuation joint model can be achieved by fitting the model parameters using nonlinear least squares or by updating the model parameters online using particle filtering to adapt to new data, thus enabling a unified mathematical representation of the degradation degree. Determining whether the marine cable poses a corrosion or insulation aging risk affecting the safe operation of the ship's electrical system can be achieved by comparing the calculated parameters with preset thresholds and triggering a risk alarm. Furthermore, determining whether the marine cable poses a corrosion or insulation aging risk affecting the safe operation of the ship's electrical system can provide a clear basis for operation and maintenance decisions.

[0070] Taking the health monitoring of the main power distribution cable of an ocean-going cargo ship as an example, the marine cable life prediction and health management method based on multi-source data fusion in this embodiment can be implemented by deploying a multi-modal sensor array on the main generator feeder cable to simultaneously collect electrochemical impedance spectroscopy and distributed humidity data during transit through the high-humidity and high-salinity equatorial sea. The system can also acquire the engine room temperature and humidity, salt spray concentration, and load current in real time and dynamically compensate the original signals. Through electrochemical coupling field tomography reconstruction, a local corrosion hotspot was found 2.3 meters away from the junction box. Through spatiotemporal adaptive segmentation and dynamic verification, it was confirmed that the area was a real corrosion defect. The hyperbolic tangent-power law model calculation showed that the corrosion depth had reached 85% of the safety threshold. The system issued a level three warning and recommended that the cable section be prioritized for maintenance after berthing.

[0071] In one embodiment, reference Figure 2 Step S2 specifically includes:

[0072] S21. Real-time collection of salt spray deposition density and micro-environmental humidity data on the surface of marine cables through a distributed environmental sensing network deployed on the cable surface. Based on the salt spray deposition density and micro-environmental humidity data, query the pre-built environmental parameter-electrochemical disturbance mapping table to obtain the electrochemical interference compensation coefficient corresponding to the salt spray concentration and environmental humidity value at each cable segment.

[0073] The distributed environmental sensing network can be a sensor network deployed on the cable surface for real-time monitoring of local marine microenvironment parameters. It can acquire salt spray deposition density and microenvironmental humidity directly in contact with different cable segments, providing high spatial resolution disturbance source data for electrochemical interference compensation. In this embodiment, the distributed environmental sensing network can integrate miniature salt spray deposition sensors and humidity-sensitive elements on the cable outer sheath surface, arranged at fixed intervals to form a continuous sensing chain. The salt spray deposition density on the cable surface can be the mass of chloride salts accumulated per unit area of ​​the cable surface, which can be used to characterize the concentration of local corrosive media and directly affect the background noise level of electrochemical impedance spectroscopy measurements. The microenvironmental humidity data can be the local relative or absolute humidity value of the air close to the cable outer surface, which can reflect the driving force strength of water molecule penetration into the insulation layer, affecting dielectric response and ion migration rate.

[0074] The environmental parameter-electrochemical disturbance mapping table can be a lookup function established through experiments or simulations, relating the combination of salt spray concentration and humidity to the corresponding electrochemical measurement deviation. This table can be used to quickly convert measured environmental parameters into electrochemical interference coefficients suitable for signal compensation. Furthermore, the environmental parameter-electrochemical disturbance mapping table can be constructed by applying different salt spray-humidity combinations to standard cable samples in a simulation chamber, recording their impact on impedance spectrum amplitude, and building a multidimensional lookup table. The salt spray concentration at each cable segment can be the local equivalent chloride ion concentration obtained by inverting the salt spray deposition density, and can be used as one of the input variables for the mapping table to determine the interference compensation intensity for that segment. The ambient humidity value can be a numerical representation corresponding to the microenvironment humidity data, and can be used together with the salt spray concentration to form a two-dimensional lookup index for the mapping table. The electrochemical interference compensation coefficient can be a scaling factor used to correct the electrochemical impedance amplitude shift caused by local salt spray and humidity, and can be used to normalize the amplitude of the original impedance signal, eliminating spurious fluctuations caused by non-degradation factors.

[0075] Real-time acquisition of salt spray deposition density and micro-environmental humidity data on the surface of marine cables is achieved through a distributed environmental sensing network deployed on the cable surface. This can be accomplished by continuously monitoring local salt spray and humidity using surface-integrated micro-sensors. Furthermore, this operation can be achieved by printing interdigital electrode arrays onto the cable sheath surface using printed electronics technology, inverting the salt spray deposition amount using AC impedance spectroscopy, or by embedding a micro-temperature and humidity module and combining it with a Fick diffusion model to estimate the surface water film thickness. This allows for the acquisition of high spatial resolution disturbance source data, supporting segmented and refined compensation. A pre-built environmental parameter-electrochemical disturbance mapping table is then queried based on the salt spray deposition density and micro-environmental humidity data. This can be done by using measured salt spray density and humidity as indexes to retrieve the corresponding compensation coefficients from the pre-stored mapping table. Furthermore, this operation can be achieved by using bilinear interpolation to obtain continuous compensation values ​​from a discrete mapping table, or by calling a lightweight neural network model to predict compensation coefficients online, thus enabling rapid mapping of environmental disturbances to electrochemical deviations. The electrochemical interference compensation coefficients corresponding to the salt spray concentration and ambient humidity values ​​at each cable segment can be obtained. The compensation coefficients can be independently assigned to each sensing segment to support differentiated signal correction in spatially heterogeneous environments.

[0076] S22. For the time-domain signals collected by each sensor channel, the electrochemical interference compensation coefficient is applied to the amplitude normalization process of the electrochemical impedance signal, and the compensation amount of the effective ion diffusion path is calculated in combination with the design wall thickness parameters of the marine cable insulation layer.

[0077] The time-domain signal acquired by the sensor channel can be the raw time-series output from a specific sensing unit (such as a segment of electrochemical impedance probe), which can be used as the basic data unit for compensation processing and normalized independently by channel. The amplitude normalization process of the electrochemical impedance signal can be a mathematical operation of dividing the original impedance amplitude by the corresponding electrochemical interference compensation coefficient, which can be used to recover the intrinsic amplitude of the signal and improve the comparability of data under different environmental conditions. The design wall thickness parameter of the marine cable insulation layer can be the nominal radial thickness of the insulation material specified during cable manufacturing, which can be used to calculate the effective path length required for ions to diffuse from the surface to the conductor, correcting the impact of structural geometric differences on aging assessment. The effective ion diffusion path compensation amount can be a geometric correction factor derived based on the design wall thickness parameter, used to correct the ion migration time constant, which can be used to eliminate diffusion path differences caused by different cable models or batches, enhancing the consistency of aging rate assessment.

[0078] Applying the electrochemical interference compensation coefficient to the amplitude normalization process of the electrochemical impedance signal can be achieved by dividing the original impedance amplitude of each channel by its corresponding compensation coefficient. Furthermore, this operation can be performed independently at each frequency point in the frequency domain, or by scaling the envelope signal as a whole in the time domain, thereby suppressing spurious amplitude fluctuations caused by salt spray and improving signal fidelity. Calculating the compensation amount for the effective ion diffusion path by combining the design wall thickness parameters of the marine cable insulation layer can be achieved by substituting the design wall thickness into the analytical solution or numerical model of Fick's second law to derive the diffusion path correction factor. Furthermore, this operation can be achieved by assuming one-dimensional steady-state diffusion that makes the compensation amount proportional to the square of the wall thickness, or by establishing a nonlinear mapping between wall thickness and time constant based on finite element simulation, thereby correcting ion migration time deviations caused by differences in cable structure.

[0079] S23. The normalized time-domain signal is fused with the effective ion diffusion path compensation amount at the data level to generate a compensated state characteristic parameter set containing the electrochemical impedance attenuation coefficient, ion migration time constant and humidity diffusion characteristic parameters.

[0080] The normalized time-domain signal can be electrochemical impedance time-domain data with normalized amplitude, which can be used as one of the inputs for data fusion, preserving true degradation characteristics. The electrochemical impedance attenuation coefficient can be a quantitative parameter characterizing the rate of impedance amplitude decay with frequency or time, reflecting the degree of degradation in interfacial charge transfer capability and used for corrosion process modeling. The ion migration time constant can be a physical quantity describing the characteristic time required for ions to traverse the insulating layer under electric field drive, which can be used to correlate aging depth with dielectric degradation, supporting lifetime prediction. The humidity diffusion characteristic parameter can be the water molecule diffusion rate or gradient index obtained from distributed fiber optic humidity sensor data, which can be used to characterize the hygroscopic aging kinetics, complementing the electrochemical parameters. The compensated state characteristic parameter set can be a high-dimensional feature set generated by fusing the normalized signal and diffusion path compensation, containing the electrochemical impedance attenuation coefficient, ion migration time constant, and humidity diffusion characteristic parameters, providing high-quality input for subsequent tomographic reconstruction.

[0081] Data-level fusion of the normalized time-domain signal and the effective ion diffusion path compensation can be achieved by jointly encoding the normalized signal features and the diffusion path compensation into a unified feature vector. Furthermore, this operation can be implemented by concatenating the compensation as an additional channel to the signal feature matrix, or by scaling the ion migration time constant using the compensation, thereby achieving synergistic elimination of both environmental and structural interferences. Generating a compensated set of state feature parameters, including the electrochemical impedance attenuation coefficient, ion migration time constant, and humidity diffusion characteristic parameters, can be achieved by extracting three key degradation indices from the fused data to form a feature set, resulting in a high-dimensional state characterization that integrates electrochemistry, transport kinetics, and humidity response.

[0082] For example, in the scenario of health assessment of main propulsion motor feeder cables during transoceanic voyages, the marine cable life prediction and health management method based on multi-source data fusion in this embodiment could be: when a ship passes through a high-salt-fog sea area, a distributed environmental sensor network deployed on the cable surface detects a sudden increase in salt fog deposition density to 1.8 mg / cm³ in the 5th segment. 2 The microenvironment humidity reached 92%; the system queried the environmental parameter-electrochemical disturbance mapping table and obtained the electrochemical interference compensation coefficient of 1.35 for this segment; the amplitude of the electrochemical impedance signal of this segment was normalized, and the effective ion diffusion path compensation was calculated in combination with the 0.8mm insulation wall thickness of the cable model; the state characteristic parameter set generated after fusion showed that the ion migration time constant was abnormally shortened, triggering subsequent chromatographic reconstruction, and finally confirmed the existence of early water tree aging risk.

[0083] In one embodiment, reference Figure 3 Step S3 specifically includes:

[0084] S31. The compensated set of state characteristic parameters is used as input variables and transmitted to the pre-constructed electro-chemical coupling iterative tomographic reconstruction model. The electro-chemical coupling iterative tomographic reconstruction model has built-in prior physical constraints based on finite element corrosion simulation analysis.

[0085] The electro-chemical coupling iterative tomographic reconstruction model can be a computational model that integrates electrical and chemical field responses and iteratively optimizes the inversion of the internal degradation distribution of the cable. It can be used to map the compensated set of state characteristic parameters into a high-resolution spatial corrosion-aging distribution map. In this embodiment, the electro-chemical coupling iterative tomographic reconstruction model can be constructed based on forward finite element simulation and inverse optimization algorithms, with built-in physical constraints guiding the iterative solution process. The prior physical constraints based on finite element corrosion simulation analysis can be physical regularities obtained from the multiphysics evolution of corrosion and aging through finite element simulation, encoded as constraint terms in the inversion optimization. This can be used to improve the physical rationality of the inversion solution and avoid non-physical solutions or overfitting. In an exemplary embodiment, the prior physical constraints based on finite element corrosion simulation analysis can be used to establish a corrosion diffusion model of the cable insulation layer under the coupling of salt spray, humidity, and electric field in a multiphysics simulation platform, and extract typical degradation modes as priors. For example, the prior physical constraints based on finite element corrosion simulation analysis may include constraints on the continuity of ion concentration gradient, upper limit constraints on the corrosion front advance rate, and monotonic mapping constraints between dielectric constant and water content.

[0086] The compensated set of state feature parameters is used as input variables and transmitted to a pre-constructed electrochemical coupling iterative tomographic reconstruction model. This can be achieved by loading the feature vectors output from previous steps into the input interface of the reconstruction model. Furthermore, this operation can initiate a high-fidelity spatial reconstruction process. The electrochemical coupling iterative tomographic reconstruction model incorporates prior physical constraints based on finite element corrosion simulation analysis. This can be achieved by transforming the degradation laws obtained from finite element simulation into regularization terms or feasible region constraints in the optimization problem. In one specific embodiment, this operation can be implemented by embedding the upper limit of the corrosion front velocity as an inequality constraint into the optimizer; alternatively, simulation data can be used to train a discriminator in a generative adversarial network (GAN) as a physical rationality criterion, thereby enhancing the physical consistency of the inverted solution and improving the stability of ill-conditioned problems.

[0087] S32. By constructing an objective function to minimize the weighted mean square error between the actual measured signal and the model simulation signal, the spatial distribution matrix of the corrosion field and aging field inside the insulation layer of the marine cable is iteratively inverted to solve the spatial corrosion-aging distribution map with high resolution.

[0088] The objective function can be a mathematical expression used to quantify the difference between the actual measured signal and the model simulation signal, guiding the iterative inversion towards minimizing signal mismatch. For example, the objective function can include a weighted L2 norm objective function, a Huber robust loss function, or a multi-task joint objective function. The actual measured signal can be multi-source sensor data obtained after dynamic compensation processing, serving as the observation input for inversion and providing realistic boundary conditions to constrain the model output. The model simulation signal can be the boundary response calculated by the forward model based on the assumed internal degradation distribution in the current iteration state, used for comparison with the actual measured signal to drive inversion updates. The weighted mean square error can be a mean square error metric assigning different weights to different frequencies or sensor channels, used to emphasize the fitting accuracy of high signal-to-noise ratio bands or key sensor channels, improving inversion stability. The spatial distribution matrix of the corrosion and aging fields inside the marine cable insulation layer can be a two-dimensional / three-dimensional numerical matrix representing the corrosion degree and aging level on the cable cross-section or axial profile in discrete grid form, serving as a direct data basis for generating a spatial corrosion-aging distribution map.

[0089] By constructing an objective function to minimize the weighted mean square error between the actual measured signal and the model simulation signal, the loss function can be defined as a weighted sum of the squared errors of each channel, where the weights reflect the importance of the channels. Further, this operation can be achieved by adaptively assigning weights according to the signal-to-noise ratio (SNR), giving higher weights to channels with high SNR; in a specific embodiment, frequency-selective weights can also be introduced to focus on degradation-sensitive frequency bands, thereby achieving optimal matching between measurement and simulation and driving inversion convergence. Iteratively inverting and solving the spatial distribution matrix of the corrosion and aging fields within the marine cable insulation layer can involve repeatedly executing a forward simulation-error calculation-parameter update loop until convergence. Exemplarily, this operation can use the conjugate gradient method to accelerate the solution of linear subproblems; in an exemplary embodiment, the Levenberg-Marquardt algorithm can also be used to handle nonlinear inversion, thereby reconstructing the internal degradation state from the boundary signals. Generating a high-resolution spatial corrosion-aging distribution map can be achieved by visualizing the converged spatial distribution matrix as a heatmap or pseudo-color image. Further, this operation can provide an intuitive, high-resolution view of the defect distribution.

[0090] S33. Perform spatiotemporal joint filtering on the spatial corrosion-aging distribution map to effectively suppress artifacts caused by noise interference during tomographic reconstruction while preserving the sharpness of corrosion defect edge features.

[0091] Among these, the sharpness of corrosion defect edge features, which can be described as the steepness of the gradient change at the boundary between the defective and healthy regions, reflects the clarity of the defect boundary and is a key visual indicator for judging reconstruction quality. Artifacts generated during tomographic reconstruction due to noise interference can be non-realistic structural anomalies caused by sensor noise, model simplification, or iteration non-convergence. These can characterize factors that reduce image credibility and need to be suppressed through post-processing. Spatiotemporal joint filtering is a post-processing method that simultaneously utilizes temporal continuity and spatial local structural features for image enhancement. It can be used to suppress reconstruction artifacts while preserving realistic defect edges.

[0092] Performing spatiotemporal joint filtering on the spatial corrosion-aging distribution map can be achieved by combining the current frame with historical reconstruction results and applying structure-preserving smoothing in the spatiotemporal domain. Further, this operation can be implemented using bilateral filtering to preserve edges in the spatial domain and Kalman filtering to smooth trajectories in the temporal domain; alternatively, a 3D convolutional neural network can be constructed to perform end-to-end artifact removal on consecutive reconstructed frames, thereby improving the image signal-to-noise ratio while preserving key edges. To effectively suppress artifacts caused by noise interference during tomographic reconstruction while preserving the sharpness of corrosion defect edge features, a low smoothing intensity can be applied to high-gradient regions and a high smoothing intensity to flat regions during the filtering process. This operation achieves a balance between structure fidelity and noise suppression.

[0093] For example, in the scenario of health assessment of cables in the main power distribution system of a polar research vessel, the life prediction and health management method for marine cables based on multi-source data fusion in this embodiment can be as follows: Under low temperature and high humidity environment, the system inputs the compensated set of state characteristic parameters into an electrochemical coupling iterative tomographic reconstruction model, which embeds an upper limit constraint on ion migration rate based on finite element simulation; by minimizing the weighted mean square error, the model converges after 12 iterations, generating a cross-sectional corrosion-aging distribution map, showing that there is a local high corrosion index area near the joint; then, spatiotemporal joint filtering is performed, utilizing the temporal continuity of the past 5 reconstruction results to effectively eliminate stripe artifacts caused by fiber micro-bending, while preserving the sharpness of defect boundaries; finally, the high-fidelity distribution map is used to accurately delineate the maintenance scope.

[0094] In one embodiment, step S4 specifically includes:

[0095] S41. Based on the statistical distribution characteristics of the healthy substrate area of ​​marine cables in the spatial corrosion-aging distribution map, determine the dynamic segmentation threshold. The dynamic segmentation threshold is adjusted spatiotemporally according to the local area standard deviation and the cumulative operating time of the ship.

[0096] The healthy substrate region of marine cables can be an area of ​​insulation material that has not undergone significant degradation in the spatial corrosion-aging distribution map, which can be used as a statistical reference benchmark to dynamically determine the segmentation threshold. Furthermore, the healthy substrate region of marine cables can include, but is not limited to, one or more of the following: low corrosion index regions, stable dielectric response regions, and gradient-free abnormal background regions. The statistical distribution characteristics in the spatial corrosion-aging distribution map can be statistical measures such as the mean, variance, and skewness of the pixel values ​​of the healthy substrate region in the distribution map, which can be used to reflect the background noise and normal fluctuation levels under current service conditions to support adaptive threshold adjustment.

[0097] A dynamic segmentation threshold can be a critical image segmentation value dynamically calculated based on local statistical characteristics and runtime. It can be used to achieve sensitive detection of early, subtle defects while suppressing missegmentation caused by environmental or aging stage changes. In an exemplary embodiment, the dynamic segmentation threshold's acquisition method and operating principle can be explained in context: it is based on the mean of the healthy matrix region plus a certain multiple of the local standard deviation, with attenuation or growth factors introduced as the ship's cumulative runtime increases. The local standard deviation can be the standard deviation of the distribution map values ​​within the neighborhood centered on the candidate pixel, which can be used to quantify local degradation inhomogeneity or noise intensity, serving as a sensitivity factor for threshold adjustment. The ship's cumulative runtime can be the total sailing time or equivalent service time since the ship was put into service, which can be used to characterize the degree of cumulative material damage, adjusting the threshold's tolerance to the aging background.

[0098] Based on the statistical distribution characteristics of the healthy substrate region of marine cables in the spatial corrosion-aging distribution map, a dynamic segmentation threshold is determined. This can be achieved by calculating the statistics of pixel values ​​in the healthy region and setting an initial threshold. Furthermore, this operation can be implemented by automatically determining the initial threshold on the healthy region submap using the Otsu method, or by fitting the healthy region distribution using kernel density estimation and taking the 95th quantile as the threshold, thereby establishing a segmentation benchmark that matches the current service status. The dynamic segmentation threshold is then spatiotemporally adaptively adjusted based on the local region standard deviation and the ship's cumulative operating time. This can be achieved by representing the threshold as the product of the healthy region mean plus the local standard deviation and a time modulation function. In a specific embodiment, this operation can be achieved by having the time modulation function simulate the slow rise in background caused by aging, or by using a piecewise linear function to adjust the sensitivity at key service nodes, thereby allowing the threshold to dynamically evolve with the local noise level and the degree of material aging.

[0099] S42. Based on the dynamic segmentation threshold, a multi-dimensional connected region labeling algorithm is performed on the spatial corrosion-aging distribution map to obtain the initial set of candidate corrosion defect regions and the initial set of candidate aging defect regions.

[0100] Multidimensional connected component labeling algorithms are image segmentation methods that identify and label spatially connected pixel clusters in a high-dimensional feature space. They can be used to cluster continuous abnormal regions in a distribution map as candidate defects, avoiding misclassification of isolated noise points. For example, multidimensional connected component labeling algorithms can employ 8-neighborhood connected component labeling, morphological reconstruction-based region growing algorithms, or spectral clustering-guided connected component analysis. The initial set of corrosion defect candidate regions can be a set of potential corrosion regions initially identified after dynamic thresholding and connected component labeling, and can be used as input for corrosion authenticity determination. The initial set of aging defect candidate regions can be a set of potential aging regions initially identified after dynamic thresholding and connected component labeling, and can be used as input for aging authenticity determination.

[0101] Based on a dynamic segmentation threshold, a multidimensional connected region labeling algorithm is applied to the spatial corrosion-aging distribution map. This can involve performing connectivity analysis on pixels exceeding the threshold and merging adjacent pixels to form region labels. Furthermore, this operation can be achieved efficiently using a disjoint-set data structure to implement 8-neighborhood connected region labeling, or by combining morphological opening and closing operations for preprocessing to eliminate salt-and-pepper noise before labeling. This generates a structurally complete set of candidate defects, avoiding fragmentation. Obtaining the initial set of corrosion defect candidate regions and the initial set of aging defect candidate regions can be achieved by extracting candidate sets separately based on different physical quantities in the distribution map, thus providing separate inputs for subsequent differential feature extraction.

[0102] S43. For each candidate region in the initial set of candidate corrosion defect regions, extract its impedance spectrum phase entropy characteristic parameters and time-domain relaxation characteristic parameters; for each candidate region in the initial set of candidate aging defect regions, extract its dielectric loss growth characteristic parameters and humidity hysteresis loop characteristic parameters; and match and verify the extracted characteristic parameters with the pre-constructed defect dynamics characteristic database. When the impedance spectrum phase entropy characteristic parameter is within the preset corrosion characteristic range and the time-domain relaxation characteristic parameter is below the preset threshold, the candidate region is confirmed as a real corrosion defect region; when the dielectric loss growth characteristic parameter shows a continuous growth trend and the humidity hysteresis loop characteristic parameter exceeds the preset creep critical value, the candidate region is confirmed as a real aging defect region.

[0103] The phase entropy characteristic parameter of impedance spectroscopy can be the information entropy calculated from the phase angle sequence of electrochemical impedance spectroscopy. It can be used to characterize the degree of disorder in the interfacial electrochemical process, distinguishing between real corrosion and random interference. The time-domain relaxation characteristic parameter can be the time constant required for the current or impedance to recover to a steady state after applying a step voltage. It can be used to reflect the ion migration rate; the relaxation of the corrosion region is accelerated due to channel formation. The dielectric loss growth characteristic parameter can be the increment or trend slope of the dielectric loss tangent per unit time. It can be used to indicate the intensification of polarization loss and is a typical indicator of hygroscopic aging. The humidity hysteresis loop characteristic parameter can be the area or asymmetry of the hysteresis loop formed in the dielectric response curve during humidity cyclic loading-unloading. It is used to characterize the irreversibility of water molecule adsorption / desorption; the hysteresis is enhanced due to the destruction of the microporous structure in the aged material.

[0104] The defect dynamics feature database can be a structured knowledge base storing multi-dimensional dynamic feature parameters corresponding to various real corrosion and aging defects. It can provide a benchmark for matching and verification, supporting mechanism-driven authenticity determination. Furthermore, the defect dynamics feature database can be obtained by accumulating samples through accelerated aging experiments and long-term shipboard monitoring, and by labeling the range and evolution of its feature parameters. The preset corrosion feature interval can be the typical range of impedance spectrum phase entropy under real corrosion scenarios, used as the first criterion for corrosion authenticity determination. The preset threshold can be the upper limit of the time-domain relaxation feature parameter used to determine the authenticity of corrosion, and can be used in conjunction with phase entropy to form a composite condition for corrosion confirmation. The continuous growth trend can be the monotonically increasing pattern of the dielectric loss growth feature parameter over multiple consecutive sampling periods, used to exclude instantaneous fluctuations and confirm the irreversibility of the aging process. The preset creep critical value can be the critical level of the humidity hysteresis loop feature parameter characterizing irreversible damage to the material structure, and can be used as the second criterion for aging authenticity determination.

[0105] For each candidate region in the initial set of candidate corrosion defect regions, its impedance spectrum phase entropy characteristic parameters and time-domain relaxation characteristic parameters are extracted. This can be achieved by retracing the original electrochemical impedance data and calculating the phase sequence entropy and time-domain response time constant of the corresponding spatial region. Further, this operation can be achieved by calculating the Shannon entropy of the phase angle sequence or by extracting the dominant time constant by fitting the current relaxation curve, thereby obtaining corrosion discrimination criteria with electrochemical mechanism significance. For each candidate region in the initial set of candidate aging defect regions, its dielectric loss growth characteristic parameters and humidity hysteresis loop characteristic parameters are extracted. This can be achieved by using historical data of distributed optical fiber humidity and dielectric response and calculating the loss growth slope and humidity cycle hysteresis characteristics. In an exemplary embodiment, this operation can be achieved by performing linear regression on the tanδ time series to obtain the growth slope, or by integrating the dielectric response difference during the humidity rise and fall cycle to calculate the hysteresis loop area, thereby obtaining aging discrimination criteria reflecting the degradation of the material's microstructure.

[0106] The extracted feature parameters are matched and verified against a pre-built defect dynamics feature database. This can be done by querying the feature parameter range for the corresponding defect type in the database and determining whether the current value falls within the valid range. Furthermore, this operation can be achieved by calculating the Mahalanobis distance between the current feature vector and the database samples, or by using a support vector machine classifier for binary discrimination, thus enabling mechanism-driven true / false discrimination.

[0107] When the impedance spectrum phase entropy characteristic parameter is within a preset corrosion characteristic range and the time-domain relaxation characteristic parameter is below a preset threshold, the candidate region is confirmed as a real corrosion defect region. This can be achieved by performing a logical AND judgment: if the phase entropy falls within the effective range and the relaxation time is below the upper limit, it is determined to be real corrosion, thus reducing the false alarm rate of corrosion through dual physical criteria. When the dielectric loss growth characteristic parameter shows a continuous increasing trend and the humidity hysteresis loop characteristic parameter exceeds a preset creep critical value, the candidate region is confirmed as a real aging defect region. This can be achieved by verifying that the dielectric loss monotonically increases over multiple consecutive cycles and the hysteresis parameter exceeds the critical level, thus confirming real aging through a combination of dynamic trends and structural damage thresholds.

[0108] For example, in the scenario of mid-term health assessment of main power distribution cables on ocean-going merchant ships, the marine cable life prediction and health management method based on multi-source data fusion in this embodiment can be as follows: Based on the statistical characteristics of the healthy substrate area, combined with the current local standard deviation of 0.12 and the cumulative operating hours of 32,000 hours, the system dynamically calculates the segmentation threshold to be 1.85; through multi-dimensional connectivity marking, 7 initial corrosion candidate areas and 4 aging candidate areas are identified; for the corrosion candidate area numbered C3, the impedance spectrum phase entropy is extracted to be 2.31 (falling into the corrosion interval of [2.0, 2.8]), and the time domain relaxation time is 8.7ms (below the 10ms threshold), confirming it as real corrosion; for the aging candidate area numbered A2, the dielectric loss has a continuous growth slope of 0.012 / week for 5 weeks, and the humidity hysteresis loop area reaches 0.45 (exceeding the 0.4 creep critical value), confirming it as real aging; the remaining candidate areas are eliminated because they do not meet the composite criteria.

[0109] In one embodiment, the process of constructing the defect dynamics feature database includes:

[0110] B1. Collect standard-sized marine cable samples, prepare defect samples with known corrosion depth and insulation aging rate, and accurately measure the impedance spectrum phase entropy characteristic parameters, time-domain relaxation characteristic parameters, dielectric loss growth characteristic parameters, and humidity hysteresis loop characteristic parameters corresponding to each defect sample in a simulated marine environment.

[0111] Among these, standard-sized marine cable samples can be cable specimens conforming to industry standards or uniform geometric specifications, ensuring the comparability and generalizability of experimental results and eliminating the interference of structural differences on feature extraction. Defect samples with known corrosion depth and insulation aging rate can be cable samples prepared through artificial accelerated aging or spot corrosion processes, with their degradation degree confirmed by precise measurement. These can provide training and validation data with truth labels, supporting feature-degradation relationship modeling. Furthermore, defect samples with known corrosion depth and insulation aging rate can have their corrosion depth controlled by electrochemical etching or aged in a temperature, humidity, and salt spray chamber for a preset time, followed by confirmation of degradation parameters through microscopic sectioning or dielectric spectroscopy inversion. Simulated marine environments can be environmental chambers that reproduce typical marine service conditions such as high salt spray, high humidity, and high temperature in a laboratory, allowing for the acquisition of sensor response data consistent with the actual ship environment under controlled conditions.

[0112] Collecting standard-sized marine cable samples can be achieved by selecting or preparing cable samples according to uniform specifications. Furthermore, collecting standard-sized marine cable samples can be accomplished by cutting standard lengths from the same batch of mass-produced cables, or by customizing dedicated test cables according to classification society specifications, thus ensuring the consistency and transferability of experimental data. Preparing defect samples with known corrosion depth and insulation aging rate can be achieved by introducing a specific degree of corrosion or aging through controlled experimental methods. Furthermore, preparing defect samples with known corrosion depth and insulation aging rate can be achieved by using a constant potential polarization method to generate pitting corrosion of a predetermined depth at a specified location, or by aging in an 85°C / 85%RH salt spray environment for different periods to obtain gradient-aged samples, thereby obtaining precisely labeled degradation samples.

[0113] Accurately measuring the impedance spectrum phase entropy characteristic parameters, time-domain relaxation characteristic parameters, dielectric loss growth characteristic parameters, and humidity hysteresis loop characteristic parameters corresponding to each defect sample in a simulated marine environment can be achieved by simultaneously acquiring multimodal sensor data and calculating four types of dynamic characteristics within the environmental chamber. Furthermore, the accurate measurement of these parameters in a simulated marine environment can be achieved by using a frequency response analyzer (FRA) to obtain a broadband impedance spectrum and then calculating the phase entropy, or by using step voltage excitation combined with high-speed sampling to obtain relaxation curves. This allows for the establishment of a reliable mapping between the degradation true value and the observable characteristics.

[0114] B2. Based on corrosion depth parameters and aging rate parameters, establish a hierarchical mapping table of defect dynamic characteristic parameters, where micro corrosion defects correspond to the first phase entropy characteristic range, deep corrosion defects correspond to the second phase entropy characteristic range, initial insulation aging corresponds to the first dielectric loss growth range, and accelerated insulation aging corresponds to the second dielectric loss growth range.

[0115] The hierarchical mapping table can be a structured table that discretizes the continuous degradation degree into engineering levels and associates them with corresponding characteristic parameter ranges. It can provide clear and operable discrimination intervals for online identification, supporting defect classification and severity grading. The first phase entropy feature range can be the impedance spectrum phase entropy value range corresponding to minor corrosion defects, which can be used as a quantitative basis for minor corrosion discrimination. The second phase entropy feature range can be the impedance spectrum phase entropy value range corresponding to deep corrosion defects, which can be used as a quantitative basis for severe corrosion discrimination. The first dielectric loss growth range can be the typical change range of dielectric loss growth characteristic parameters in the initial insulation aging stage, which can be used to identify early reversible or slow aging. The second dielectric loss growth range can be the significant increase range of dielectric loss growth characteristic parameters in the accelerated insulation aging stage, which can be used to warn of irreversible aging processes.

[0116] Based on corrosion depth and aging rate parameters, a hierarchical mapping table of defect dynamic characteristic parameters can be established. This can be achieved by dividing continuous degradation parameters into discrete levels and statistically analyzing the distribution range of characteristic parameters corresponding to each level. Furthermore, this hierarchical mapping table can be established by automatically grouping feature-degradation data using K-means clustering or by setting theoretical boundary points based on a failure physics model, thereby enabling the transformation from continuous physical quantities to engineering criteria.

[0117] Associating minute corrosion defects with the first phase entropy feature range can be achieved by recording the phase entropy statistical interval of micro-corrosion samples in a mapping table, thus providing a criterion for identifying weak corrosion. Associating deep corrosion defects with the second phase entropy feature range can also be achieved by recording the phase entropy statistical interval of deep corrosion samples in a mapping table, thus providing a criterion for identifying severe corrosion. Associating initial insulation aging with the first dielectric loss growth range can be achieved by recording the loss growth slope interval of initial aging samples in a mapping table, thus supporting early aging warnings. Associating accelerated insulation aging with the second dielectric loss growth range can be achieved by recording the loss growth slope interval of accelerated aging samples in a mapping table, thus supporting critical aging alarms.

[0118] B3. Establish a correlation between the defect dynamic characteristic parameters in the hierarchical mapping table and the critical influence threshold of the remaining service life of marine cables to form a defect dynamic characteristic database for defect authenticity and severity assessment.

[0119] Specifically, the critical impact threshold for the remaining service life of marine cables can be defined as the degradation critical point where the cable's remaining service life falls below the safety margin when defects reach a certain level. This can be used to directly link characteristic parameters with service life risk, supporting decision-level output. Defect authenticity detection can be the process of distinguishing between genuine material degradation and false signals such as measurement noise and reconstruction artifacts. This can be used to reduce false alarm rates and improve system reliability. Severity assessment can be the process of determining the level of impact of defect characteristic parameters on the safe operation of the cable. This can be used to support tiered early warning and differentiated operation and maintenance strategies.

[0120] Establishing a correlation between the defect dynamics characteristic parameters in the grading mapping table and the critical impact threshold of the remaining service life of marine cables can be achieved through life tests or failure models to determine the remaining service life level corresponding to each characteristic interval. Furthermore, this correlation can be established by extrapolating the remaining service life at different aging rates based on the Arrhenius model, or by inversely deriving the critical characteristic threshold through accelerated breakdown experiments. This allows the feature identification results to be directly converted into life risk assessment. A defect dynamics characteristic database for defect authenticity discrimination and severity assessment can be formed by integrating the grading mapping table and life thresholds into a structured knowledge base, thereby providing a mechanism-driven discrimination engine for online health management systems.

[0121] Taking the database construction and deployment of a marine cable health management system as an example, the method for predicting the lifespan and managing the health of marine cables based on multi-source data fusion in this embodiment can be as follows: the research team collects 10kV marine power cable samples conforming to the IEC60092 standard, prepares samples with corrosion depths of 0.1mm (micro) and 0.3mm (deep) in a simulated marine environment chamber, and aging samples with aging rates of 0.005 / month (initial) and 0.02 / month (accelerated); and conducts aging tests at 85% RH, 35°C, and 5% NaC. Four types of kinetic characteristics were measured under salt spray conditions, and a hierarchical mapping table was established: micro-corrosion corresponds to phase entropy [1.8, 2.2], and deep corrosion corresponds to [2.5, 3.0]; the initial aging dielectric loss growth slope is <0.01 / week, and accelerated aging is >0.015 / week; further, the critical threshold of remaining lifetime before breakdown <6 months was correlated; after embedding the database into step S4, a candidate area (phase entropy 2.7, relaxation time 7ms) in a cargo ship cable was successfully identified as deep corrosion with a remaining lifetime of only 4 months, triggering a first-level warning.

[0122] In one embodiment, the following steps are also included after step S5:

[0123] S6. Based on the spatial location and distribution density of the actual corrosion defect area and the actual aging defect area, generate a heat map of the equipment health status, and evaluate the remaining life coefficient of the marine cable based on the heat map of the equipment health status and the Arrhenius model.

[0124] The equipment health status heatmap can be a two-dimensional or three-dimensional image that visualizes the overall health status of the cable in the form of a color gradient, based on the spatial location and distribution density of actual corrosion and aging defects. It can be used to intuitively reflect the degradation hotspots and their severity along the cable, assisting maintenance personnel in quickly identifying high-risk sections. In this embodiment, the equipment health status heatmap can map the spatial coordinates and distribution density parameters of actual corrosion and aging defect areas to the cable geometric model, generating a continuous thermal distribution through interpolation and normalization. The Arrhenius model can be a physicochemical model describing the exponential relationship between the material aging rate and absolute temperature, commonly used for accelerated life testing extrapolation. It can be used to combine the local degradation intensity reflected in the heatmap with historical temperature, humidity, and load data to achieve a mechanism-driven assessment of the remaining life coefficient. In an exemplary embodiment, the Arrhenius model can establish a mapping between the aging rate and the local equivalent temperature based on the exponential relationship between the reaction rate constant and absolute temperature, where the reaction rate constant changes negatively exponentially with the ratio of activation energy to gas constant and temperature. The remaining life factor can be a dimensionless quantitative indicator that represents the proportion of the current health status of a cable to its remaining design life. It can be used as a core criterion for life assessment and operation and maintenance decisions, reflecting the time margin by which the cable can operate safely.

[0125] The preset safe operation threshold can be a lower limit of the remaining life coefficient set according to the reliability requirements of the ship's power system, which can be used as a critical criterion to trigger the automatic generation of health analysis reports. Based on the spatial location and distribution density of actual corrosion defect areas and actual aging defect areas, a heat map of the equipment's health status is generated. This can be achieved by inputting the spatial coordinates and density parameters of the identified actual defect areas into the cable's geometric model, and generating the heat map through spatial interpolation and color mapping. Furthermore, this operation can be achieved by constructing a continuous health index field using radial basis function interpolation, or by generating a heat distribution with confidence intervals based on the Kriging spatial statistical method, thereby enabling spatial visualization of the cable's health status and highlighting localized degradation hotspots.

[0126] Assessing the remaining life coefficient of marine cables based on equipment health status heatmaps combined with the Arrhenius model can be achieved by converting local health indices in the heatmap into equivalent aging rates, and then integrating historical temperature-load data into the Arrhenius model to calculate the remaining life ratio. Furthermore, this operation can be achieved by integrating the heatmap into different regions and weighting the sum by activation energy of each region to obtain the overall cable life coefficient, or by using Monte Carlo simulation to consider temperature fluctuation uncertainties and outputting a probability distribution of the life coefficient. This allows for a quantitative assessment of lifespan that integrates spatial degradation distribution and physical aging mechanisms.

[0127] S7. When the remaining life coefficient is lower than the preset safe operating threshold, an equipment health analysis report is automatically generated, which includes defect spatial distribution characteristic parameters, severity level, and corresponding ship operating condition optimization suggestions.

[0128] The defect spatial distribution characteristic parameters can be a set of quantitative indicators describing the geometric and statistical characteristics of actual corrosion and aging defects in the cable space, which can be used to provide structured defect information for health analysis reports. The severity level can be a classification of the degree of defect hazard based on a comprehensive assessment of corrosion depth, aging rate, and distribution density, which can be used to guide the prioritization of maintenance and the selection of handling strategies. Ship operating condition optimization recommendations can be adjustments to the ship's power system operation proposed for specific defect states, which can be used to slow down the degradation process and extend the safe service life of cables. In an exemplary embodiment, ship operating condition optimization recommendations may include one or more of the following: load current limiting recommendations, enhanced cabin ventilation recommendations, and environmental isolation recommendations for wiring paths. The equipment health analysis report can be a structured electronic document containing defect spatial distribution characteristic parameters, severity levels, and operating condition optimization recommendations, which can be used to provide decision support for ship-shore collaborative maintenance and achieve a closed loop of health management.

[0129] When the remaining lifespan factor falls below a preset safe operating threshold, an equipment health analysis report is automatically generated, containing defect spatial distribution characteristic parameters, severity levels, and corresponding ship operating condition optimization suggestions. This can be achieved by triggering a conditional judgment, calling a template engine to populate defect data, severity determination results, and preset optimization strategy library content to generate a standardized report. Furthermore, this operation can be implemented by matching defect types with the optimization strategy knowledge base using a rule engine, or by using a large language model to generate natural language optimization suggestions based on defect context, thereby achieving automated closed-loop output from condition assessment to maintenance recommendations.

[0130] Taking the health management of the main propulsion cable of a research vessel as an example, the method for predicting and managing the lifespan of marine cables based on multi-source data fusion in this embodiment can be as follows: After the system confirms in step S5 that there is a high-density aging defect in the 3.1-3.4 meter section of the starboard propeller feeder, a heat map of the equipment health status is generated in step S6, showing that the health index of this section is only 0.32; combined with the average temperature of 42℃ and load fluctuation data of this area over the past 30 days, the remaining life coefficient is calculated to be 0.41 using the Arrhenius model, which is lower than the preset safe operation threshold of 0.5; the system automatically executes step S7 to generate an equipment health analysis report, indicating that the defect is at the level of a level two warning, and recommends: ① limiting the propulsion power to no more than 85% of the rated value during navigation; ② strengthening dehumidification and ventilation in the cable tray area; ③ prioritizing infrared thermal imaging re-inspection after berthing. The report is simultaneously pushed to the ship's electromechanical officer and the shore-based maintenance center.

[0131] In one embodiment, step S1 specifically includes:

[0132] S11. Obtain the insulation material type parameters and design environmental resistance level parameters of the marine cable to be monitored. Based on the insulation material type parameters and design environmental resistance level parameters, query the pre-built material-sensor configuration mapping table to determine the excitation frequency range of the electrochemical impedance spectroscopy sensor and the spatial sampling frequency of the fiber optic humidity sensor suitable for this type of marine cable.

[0133] Among these parameters, the insulation material type parameter can be an identifying parameter describing the type of polymer material used in the insulation layer of marine cables. This can serve as a basis for sensor configuration adaptation, ensuring that the excitation and sampling strategies match the electrochemical and dielectric properties of the materials. The design environmental resistance level parameter can characterize the cable's ability to withstand the harshness of the marine environment as specified in the design phase. This can reflect the upper limits of salt spray, humidity, and temperature in the cable's expected service environment, guiding the configuration of sensor sensitivity. The material-sensor configuration mapping table can be a pre-established structured data table that maps insulation material types and environmental resistance levels to optimal sensor operating parameters. This can be used to achieve adaptive selection of sensor excitation frequency and spatial sampling frequency, improving sensing specificity. Furthermore, the material-sensor configuration mapping table can be statistically analyzed based on laboratory accelerated aging tests and simulation analysis, statistically determining the optimal sensor parameter combinations under different material-environment combinations and solidifying this into lookup table rules.

[0134] The excitation frequency range of an electrochemical impedance spectroscopy (EIS) sensor can be a frequency range of AC excitation signals determined for specific insulation materials and environmental levels. This ensures that impedance measurements cover the corrosion-sensitive frequency bands of materials, improving the efficiency of capturing early degradation characteristics. The spatial sampling frequency of a fiber optic humidity sensor can be the density of spatial sampling points per unit length of a distributed fiber optic humidity sensor. This determines the spatial resolution of changes in dielectric property gradients; higher sampling frequencies can capture localized micro-area moisture absorption anomalies. The cable under monitoring can be a marine power cable currently undergoing health monitoring, serving as the object of multimodal sensing and lifetime assessment.

[0135] Obtaining the insulation material type parameters and design environmental resistance rating parameters of the marine cable to be monitored can be achieved by reading material and environmental rating information from cable technical files, equipment nameplates, or BIM models. Furthermore, this operation can be achieved by automatically identifying the cable model and parsing the corresponding material and rating parameters using RFID tags, or by manually entering the parameters and then having the system verify their validity, thus providing input for adaptive configuration of sensor parameters. A pre-built material-sensor configuration mapping table is then queried based on the insulation material type parameters and design environmental resistance rating parameters, using material type and environmental rating as joint keys to retrieve matching items in the mapping table. In a specific embodiment, this operation can be accelerated by using a hash index or by using fuzzy matching to handle parameter boundary cases, thereby achieving automated mapping from cable properties to sensor parameters. Determining the excitation frequency range of the electrochemical impedance spectroscopy sensor and the spatial sampling frequency of the fiber optic humidity sensor suitable for this type of marine cable can be achieved by extracting the corresponding frequency and sampling rate configuration values ​​from the mapping table and sending them to the sensor controller. Furthermore, this operation can be achieved by dynamically interpolating intermediate parameters between preset levels, or by combining an online learning mechanism to fine-tune the initial configuration, thereby enabling personalized adaptation of the sensing strategy.

[0136] S12. Under the excitation frequency range and spatial sampling frequency, control the multi-modal sensor array to scan the cable to be monitored synchronously, and capture the electrochemical transient response characteristics caused by corrosion and the dynamic change sequence of insulation humidity caused by environmental fluctuations in real time through the high-speed data acquisition system.

[0137] Electrochemical transient response characteristics can be the dynamic change mode of unsteady electrochemical impedance induced by the corrosion process under optimized excitation frequency, which can be used to reflect early corrosion kinetics, such as interfacial charge accumulation and ion migration rate. Insulation humidity dynamic change sequence can be a continuously recorded spatiotemporal evolution data stream of humidity along the cable at a set spatial sampling frequency, which can be used to reflect the water molecule penetration path and rate, supporting dielectric aging modeling. A high-speed data acquisition system can be a data acquisition hardware platform with high sampling rate and multi-channel synchronization capabilities, which can be used to ensure strict temporal alignment between the electrochemical transient response and humidity dynamic changes, ensuring the fidelity of multi-source signals. Furthermore, the high-speed data acquisition system can employ a multi-channel ADC synchronous sampling architecture, combined with an FPGA, to achieve microsecond-level timestamp alignment and real-time caching.

[0138] Within the excitation frequency range and spatial sampling frequency, controlling the multimodal sensor array to synchronously scan the cable under monitoring can be achieved by driving the electrochemical impedance spectroscopy sensor to apply excitation according to determined frequency parameters, while simultaneously triggering the fiber optic system to collect humidity data at a specified spatial density. Furthermore, this operation can be achieved by extracting the impedance response within a specified frequency band using lock-in amplification technology, or by utilizing optical frequency domain reflectometry (OFDR) technology to achieve high-density spatial sampling, thereby enabling material-adaptive collaborative sensing. Real-time capture of the electrochemical transient response characteristics caused by corrosion and the dynamic change sequence of insulation humidity caused by environmental fluctuations through a high-speed data acquisition system can be achieved by synchronously recording the impedance response voltage / current and fiber backscattered signals through multiple channels. In a specific embodiment, this operation can be achieved by using a 16-bit or higher ADC to capture transient signals at a sampling rate of 1 MS / s, or by employing a zero-copy DMA transfer mechanism to reduce the risk of data loss, thereby obtaining high-fidelity, time-aligned multiphysics raw data.

[0139] Taking the deployment of high-voltage shore power cables on a new type of LNG carrier as an example, the method for predicting the lifespan and health management of marine cables based on multi-source data fusion in this embodiment can be as follows: The ship selects a 6.6kV shore power cable with silicone rubber insulation and an exposed deck environmental resistance design. The system automatically reads its material type parameter as "SiR" and environmental resistance level as "exposed deck level". After querying the material-sensor configuration mapping table, it determines that the excitation frequency range of the electrochemical impedance spectroscopy is 0.1-5kHz (covering the SiR interface polarization frequency band), and the spatial sampling frequency of the fiber optic humidity sensor is set to 2cm. The high-speed data acquisition system synchronously records the impedance transient response and humidity dynamic sequence at 500kS / s, successfully capturing the periodic micro-condensation at the joint caused by the diurnal temperature difference and the accompanying interface charge accumulation phenomenon, providing a high-quality data foundation for subsequent accurate modeling.

[0140] In one embodiment, the sampling rate of the electrochemical impedance spectroscopy sensor is configured to match the typical response timescale of the electrochemical reaction of metal corrosion in marine cable insulation to ensure that the acquired time-domain signal has sufficient corrosion source localization resolution; the humidity sensitivity of the fiber optic humidity sensor is configured to match the moisture diffusion rate in the early stages of insulation material aging to ensure effective detection of subtle abnormal changes in the humidity field.

[0141] The sampling rate of the electrochemical impedance spectroscopy (EIS) sensor can be defined as the number of times the electrochemical impedance response signal is sampled per unit time, used to characterize temporal resolution. In this embodiment, the sampling rate of the EIS sensor can be used to ensure sufficient detail in capturing the transient process of the electrochemical corrosion reaction of metals, supporting high-precision corrosion source localization. The typical response timescale of the electrochemical corrosion reaction of marine cable insulation can be the range of characteristic time constants exhibited by the interface charge transfer or diffusion process when the cable shielding or armor metal undergoes electrochemical corrosion in a marine environment. For example, the typical response timescale of the electrochemical corrosion reaction of marine cable insulation can be used as the physical basis for configuring the sampling rate of the EIS sensor. The corrosion source localization resolution can be the smallest resolvable spatial interval for retrieving the corrosion location based on the time-domain impedance signal. In a specific embodiment, the corrosion source localization resolution can be used to measure the precision with which the system can locate corrosion defects axially or radially.

[0142] Configuring the sampling rate of the electrochemical impedance spectroscopy (EIS) sensor to match the typical response timescale of the electrochemical reaction of metal corrosion in marine cable insulation can be achieved by determining the target timescale based on a corrosion kinetic model and setting the sampling period to be less than 1 / 5 to 1 / 10 of this timescale to meet the Nyquist sampling criterion. Furthermore, configuring the sampling rate of the EIS sensor to match the typical response timescale of the electrochemical reaction of metal corrosion in marine cable insulation can be achieved by predicting the corrosion response time at different temperatures based on the Arrhenius equation and dynamically adjusting the sampling rate, or by using an event-triggered sampling mechanism to automatically increase the local sampling density when an impedance abrupt change is detected. This avoids temporal aliasing, preserves the transient characteristics of corrosion, and improves the spatial positioning accuracy of the corrosion source. The humidity sensitivity of the fiber optic humidity sensor can be a measure of the sensor's response amplitude to optical signals (such as wavelength shift and light intensity attenuation) caused by a unit change in humidity. In this embodiment, the humidity sensitivity of the fiber optic humidity sensor can be used to determine the detection limit for the weak moisture penetration in the early stages of insulation material aging.

[0143] The moisture diffusion rate in the early stages of insulation material aging can be characterized by the migration speed of water molecules during the initial penetration phase in the insulating polymer matrix. In one specific embodiment, the moisture diffusion rate in the early stages of insulation material aging can be used as a physical benchmark for configuring the sensitivity of a humidity sensor, ensuring coverage of the early aging sensitive range. Subtle anomalous changes in the humidity field can be subtle disturbances in the local dielectric environment caused by early moisture penetration, manifesting spatially or temporally as humidity fluctuations below the conventional detection threshold. In an exemplary embodiment, subtle anomalous changes in the humidity field can be used as a key signal feature for early aging warning.

[0144] Configuring the humidity sensitivity of a fiber optic humidity sensor to match the moisture diffusion rate in the early stages of insulation material aging can be achieved by adjusting the fiber coating thickness, grating length, or demodulation algorithm gain based on the material's diffusion coefficient, enabling the system to distinguish humidity changes of a corresponding magnitude. Furthermore, matching the humidity sensitivity of the fiber optic humidity sensor to the moisture diffusion rate in the early stages of insulation material aging can be achieved by using a highly hydrophilic nanoporous coating to amplify refractive index changes caused by slight humidity, or by employing differential demodulation technology to suppress common-mode noise and improve the signal-to-noise ratio to identify minute signal shifts. This enhances the ability to detect early water molecule penetration and reduces the risk of missed detections during aging.

[0145] For example, in the scenario of monitoring the main propulsion motor cable in the low-temperature and high-humidity environment of a polar research vessel, the marine cable life prediction and health management method based on multi-source data fusion in this embodiment could be as follows: The cable uses XLPE insulation and galvanized steel tape armor. According to laboratory data, the corrosion response timescale of its armor layer at -10℃ and 95%RH is approximately 2.3 seconds. Based on this, the system sets the sampling rate of the electrochemical impedance spectroscopy sensor to 1Hz (period 0.5 seconds) to ensure that each corrosion transient process is covered by at least 4 sampling points; simultaneously, because the moisture diffusion rate of XLPE in the early stage of aging is 3×10 -11 m 2 The system is equipped with a fiber optic humidity sensor featuring a nanoporous TiO2 coating, enabling it to produce a resolvable wavelength shift of 0.8 pm to a 0.1% RH change. In actual operation, it successfully detected the early corrosion initiation point caused by condensation at a distance of 1.7 meters from the terminal and the slight humidity gradient anomaly in the adjacent area, achieving an early warning 14 days in advance.

[0146] In addition, to achieve the above objectives, refer to Figure 4 The present invention also provides a marine cable life prediction and health management system based on multi-source data fusion, the system comprising:

[0147] Sensor array module 10 is used to simultaneously acquire multi-source time-domain signals characterizing the corrosion and aging behavior of cable insulation by utilizing an electrochemical impedance spectroscopy sensor and a distributed optical fiber humidity sensor deployed on marine cables.

[0148] Signal compensation module 20 is used to dynamically compensate the multi-source time-domain signal based on ship marine environment parameters and cable load condition parameters, and generate a set of compensated state feature parameters.

[0149] The tomographic reconstruction module 30 is used to construct a spatial corrosion-aging distribution map of the cable insulation layer based on the state feature parameter set and through an electro-chemical coupled field tomographic reconstruction algorithm. The electro-chemical coupled field tomographic reconstruction algorithm is an algorithm that integrates electrical and chemical field response data and reconstructs the internal corrosion-aging spatial distribution of the cable insulation layer through inversion calculation.

[0150] The defect identification module 40 is used to extract defect areas and distinguish between true and false in the spatial corrosion-aging distribution map, and to determine the true corrosion defect area and the true aging defect area.

[0151] The risk assessment module 50 is used to calculate the corrosion depth and insulation aging rate based on the actual corrosion defect area and the actual aging defect area using a joint degradation model, and to conduct risk assessment in conjunction with a preset safety threshold. The joint degradation model is a composite mathematical model that combines the hyperbolic tangent function to describe the initial rapid degradation stage and the power law function to describe the long-term gradual degradation stage, and is used to perform unified quantitative modeling of corrosion depth and insulation aging rate.

[0152] Other embodiments or specific implementations of the marine cable life prediction and health management system based on multi-source data fusion described in this invention can be referred to the above-mentioned method embodiments, and will not be repeated here.

[0153] Furthermore, to achieve the above objectives, the present invention also provides a computer-readable storage medium storing a marine cable life prediction and health management program based on multi-source data fusion, wherein when the marine cable life prediction and health management program based on multi-source data fusion is executed by a processor, it implements the steps of the marine cable life prediction and health management method based on multi-source data fusion as described above.

[0154] The above are merely preferred embodiments of the present invention and do not limit the scope of the patent. Any equivalent structural or procedural transformations made based on the description and drawings of the present invention, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of the present invention.

Claims

1. A method for predicting the lifespan and managing the health of marine cables based on multi-source data fusion, characterized in that, The method includes: S1. Using an electrochemical impedance spectroscopy sensor and a distributed optical fiber humidity sensor deployed on marine cables, multi-source time-domain signals characterizing the corrosion and aging behavior of cable insulation are collected simultaneously. S2. Based on the ship's marine environment parameters and cable load conditions, the multi-source time-domain signals are dynamically compensated to generate a set of compensated state characteristic parameters. S3. Based on the state characteristic parameter set, a spatial corrosion-aging distribution map of the cable insulation layer is constructed using an electro-chemical coupled field tomography reconstruction algorithm. The electro-chemical coupled field tomography reconstruction algorithm is an algorithm that integrates electrical and chemical field response data and reconstructs the spatial distribution of corrosion-aging inside the cable insulation layer through inversion calculation. S4. Extract defect areas and determine authenticity of the spatial corrosion-aging distribution map to identify the real corrosion defect areas and the real aging defect areas. S5. Based on the actual corrosion defect area and the actual aging defect area, the corrosion depth and insulation aging rate are calculated using a joint degradation model, and a risk assessment is performed in conjunction with a preset safety threshold. The joint degradation model is a composite mathematical model that combines the hyperbolic tangent function to describe the initial rapid degradation stage and the power law function to describe the long-term gradual degradation stage, and is used to perform unified quantitative modeling of corrosion depth and insulation aging rate. Specifically, step S2 includes: S21. Real-time collection of salt spray deposition density and micro-environmental humidity data on the surface of marine cables through a distributed environmental sensing network deployed on the cable surface; querying a pre-built environmental parameter-electrochemical disturbance mapping table based on the salt spray deposition density and micro-environmental humidity data to obtain the electrochemical interference compensation coefficient corresponding to the salt spray concentration and environmental humidity value at each cable segment. S22. For the time-domain signals collected by each sensor channel, the electrochemical interference compensation coefficient is applied to the amplitude normalization process of the electrochemical impedance signal, and the compensation amount of the effective ion diffusion path is calculated in combination with the design wall thickness parameters of the marine cable insulation layer. S23. The normalized time-domain signal is fused with the compensation amount of the effective ion diffusion path at the data level to generate a compensated set of state characteristic parameters including electrochemical impedance attenuation coefficient, ion migration time constant and humidity diffusion characteristic parameters. Specifically, step S4 includes: S41. Based on the statistical distribution characteristics of the healthy substrate area of ​​marine cables in the spatial corrosion-aging distribution map, a dynamic segmentation threshold is determined. The dynamic segmentation threshold is adjusted spatiotemporally according to the local area standard deviation and the cumulative operating time of the ship. S42. Based on the dynamic segmentation threshold, perform a multidimensional connected region labeling algorithm on the spatial corrosion-aging distribution map to obtain an initial set of candidate corrosion defect regions and an initial set of candidate aging defect regions. S43. For each candidate region in the initial set of candidate corrosion defect regions, extract its impedance spectrum phase entropy characteristic parameters and time-domain relaxation characteristic parameters; for each candidate region in the initial set of candidate aging defect regions, extract its dielectric loss growth characteristic parameters and humidity hysteresis loop characteristic parameters; and match and verify the extracted characteristic parameters with the pre-constructed defect dynamics characteristic database. When the impedance spectrum phase entropy characteristic parameter is within the preset corrosion characteristic range and the time-domain relaxation characteristic parameter is below the preset threshold, the candidate region is confirmed as a real corrosion defect region; when the dielectric loss growth characteristic parameter shows a continuous growth trend and the humidity hysteresis loop characteristic parameter exceeds the preset creep critical value, the candidate region is confirmed as a real aging defect region.

2. The method for predicting the lifespan and managing the health of marine cables based on multi-source data fusion as described in claim 1, characterized in that, Step S3 specifically includes: S31. The compensated set of state characteristic parameters is used as an input variable and transmitted to the pre-constructed electrochemical coupling iterative tomographic reconstruction model. The electrochemical coupling iterative tomographic reconstruction model has built-in prior physical constraints based on finite element corrosion simulation analysis. S32. By constructing an objective function to minimize the weighted mean square error between the actual measured signal and the model simulation signal, the spatial distribution matrix of the corrosion field and aging field inside the insulation layer of the marine cable is iteratively inverted to solve the spatial corrosion-aging distribution map with high resolution. S33. Perform spatiotemporal joint filtering on the spatial corrosion-aging distribution map to effectively suppress artifacts caused by noise interference during the tomographic reconstruction process while preserving the sharpness of the corrosion defect edge features.

3. The method for predicting the lifespan and managing the health of marine cables based on multi-source data fusion as described in claim 1, characterized in that, The process of constructing the defect dynamics feature database includes: B1. Collect standard-sized marine cable samples, prepare defect samples with known corrosion depth and insulation aging rate, and accurately measure the impedance spectrum phase entropy characteristic parameters, time-domain relaxation characteristic parameters, dielectric loss growth characteristic parameters, and humidity hysteresis loop characteristic parameters corresponding to each defect sample in a simulated marine environment. B2. Based on corrosion depth parameters and aging rate parameters, establish a hierarchical mapping table of defect dynamic characteristic parameters, where micro corrosion defects correspond to the first phase entropy characteristic range, deep corrosion defects correspond to the second phase entropy characteristic range, initial insulation aging corresponds to the first dielectric loss growth range, and accelerated insulation aging corresponds to the second dielectric loss growth range. B3. Establish a correlation between the defect dynamic characteristic parameters in the hierarchical mapping table and the critical influence threshold of the remaining service life of the marine cable to form a defect dynamic characteristic database for defect authenticity and severity assessment.

4. The method for predicting the lifespan and managing the health of marine cables based on multi-source data fusion as described in claim 1, characterized in that, It also includes the following steps performed after step S5: S6. Based on the spatial location and distribution density of the actual corrosion defect area and the actual aging defect area, generate a heat map of the equipment health status, and evaluate the remaining life coefficient of the marine cable based on the heat map of the equipment health status and the Arrhenius model. S7. When the remaining life coefficient is lower than the preset safe operation threshold, an equipment health analysis report is automatically generated, which includes defect spatial distribution characteristic parameters, severity level and corresponding ship operation condition optimization suggestions.

5. The method for predicting the lifespan and managing the health of marine cables based on multi-source data fusion as described in claim 1, characterized in that, Step S1 specifically includes: S11. Obtain the insulation material type parameters and design environmental resistance level parameters of the marine cable to be monitored. Based on the insulation material type parameters and design environmental resistance level parameters, query the pre-built material-sensor configuration mapping table to determine the excitation frequency range of the electrochemical impedance spectroscopy sensor and the spatial sampling frequency of the fiber optic humidity sensor suitable for this type of marine cable. S12. Under the excitation frequency range and spatial sampling frequency, control the multi-modal sensor array to scan the cable to be monitored synchronously, and capture the electrochemical transient response characteristics caused by corrosion and the dynamic change sequence of insulation humidity caused by environmental fluctuations in real time through the high-speed data acquisition system.

6. The method for predicting the lifespan and managing the health of marine cables based on multi-source data fusion as described in claim 5, characterized in that, The sampling rate of the electrochemical impedance spectroscopy sensor is configured to match the typical response timescale of the electrochemical reaction of metal corrosion in marine cable insulation; the humidity sensitivity of the distributed fiber optic humidity sensor is configured to match the moisture diffusion rate in the early stages of insulation material aging.

7. A marine cable life prediction and health management system based on multi-source data fusion, characterized in that, The system includes: The sensor array module is used to simultaneously acquire multi-source time-domain signals characterizing the corrosion and aging behavior of the cable insulation layer by utilizing an electrochemical impedance spectroscopy sensor and a distributed fiber optic humidity sensor deployed on marine cables. The signal compensation module is used to dynamically compensate the multi-source time-domain signal based on the ship's marine environment parameters and cable load conditions, and generate a set of compensated state feature parameters. The tomographic reconstruction module is used to construct a spatial corrosion-aging distribution map of the cable insulation layer based on the state feature parameter set and through an electro-chemical coupled field tomographic reconstruction algorithm. The electro-chemical coupled field tomographic reconstruction algorithm is an algorithm that integrates electrical and chemical field response data and reconstructs the spatial distribution of corrosion-aging inside the cable insulation layer through inversion calculation. The defect identification module is used to extract defect areas and distinguish between genuine and fake data from the spatial corrosion-aging distribution map, and to determine the real corrosion defect areas and the real aging defect areas. The risk assessment module is used to calculate the corrosion depth and insulation aging rate based on the actual corrosion defect area and the actual aging defect area using a joint degradation model, and to conduct risk assessment in combination with a preset safety threshold. The joint degradation model is a composite mathematical model that combines the hyperbolic tangent function to describe the initial rapid degradation stage and the power law function to describe the long-term gradual degradation stage, and is used to perform unified quantitative modeling of corrosion depth and insulation aging rate. Specifically, based on ship marine environmental parameters and cable load parameters, dynamic compensation is performed on the multi-source time-domain signals to generate a compensated set of state characteristic parameters, including: The salt spray deposition density and micro-environment humidity data of the marine cable surface are collected in real time by a distributed environmental sensing network deployed on the cable surface. Based on the salt spray deposition density and micro-environment humidity data, a pre-constructed environmental parameter-electrochemical disturbance mapping table is queried to obtain the electrochemical interference compensation coefficient corresponding to the salt spray concentration and environmental humidity value at each cable segment. For the time-domain signals acquired by each sensor channel, the electrochemical interference compensation coefficient is applied to the amplitude normalization process of the electrochemical impedance signal, and the compensation amount of the effective ion diffusion path is calculated in combination with the design wall thickness parameters of the marine cable insulation layer. The normalized time-domain signal is fused with the compensation amount of the effective ion diffusion path at the data level to generate a compensated set of state characteristic parameters, including the electrochemical impedance attenuation coefficient, ion migration time constant, and humidity diffusion characteristic parameters. Specifically, the extraction of defect areas and the determination of authenticity of the spatial corrosion-aging distribution map to identify real corrosion defect areas and real aging defect areas include: Based on the statistical distribution characteristics of the healthy substrate area of ​​marine cables in the spatial corrosion-aging distribution map, a dynamic segmentation threshold is determined. The dynamic segmentation threshold is adjusted spatiotemporally based on the local area standard deviation and the cumulative operating time of the ship. Based on the dynamic segmentation threshold, a multidimensional connected region labeling algorithm is performed on the spatial corrosion-aging distribution map to obtain an initial set of candidate corrosion defect regions and an initial set of candidate aging defect regions. For each candidate region in the initial set of candidate corrosion defect regions, its impedance spectrum phase entropy characteristic parameter and time-domain relaxation characteristic parameter are extracted; for each candidate region in the initial set of candidate aging defect regions, its dielectric loss growth characteristic parameter and humidity hysteresis loop characteristic parameter are extracted; and the extracted characteristic parameters are matched and verified with a pre-constructed defect dynamics characteristic database. When the impedance spectrum phase entropy characteristic parameter is within a preset corrosion characteristic range and the time-domain relaxation characteristic parameter is below a preset threshold, the candidate region is confirmed as a real corrosion defect region; when the dielectric loss growth characteristic parameter shows a continuous growth trend and the humidity hysteresis loop characteristic parameter exceeds a preset creep critical value, the candidate region is confirmed as a real aging defect region.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a marine cable life prediction and health management program based on multi-source data fusion, which, when executed by a processor, implements the steps of the marine cable life prediction and health management method based on multi-source data fusion as described in any one of claims 1 to 6.

Images