A method and system for processing deepwater umbilical cable temperature monitoring data

Abstract

The present invention relates to the technical field of deep - water umbilical cable temperature monitoring, and proposes a deep - water umbilical cable temperature monitoring data processing method and system thereof. It can not only monitor the temperature change of the umbilical cable in real - time and accurately, improve the accuracy and reliability of the monitoring data, and ensure the safe operation of the deep - water umbilical cable in a complex marine environment. It includes a temperature monitoring host, a data processing unit, and an umbilical cable. An optical fiber sensor is provided in the umbilical cable, and the temperature monitoring host communicates with the optical fiber sensor. The temperature monitoring host includes a data acquisition unit and a demodulator. The data acquisition unit is used to convert the weak optical signal transmitted from the optical fiber sensor into an electrical signal, and the demodulator is used to accurately demodulate the temperature information from the electrical signal transmitted from the data acquisition unit. The data processing unit is used to receive the temperature information transmitted by the demodulator and perform data processing, and the temperature monitoring host is used to display the data information processed by the data processing unit.

Description

Technical Field

[0001] The present invention relates to the technical field of deep - water umbilical cable temperature monitoring. Specifically, it relates to a method and system for processing deep - water umbilical cable temperature monitoring data. Background Art

[0002] Today, with the increasing depth of ocean development, deep - water umbilical cables play a crucial role in ocean engineering. During the process of the umbilical cable being wound in and out by the winch, it is in a complex and changing ocean environment. Factors such as the flow characteristics of seawater (including changes in flow velocity and direction), depth - related temperature gradients, and dynamic stresses caused by winch operation will all affect the temperature of the umbilical cable.

[0003] Traditional temperature monitoring methods, such as point - type monitoring methods based on thermocouples or thermistors, have many limitations. These traditional sensors can only measure temperature at limited discrete points, making it difficult to deploy and measure temperature on the umbilical cable. Moreover, their durability in harsh ocean environments is poor, and they are easily interfered by factors such as seawater corrosion and mechanical vibration, resulting in inaccurate monitoring. In addition, the data transmission and processing speed of traditional monitoring methods is slow, making it difficult to achieve real - time monitoring. When the temperature of the umbilical cable changes abnormally, it cannot be detected and corresponding measures cannot be taken in time, thus posing a threat to the performance and safety of the umbilical cable and the stability of the entire ocean engineering system.

[0004] In the prior art, distributed fiber optic sensing technology, with its unique advantages, can achieve continuous and long - distance measurement along the optical fiber, and the optical fiber itself has excellent characteristics such as corrosion resistance and electromagnetic interference resistance, making it very suitable for application in ocean environments. However, traditional distributed fiber optic sensing technology has two major limitations in dealing with motion - state monitoring:

[0005] 1. The interference of acceleration mutations during the start - up and shut - down stages of the winch on the optical fiber signal is not considered, resulting in step - type errors in the temperature monitoring curve;

[0006] 2. There is a lack of a Rayleigh scattering frequency shift compensation mechanism for changes in seawater pressure. When the diving depth of the umbilical cable exceeds 500 meters, the temperature measurement error grows non - linearly.

[0007] Simply applying distributed fiber optic sensing technology is not sufficient to fully meet the special requirements of deep - water umbilical cable temperature monitoring. It is also necessary to develop a dedicated data - processing method for the characteristics of the umbilical cable to improve the accuracy and reliability of monitoring. Summary of the Invention

[0008] One of the objectives of the present invention is to propose a method for processing deep - water umbilical cable temperature monitoring data, which can not only monitor the temperature change of the umbilical cable in real - time and accurately, but also improve the accuracy and reliability of the monitoring data, ensuring the safe operation of the deep - water umbilical cable in a complex ocean environment.

[0009] The technical solution of the present invention is as follows:

[0010] A method for processing temperature monitoring data of a deep - water umbilical cable, comprising the following steps:

[0011] S100: Line test: Start the temperature monitoring host and check the connectivity of the line;

[0012] S200: Signal reception: Obtain the data signal of the fiber optic sensor through the data acquisition unit;

[0013] S300: Signal demodulation: Demodulate the temperature - related information from the electrical signal transmitted by the data acquisition unit through the demodulator, and pre - process the demodulated data to remove significantly abnormal data;

[0014] S400: Data pre - processing: Pre - process the adjusted data through the data processing unit to extract the temperature information transmitted by the fiber optic sensor;

[0015] S400: Data processing: Filter and denoise the extracted temperature information through the data processing unit;

[0016] S500: Feature extraction: Extract the signal features related to temperature changes in the processed data;

[0017] S600: Temperature estimation: Estimate the temperature through the thermo - optic effect and elasto - optic effect of the fiber optic sensor and the heat conduction model of the umbilical cable;

[0018] S700: Result verification: Verify the accuracy of the temperature by comparing the real - time temperature evaluation data and historical data;

[0019] S800: Result output: Output and display the temperature monitoring result of the umbilical cable.

[0020] Further, step S300 includes:

[0021] S310: Adopt an adaptive demodulation algorithm to automatically adjust the demodulation parameters according to the real - time characteristics of the signal to obtain the best demodulation effect.

[0022] S320: Conduct a preliminary quality assessment on the demodulated signal and remove significantly abnormal demodulated data.

[0023] Further, step S400 includes:

[0024] Adopt the cumulative average and wavelet denoising algorithm to process the demodulated data and extract the temperature information monitored by the fiber optic sensor.

[0025] Further, step S400 includes:

[0026] S410: Filter the extracted temperature signal using an adaptive filter;

[0027] S420: Denoise using a noise suppression technique based on a CNN (Convolutional Neural Network).

[0028] Furthermore, step S410 includes:

[0029] S411: Generate multi-source reference signals. The adaptive filter introduces the extracted original input temperature signal, fuses water temperature data and water flow data, and generates a composite environmental noise multi-source reference template signal.

[0030] S412: Dynamically adjust the step size. The adaptive filter introduces the umbilical cable winch speed signal, establishes a step size factor dynamic adjuster. When the winch speed v is less than or equal to 30 revolutions per minute, the step size factor μ = 0.02 + 0.03×v / 60; when the winch speed v is greater than 30 revolutions per minute, the step size factor μ = 0.05.

[0031] S413: Error feedback closed-loop. Filter the output data signal through the adaptive filter, and output the error signal through the error calculation module. The error signal is simultaneously input into the weight update module and the historical database; the weight update frequency of the adaptive filter is dynamically associated with the winch speed. When the speed sensor detects a speed > 30 rpm, start the fast convergence mode, adjust the step size factor μ of the adaptive filtering algorithm from 0.01 to 0.05, and shorten the weight update period to 5 milliseconds.

[0032] Furthermore, the convolutional neural network in step S420 adopts a three-layer progressive convolutional kernel group design, including a convolutional kernel group matching the fundamental vibration frequency of the umbilical cable:

[0033] The first layer is a 3×1 convolutional kernel, capturing the fundamental vibration characteristics of the umbilical cable, corresponding to a fundamental frequency bandwidth of 2 - 5 Hz;

[0034] The second layer is a 5×1 convolutional kernel, extracting the harmonic characteristics of material deformation, corresponding to a harmonic bandwidth of 5 - 15 Hz;

[0035] The third layer is a 7×1 convolutional kernel, capturing the characteristics of vortex-induced vibration of seawater, corresponding to a fundamental frequency bandwidth of 15 - 50 Hz;

[0036] The specific steps are as follows:

[0037] S421: Signal standardization preprocessing:

[0038] The input layer receives the time-domain vibration signal fused by multiple sensors and performs the following preprocessing operations on the original signal:

[0039] Baseline calibration: Use a moving average filter to eliminate baseline drift. The filter window length is preferably 1024 sampling points;

[0040] Dynamic normalization: Perform window normalization according to the formula where and are the mean and standard deviation within the current sliding window respectively, and is a small constant to prevent division by zero;

[0041] S422: Multi-scale vibration feature extraction:

[0042] Extract vibration features in different frequency bands successively through a three-level convolutional kernel group:

[0043] Fundamental frequency vibration layer:

[0044] Configure a 3×1 convolutional kernel, with 32 channels, a stride of 1, and the activation function is Leaky ReLU with a leakage coefficient of 0.2;

[0045] The frequency response range of the convolutional kernel covers the fundamental natural vibration frequency of the umbilical cable, 2 - 5 Hz, and is used to capture the overall vibration mode of the cable structure;

[0046] Harmonic feature layer:

[0047] Configure a 5×1 convolutional kernel, with 64 channels, a stride of 2, and the activation function is ReLU;

[0048] The frequency domain coverage range is 5 - 15 Hz, and it is specifically used to extract the second harmonic and higher harmonic features generated by material deformation;

[0049] Vortex-induced vibration layer:

[0050] Configure a 7×1 dilated convolutional kernel, with a dilation rate of 2, 128 channels, and a stride of 1;

[0051] The effective receptive field covers the high-frequency band of 15 - 50 Hz, and is specifically used to capture the broadband perturbation features generated by vortex-induced vibration caused by seawater flow;

[0052] S423: Adaptive feature enhancement:

[0053] Set a multi-modal feature fusion module at the output end of the third convolutional layer, including:

[0054] Channel attention mechanism: Dynamically calculate the weight values of each feature channel through the SE module to achieve autonomous recalibration of the feature channels;

[0055] Cross-layer residual connection: Weightedly superimpose the fundamental frequency features of the first convolutional layer and the high-frequency features of the third convolutional layer. The formula is expressed as: ;

[0056] where is the adaptive weight coefficient, and the value range is [0.3, 0.7];

[0057] Noise perception regularization: Dynamically adjust the Dropout ratio according to the real-time signal-to-noise ratio. Strong regularization is enabled when the signal-to-noise ratio is below 20 dB;

[0058] S424: Temperature regression output:

[0059] The fully connected layer adopts a physically constrained network structure:

[0060] Dimension reduction design: Sequentially set hidden layers with dimensions of 128 and 64, and the dimension of the final output layer is 1;

[0061] Nonlinear mapping: The last layer uses the Sigmoid function to constrain the output value to the interval [0, 1], and maps it to the material tolerance temperature range through linear transformation:

[0062] ;

[0063] Where 、 are respectively the minimum tolerance temperature and the maximum critical temperature of the umbilical cable material;

[0064] Physical constraint mechanism: Embed a hard boundary constraint layer to forcefully correct abnormal predicted values that exceed the material tolerance range;

[0065] Furthermore, step S600 includes:

[0066] S610: According to the umbilical cable structure model, use machine learning methods to correct and optimize the temperature model;

[0067] S620: Train and verify the hybrid model through experimental data and on-site data to improve the accuracy of temperature estimation.

[0068] Furthermore, step S700 includes:

[0069] S710: Use data mining techniques to analyze historical data, extract temperature change patterns under different working conditions, and compare the differences in temperature values, the trends of temperature changes, and the fluctuation ranges;

[0070] S720: Combine the above analysis to perform a correlation analysis between the real-time monitoring data and the current working condition. If the correlation is lower than the set threshold, it is prompted that there may be measurement errors or system failures.

[0071] Another object of the present invention is to propose a deep-water umbilical cable temperature monitoring system, including a temperature monitoring host, a data processing unit, a winch for coiling the umbilical cable, and an umbilical cable for connecting underwater equipment. An optical fiber sensor is provided in the umbilical cable, and the temperature monitoring host communicates with the optical fiber sensor;

[0072] The temperature monitoring host includes a data acquisition unit and a demodulator. The data acquisition unit is used to convert the weak optical signal transmitted by the fiber optic sensor into an electrical signal, and the demodulator is used to accurately demodulate the temperature information from the electrical signal transmitted by the data acquisition unit;

[0073] The data processing unit is used to receive the temperature information transmitted by the demodulator and perform data processing. The temperature monitoring host is used to display the data information processed by the data processing unit.

[0074] Furthermore, it includes a slip ring for connecting the temperature monitoring host and the fiber optic sensor. One side of the tail fiber of the slip ring is connected to the interface of the temperature monitoring host, and the tail fiber on the other side is connected to the fiber optic sensor through fusion splicing or a fiber optic connector.

[0075] Furthermore, it includes a fixed pulley group and underwater equipment;

[0076] The fixed pulley group is used for guiding during the bending, entering the water, and exiting the water processes of the umbilical cable;

[0077] The underwater equipment is connected to the end of the umbilical cable located underwater and is powered, communicated with, and deployed and recovered by traction through the umbilical cable to carry out underwater operations.

[0078] The working principle and beneficial effects of the present invention are as follows:

[0079] The process of the present invention mainly includes signal reception, signal demodulation, data preprocessing, data processing, feature extraction, temperature estimation, result verification, data visualization, and reporting data processing links. Special filtering and denoising techniques are adopted to improve the signal-to-noise ratio of the signal; in the feature extraction link, features such as wavelength drift and light intensity change rate are extracted, and the high-order statistical features of the signal are analyzed to more accurately determine the temperature change pattern; in the temperature estimation link, a hybrid model based on machine learning is used for temperature estimation to improve the accuracy of temperature estimation; in the result verification link, the accuracy of the temperature is verified by comparing the real-time monitoring data and historical data. This technology is of great significance in marine engineering monitoring. It can not only monitor the temperature change of the umbilical cable in real time and accurately, but also improve the accuracy and reliability of the monitoring data through innovative data processing methods, ensuring the safe operation of the deepwater umbilical cable in a complex marine environment.

[0080] In addition, the present invention conducts high-precision monitoring of the temperature of the deepwater umbilical cable through the special measurement principle of the fiber optic sensor, high-precision signal acquisition and demodulation, and advanced data processing algorithms, and can control the error within a very small range, providing accurate data support for the condition assessment of the umbilical cable.

[0081] Meanwhile, from the rapid acquisition of signals, real-time demodulation to the rapid processing of data and result output, the whole process is completed in a short time, which can timely reflect the change of the temperature of the umbilical cable. Once the temperature is abnormal during the winding and unwinding process of the umbilical cable winch, a rapid response can be made.

[0082] The long-distance monitoring ability of the distributed optical fiber sensing technology itself, combined with the design of each component in the present invention (such as the smooth ring to ensure the stability of signal transmission), enables the system to stably monitor the temperature of the long-distance deep-water umbilical cable, unaffected by the length and complex movement of the umbilical cable. Multiple links in the special algorithm, such as data preprocessing, result verification, etc., and the redundant design of each component of the system (such as the self-calibration function of the demodulator), jointly ensure the high reliability of the system and reduce the possibility of false alarms and missed alarms. The design of the system and algorithm fully considers the complexity of the marine environment, such as factors like high pressure, high corrosion, electromagnetic interference, etc., enabling the present invention to operate stably in the harsh deep-water marine environment and effectively monitor the temperature of the umbilical cable.

[0083] The present invention is of great significance in the state monitoring of deep-water umbilical cables in ocean engineering, especially focusing on the accurate monitoring of the temperature change of the umbilical cable under the influence of various factors in a complex marine environment. The present invention particularly relates to an adaptive data processing method for the marine environment under dynamic laying conditions, and develops a temperature monitoring algorithm system with environmental adaptability for special working conditions such as motion artifacts generated during the winding and unwinding process of the umbilical cable winch and signal distortion caused by the change of the seawater pressure gradient. Brief Description of the Drawings

[0084] The present invention will be further described in detail below in conjunction with the drawings and specific embodiments.

[0085] Figure 1 It is the flowchart of the adaptive filtering in the present invention;

[0086] Figure 2 It is the architecture diagram of the convolutional neural network in the present invention;

[0087] Figure 3 It is the overall architecture diagram of the deep-water umbilical cable temperature monitoring system in the present invention;

[0088] Figure 4 It is the schematic diagram of the cross-section of the umbilical cable and the structure of its optical fiber sensor in the present invention.

[0089] In the figure: 1. Temperature monitoring host; 2. Data acquisition unit; 3. Demodulator; 4. Data processing unit; 5. Smooth ring; 6. Umbilical cable; 7. Optical fiber sensor; 8. Winch; 9. Fixed pulley block; 10. Underwater equipment. Detailed Description of the Embodiments

[0090] The following will describe the technical solutions in the embodiments of the present invention in a clear and complete manner in combination with the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts fall within the scope of protection of the present invention.

[0091] As Figure 3-4 shown, a deep-water umbilical cable temperature monitoring system includes a temperature monitoring host 1, a data processing unit 4, a fixed pulley group 9, an underwater device 10, a winch 8 for coiling the umbilical cable 6, and an umbilical cable 6 for connecting the underwater device 10. An optical fiber sensor 7 is provided in the umbilical cable 6, and the temperature monitoring host 1 communicates with the optical fiber sensor 7.

[0092] Among them, the winch 8 is used to store and retract the umbilical cable 6, and the fixed pulley group 9 is used for guiding the umbilical cable 6 to bend, enter the water, and exit the water during the retraction and release process; the underwater device 10 is connected to the underwater end of the umbilical cable 6 and is powered, communicates, and is deployed and recovered by traction through the umbilical cable 6 to carry out underwater operations.

[0093] The temperature monitoring host 1 includes a data acquisition unit 2 and a demodulator. The data acquisition unit 2 is used to convert the weak optical signal transmitted by the optical fiber sensor 7 into an electrical signal, and the regulator 3 is used to accurately demodulate the temperature information from the electrical signal transmitted by the data acquisition unit 2;

[0094] The data processing unit 4 is used to receive the temperature information transmitted by the regulator 3 and perform data processing, and the temperature monitoring host 1 is used to display the data information processed by the data processing unit 4.

[0095] It includes a slip ring 5 for connecting the temperature monitoring host 1 and the optical fiber sensor 7. One side of the tail fiber of the slip ring 5 is connected to the interface of the temperature monitoring host 1, and the tail fiber on the other side is connected to the optical fiber sensor 7 through fusion splicing or an optical fiber connector.

[0096] The functions of each important hardware in this embodiment are as follows:

[0097] Optical fiber sensor 7: An optical fiber sensor 7 uses a single-core single-mode communication optical fiber integrated in the umbilical cable 6 as a sensing optical fiber, which can sensitively sense the internal temperature of the umbilical cable 6 during operation and its changes. The optical fiber sensor 7 is based on various optical effects of the optical fiber (such as Raman scattering, Brillouin scattering, etc.) to achieve temperature measurement. When the temperature of the umbilical cable 6 changes, the characteristics of the scattered light in the optical fiber (including light intensity, frequency, etc.) will change accordingly. By accurately measuring these changes, the temperature information can be obtained.

[0098] Data acquisition unit 2: It can convert the weak optical signal transmitted by the fiber optic sensor 7 into an electrical signal and has an automatic gain control function. It can automatically adjust the amplification factor according to the intensity of the input signal to ensure that the signal is processed subsequently within an appropriate range.

[0099] Smooth ring 5: It is a fiber optic rotary connection device used for fiber optic connection and conduction between rotating machinery and stationary machinery. During the retraction and deployment process of the umbilical cable 6, the smooth ring 5 can realize the fiber optic connection and communication between the fiber optic sensor 7 in the umbilical cable 6 that rotates with the winch 8 and the temperature monitoring host 1.

[0100] Demodulator: Through signal demodulation algorithms, it can demodulate the temperature-related information from the electrical signal transmitted by the data acquisition unit 2. The demodulator has a wide dynamic range, can process input signals of different intensities, and has high demodulation accuracy, reaching sub-degree Celsius level accuracy. In addition, the demodulator also has a self-calibration function and can be automatically calibrated regularly to ensure the accuracy of the demodulation results.

[0101] Data processing unit 4: It deeply analyzes the demodulated data using the cumulative average and wavelet denoising algorithms, integrating a high-performance processor and a large-capacity storage device. The processor can quickly run complex algorithms, and the storage device is used to store information such as parameters and historical data required by the algorithms. The data processing unit 4 implements special algorithms through software programming, and the software can be upgraded online to adapt to different application scenarios and improve algorithm performance.

[0102] Embodiment 2

[0103] As Figure 1-2 shown, based on the hardware of Embodiment 1, this embodiment relates to a method for processing deep umbilical cable 6 temperature monitoring data, including the following steps:

[0104] S100: Line test: Start the temperature monitoring host 1 and check the connectivity of the line;

[0105] S200: Signal reception: Obtain the data signal of the fiber optic sensor 7 through the data acquisition unit 2;

[0106] S300: Signal demodulation: Demodulate the temperature-related information from the electrical signal transmitted by the data acquisition unit 2 through the adjuster 3, and preprocess the demodulated data to remove significantly abnormal data;

[0107] Among them, step S300 includes:

[0108] S310: Adopt an adaptive demodulation algorithm to automatically adjust the demodulation parameters according to the real-time characteristics of the signal to obtain the best demodulation effect.

[0109] S320: Conduct a preliminary quality assessment on the demodulated signal and remove the significantly abnormal demodulated data.

[0110] S400: Data processing: Filter and denoise the extracted temperature information through data processing unit 4;

[0111] Process the demodulated data using the cumulative average and wavelet denoising algorithms to extract the temperature information monitored by the fiber optic sensor 7.

[0112] Among them, step S400 includes:

[0113] S410: Filter the extracted temperature signal using an adaptive filter;

[0114] Step S410 includes:

[0115] S411: Generation of multi-source reference signals. The adaptive filter introduces the extracted original input temperature signal, fuses the water temperature data and water flow data, and generates a composite environmental noise multi-source reference template signal.

[0116] S412: Dynamic step size adjustment. The adaptive filter introduces the rotational speed signal of the winch 8 of the umbilical cable 6, and establishes a dynamic step size factor adjuster. When the rotational speed v of the winch 8 is less than or equal to 30 revolutions per minute, the step size factor μ = 0.02 + 0.03×v / 60; when the rotational speed v of the winch 8 is greater than 30 revolutions per minute, the step size factor μ = 0.05.

[0117] S413: Error feedback closed loop. Filter the output data signal through the adaptive filter, and output the error signal through the error calculation module. The error signal is simultaneously input into the weight update module and the historical database; the weight update frequency of the adaptive filter is dynamically associated with the rotational speed of the winch 8. When the rotational speed sensor detects that the rotational speed > 30 rpm, start the fast convergence mode, adjust the step size factor μ of the adaptive filtering algorithm from 0.01 to 0.05, and shorten the weight update period to 5 milliseconds.

[0118] S420: Denoise using the noise suppression technology based on the CNN convolutional neural network.

[0119] The convolutional neural network in step S420 adopts a three-layer progressive convolutional kernel group design, including a convolutional kernel group matching the fundamental vibration frequency of the umbilical cable:

[0120] The first layer is a 3×1 convolutional kernel, which captures the fundamental vibration characteristics of the umbilical cable, corresponding to the fundamental frequency bandwidth of 2 - 5 Hz;

[0121] The second layer is a 5×1 convolutional kernel, which extracts the harmonic characteristics of material deformation, corresponding to the harmonic bandwidth of 5 - 15 Hz;

[0122] The third layer has a 7×1 convolution kernel that captures the characteristics of vortex-induced vibration of seawater, corresponding to a fundamental frequency bandwidth of 15 - 50 Hz;

[0123] The specific steps are as follows:

[0124] S421: Signal standardization preprocessing:

[0125] The input layer (P1) receives the time-domain vibration signal after multi-sensor fusion and performs the following preprocessing operations on the original signal:

[0126] Baseline calibration: A sliding mean filter is used to eliminate baseline drift, and the filter window length is preferably 1024 sampling points;

[0127] Dynamic normalization: According to the formula perform window standardization, where and are the mean and standard deviation within the current sliding window respectively, is a small constant to prevent division by zero;

[0128] S422: Multi-scale vibration feature extraction:

[0129] Extract vibration features in different frequency bands through a three-level convolution kernel group:

[0130] Fundamental frequency vibration layer (P2 - P3):

[0131] Configure a 3×1 convolution kernel, with 32 channels, a stride of 1, and the activation function is Leaky ReLU with a leakage coefficient of 0.2;

[0132] The frequency response range of the convolution kernel covers the fundamental frequency of the natural vibration of the umbilical cable, 2 - 5 Hz, and is used to capture the overall vibration mode of the cable structure;

[0133] Harmonic feature layer (P4 - P5):

[0134] Configure a 5×1 convolution kernel, with 64 channels, a stride of 2, and the activation function is ReLU;

[0135] The frequency domain coverage range is 5 - 15 Hz, and it is specifically used to extract the second harmonic and higher harmonic features generated by material deformation;

[0136] Vortex-induced vibration layer (P6 - P7):

[0137] Configure a 7×1 dilated convolution kernel, with a dilation rate of 2, 128 channels, and a stride of 1;

[0138] The effective receptive field covers the high-frequency band of 15 - 50 Hz, and is specifically used to capture the broadband disturbance features generated by seawater flow-induced vortex-induced vibration;

[0139] S423: Adaptive feature enhancement (P8):

[0140] A multi-modal feature fusion module is set at the output end of the third convolutional layer, including:

[0141] Channel attention mechanism: Dynamically calculate the weight values of each feature channel through the SE (Squeeze-and-Excitation) module to achieve self-calibration of the feature channels;

[0142] Cross-layer residual connection: Weightedly superimpose the fundamental frequency features of the first convolutional layer and the high-frequency features of the third convolutional layer. The formula is expressed as: ;

[0143] where is the adaptive weight coefficient, and the value range is [0.3, 0.7];

[0144] Noise-aware regularization: Dynamically adjust the Dropout ratio according to the real-time signal-to-noise ratio. When the signal-to-noise ratio is lower than 20 dB, strong regularization is enabled (Dropout rate ≥ 0.5);

[0145] S424: Temperature regression output (P9 - P10):

[0146] The fully connected layer adopts a physically constrained network structure:

[0147] Dimension reduction design: Set hidden layers of 128 dimensions and 64 dimensions in sequence, and the dimension of the final output layer is 1;

[0148] Nonlinear mapping: The last layer uses the Sigmoid function to constrain the output value to the interval [0, 1], and maps it to the material tolerance temperature range through linear transformation:

[0149] ;

[0150] where , are the minimum tolerance temperature and the maximum critical temperature of the umbilical cable material respectively;

[0151] Physical constraint mechanism: Embed a hard boundary limit layer to forcibly correct abnormal predicted values that exceed the material tolerance range;

[0152] Through the above-mentioned multi-band vibration feature collaborative extraction and physical constraint regression mechanism, the temperature inversion accuracy in complex marine environments is significantly improved.

[0153] S500: Feature extraction: Extract signal features related to temperature changes from the processed data;

[0154] S600: Estimate temperature: Estimate the temperature through the thermo-optic effect, elasto-optic effect of the fiber optic sensor 7 and the heat conduction model of the umbilical cable 6;

[0155] Among them, step S600 includes:

[0156] S610: According to the umbilical cable structure model, use machine learning methods to correct and optimize the temperature model;

[0157] S620: Train and verify the hybrid model through experimental data and on-site data to improve the accuracy of temperature estimation.

[0158] This section of steps estimates the temperature based on the extracted multi-dimensional features and complex temperature feature relationship models. This model not only considers the relationship between the basic optical properties of the optical fiber and temperature, but also incorporates the material properties of the umbilical cable 6 and the influence of marine environmental factors (such as seawater temperature gradient, pressure, etc.) on temperature measurement. A high-precision temperature estimation model is constructed through methods such as multiple regression analysis and neural networks.

[0159] Use a hybrid model for temperature estimation. First, establish a theoretical model based on physical principles, considering the thermo-optic effect, elasto-optic effect of the optical fiber, and the heat conduction model of the umbilical cable 6, etc. Then, use machine learning methods (such as support vector machines, random forests, etc.) to correct and optimize the theoretical model. The hybrid model is trained and verified through a large amount of experimental data and on-site data to improve the accuracy of temperature estimation. In addition, the model can also be adaptively adjusted according to changes in environmental parameters (such as seawater temperature, depth, etc.).

[0160] S700: Result verification: Verify the accuracy of the temperature by comparing the real-time temperature evaluation data and historical data;

[0161] This section of steps deeply explores various signal features related to the temperature change of the umbilical cable 6. In addition to traditional features such as wavelength drift and light intensity change rate, the joint features of the time domain and frequency domain of the signal are also analyzed, such as signal phase change, frequency modulation depth, etc. Through multi-dimensional feature extraction, the temperature change situation can be described more comprehensively.

[0162] In addition to extracting traditional features such as wavelength drift and light intensity change rate, the higher-order statistical features of the signal can also be analyzed, such as kurtosis, skewness, etc. These higher-order statistics can reflect the non-linear features of the signal and are of great significance for capturing complex signal changes caused by temperature changes.

[0163] At the same time, use time-frequency analysis methods (such as short-time Fourier transform, wavelet transform, etc.) to analyze the signal in the time-frequency domain and extract the joint time-frequency domain features. These features can reflect the distribution of temperature changes at different times and frequencies, and help to more accurately determine the temperature change pattern.

[0164] Among them, step S700 includes:

[0165] S710: Analyze historical data using data mining techniques, extract temperature change patterns under different working conditions, and compare the differences in temperature values, the trends of temperature changes, and the fluctuation ranges.

[0166] S720: Combine the above analysis to perform a correlation analysis between the real-time monitoring data and the current working conditions. If the correlation is lower than the set threshold, it is prompted that there may be measurement errors or system failures.

[0167] This section of steps verifies the accuracy of the temperature by comparing the real-time monitoring data and the historical data. Combining with the historical data database, use data mining techniques to analyze the historical data and extract the temperature change patterns under different working conditions. During the verification process, not only the differences in temperature values are compared, but also the characteristics such as the trends of temperature changes and the fluctuation ranges are compared. At the same time, perform a correlation analysis between the real-time monitoring data and the data of other sensors (such as the water temperature sensor near the umbilical cable 6). If the correlation is lower than the set threshold, it is prompted that there may be measurement errors or system failures.

[0168] S800: Result output: Output and display the temperature monitoring results of the umbilical cable 6.

[0169] This step can output the temperature monitoring results of the umbilical cable 6 in an intuitive and diverse way. In addition to the common digital display and graphical display, the data can also be transmitted to other monitoring systems or remote control centers using a standardized communication protocol (such as Modbus, OPC, etc.) for integration into the entire offshore engineering monitoring system.

[0170] This embodiment innovatively incorporates the motion parameters of the winch 8 into the design of the data processing algorithm, establishes a technical system including a filtering mechanism with rotational speed adaptability and a CNN network guided by temperature characteristics, etc., and solves the problem of temperature monitoring distortion under dynamic deployment conditions. The signal-to-noise ratio reaches 18.7 dB under sea state 4 (9.2 dB for the traditional method), the dynamic accuracy has a maximum error of 0.28 °C under the emergency stop condition of the winch 8 (2.3 °C for the traditional method), and the response speed has an average delay of <180 ms from signal acquisition to temperature output.

[0171] Through the special measurement principle of the fiber optic sensor 7, high-precision signal acquisition and demodulation, and advanced data processing algorithms, the present invention can achieve high-precision monitoring of the temperature of the deepwater umbilical cable 6, with the error controllable within a very small range, providing accurate data support for the status assessment of the umbilical cable 6. From the rapid acquisition of signals, real-time demodulation to the rapid processing of data and result output, the whole process is completed in a short time, which can timely reflect the temperature change of the umbilical cable 6. Once the temperature is abnormal during the winding and unwinding process of the umbilical cable 6 winch 8, a rapid response can be made. The long-distance monitoring ability of the distributed fiber optic sensing technology itself, combined with the design of each component in the present invention (such as the smooth ring 5 to ensure signal transmission stability), enables the system to stably monitor the temperature of the long-distance deepwater umbilical cable 6, without being affected by the length and complex movement of the umbilical cable 6. Multiple links in the special algorithm, such as data preprocessing, result verification, etc., and the redundant design of each component of the system (such as the self-calibration function of the demodulator), jointly ensure the high reliability of the system and reduce the possibility of false alarms and missed alarms. The design of the system and algorithm fully considers the complexity of the marine environment, such as factors like high pressure, high corrosion, and electromagnetic interference, enabling the present invention to operate stably in the harsh deepwater marine environment and effectively monitor the temperature of the umbilical cable 6.

[0172] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A method for processing temperature monitoring data of a deep - water umbilical cable (6), characterized in that, It includes the following steps: S100: Line test: Start the temperature monitoring host (1) and check the connectivity of the line; S200: Signal reception: Obtain the data signal of the fiber optic sensor (7) through the data acquisition unit (2); S300: Signal demodulation: Demodulate the temperature-related information from the electrical signal transmitted by the data acquisition unit (2) through the demodulator (3), and preprocess the demodulated data to remove significantly abnormal data; S400: Data processing: Filter and denoise the extracted temperature information through the data processing unit (4); Step S400 includes: S410: Filter the extracted temperature signal using an adaptive filter; Step S410 includes: S411: Generation of multi-source reference signals. The adaptive filter introduces the original input signal of the extracted temperature, fuses the water temperature data and the water flow data, and generates a composite environmental noise multi-source reference template signal; S412: Dynamic step size adjustment. The adaptive filter introduces the rotation speed signal of the winch (8) of the umbilical cable (6), and establishes a dynamic step size factor adjuster. When the rotation speed v of the winch (8) is less than or equal to 30 revolutions per minute, the step size factor μ = 0.02 + 0.03×v / 60; when the rotation speed v of the winch (8) is greater than 30 revolutions per minute, the step size factor μ = 0.05; S413: Error feedback closed loop. Filter the output data signal through the adaptive filter, and output the error signal through the error calculation module. The error signal is input into the weight update module and the historical database at the same time. The weight update frequency of the adaptive filter is dynamically associated with the rotation speed of the winch (8). When the rotation speed sensor detects that the rotation speed > 30 rpm, start the fast convergence mode, adjust the step size factor μ of the adaptive filtering algorithm from 0.01 to 0.05, and shorten the weight update period to 5 milliseconds; S420: Denoise using the noise suppression technology based on the CNN convolutional neural network; The convolutional neural network in step S420 adopts a three-layer progressive convolutional kernel group design, including a convolutional kernel group matching the fundamental vibration frequency of the umbilical cable: The first layer is a 3×1 convolutional kernel, which captures the fundamental vibration characteristics of the umbilical cable, corresponding to the fundamental frequency bandwidth of 2 - 5 Hz; The second layer is a 5×1 convolutional kernel, which extracts the harmonic characteristics of material deformation, corresponding to the harmonic bandwidth of 5 - 15 Hz; The third layer is a 7×1 convolutional kernel, which captures the characteristics of vortex-induced vibration of seawater, corresponding to the fundamental frequency bandwidth of 15 - 50 Hz; The specific steps are as follows: S421: Signal standardization preprocessing: The input layer receives the time-domain vibration signal fused by multiple sensors and performs the following preprocessing operations on the original signal: Baseline calibration: Use a sliding mean filter to eliminate baseline drift, and the filter window length is 1024 sampling points; Dynamic normalization: according to the formula perform window standardization, where are the mean and standard deviation within the current sliding window respectively, is a small constant to prevent division by zero; S422: Multi-scale vibration feature extraction: Extract vibration features in different frequency bands through a three-level convolutional kernel group in sequence: Fundamental frequency vibration layer: Configure a 3×1 convolutional kernel, the number of channels is 32, the step size is 1, and the activation function uses Leaky ReLU with a leakage coefficient of 0.2; The frequency response range of the convolutional kernel covers the fundamental vibration frequency of the umbilical cable of 2 - 5 Hz, which is used to capture the overall vibration mode of the cable structure; Harmonic feature layer: Configure a 5×1 convolutional kernel, with 64 channels, a stride of 2, and the activation function being ReLU; The frequency domain coverage range is 5 - 15 Hz, specifically extracting the characteristics of second - harmonic and higher - harmonic generated by material deformation; Vortex - induced vibration layer: Configure a 7×1 dilated convolutional kernel, with a dilation rate of 2, 128 channels, and a stride of 1; The effective receptive field covers the high - frequency band of 15 - 50 Hz, specifically capturing the broadband perturbation characteristics generated by seawater flow - induced vortex - induced vibration; S423: Adaptive feature enhancement: Set a multi - modal feature fusion module at the output end of the third convolutional layer, including: Channel attention mechanism: Dynamically calculate the weight values of each feature channel through the SE module to achieve self - recalibration of feature channels; Cross-layer residual connection: The fundamental frequency features of the first convolutional layer and the high-frequency features of the third convolutional layer are weighted and superimposed, and the formula is expressed as: ; wherein is an adaptive weight coefficient, and the value range is [0.3, 0.7]; Noise - aware regularization: Dynamically adjust the Dropout ratio according to the real - time signal - to - noise ratio, and enable strong regularization when the signal - to - noise ratio is lower than 20 dB; S424: Temperature regression output (P9 - P10): The fully - connected layer adopts a physically - constrained network structure: Dimension - decreasing design: Sequentially set hidden layers of 128 dimensions and 64 dimensions, and the dimension of the final output layer is 1; Nonlinear mapping: The last layer uses the Sigmoid function to constrain the output value to the interval [0, 1], and maps it to the material's temperature tolerance range through a linear transformation: ; wherein are respectively the minimum tolerance temperature and the maximum critical temperature of the umbilical cable material; Physical constraint mechanism: Embed a hard - boundary limiting layer to forcibly correct abnormal prediction values that exceed the material tolerance range; S500: Feature extraction: Extract signal features related to temperature changes in the processed data; S600: Estimate temperature: Estimate the temperature through the thermo - optic effect, elasto - optic effect of the fiber optic sensor (7) and the heat conduction model of the umbilical cable (6); S700: Result verification: Verify the accuracy of the temperature by comparing real - time temperature evaluation data and historical data; S800: Result output: Output and display the temperature monitoring results of the umbilical cable (6).

2. A method for processing temperature monitoring data of a deep - water umbilical cable (6) according to claim 1, characterized in that, Step S300 includes: S310: Adopt an adaptive demodulation algorithm to automatically adjust the demodulation parameters according to the real - time characteristics of the signal to obtain the best demodulation effect; S320: Conduct a preliminary quality assessment on the demodulated signal and remove significantly abnormal demodulated data.

3. A method for processing temperature monitoring data of a deep - water umbilical cable (6) according to claim 1, characterized in that, Step S400 includes: Use the cumulative average and wavelet denoising algorithms to process the demodulated data and extract the temperature information monitored by the fiber optic sensor (7).

4. A method for processing temperature monitoring data of a deep - water umbilical cable (6) according to claim 1, characterized in that, Step S600 includes: S610: According to the umbilical cable structure model, use machine learning methods to correct and optimize the temperature model; S620: Train and verify the hybrid model through experimental data and on - site data to improve the accuracy of temperature estimation.

5. A method for processing temperature monitoring data of a deepwater umbilical cable (6) according to claim 1, characterized in that, Step S700 includes: S710: Use data mining techniques to analyze historical data, extract temperature change patterns under different working conditions, and compare the differences in temperature values, the trends of temperature changes, and the fluctuation ranges; S720: Combine the above analysis to conduct a correlation analysis between real - time monitoring data and the current working condition. If the correlation is lower than the set threshold, it indicates a measurement error or a system failure.

6. A temperature monitoring system for a deep - water umbilical cable (6), characterized in that, Using the method for processing temperature monitoring data of the deepwater umbilical cable (6) according to any one of claims 1-5, the system includes a temperature monitoring host (1), a data processing unit (4), a winch (8) for coiling the umbilical cable (6), and an umbilical cable (6) for connecting an underwater device (10). An optical fiber sensor (7) is provided in the umbilical cable (6), and the temperature monitoring host (1) communicates with the optical fiber sensor (7); The temperature monitoring host (1) includes a data acquisition unit (2) and a demodulator. The data acquisition unit (2) is used to convert the weak optical signal transmitted by the optical fiber sensor (7) into an electrical signal, and the demodulator (3) is used to accurately demodulate the temperature information from the electrical signal transmitted by the data acquisition unit (2); The data processing unit (4) is used to receive the temperature information transmitted by the demodulator (3) and perform data processing. The temperature monitoring host (1) is used to display the data information processed by the data processing unit (4); It includes a slip ring (5) for connecting the temperature monitoring host (1) and the optical fiber sensor (7). One side of the slip ring (5) is connected to the interface of the temperature monitoring host (1) through a pigtail, and the pigtail on the other side is connected to the optical fiber sensor (7) by fusion splicing or an optical fiber connector; It includes a fixed pulley block (9) and an underwater device (10); The fixed pulley block (9) is used for guiding during the pay-in and pay-out process of the umbilical cable (6) when it bends, enters the water, and exits the water; The underwater device (10) is connected to the underwater end of the umbilical cable (6) and is powered, communicates, and is towed for deployment and recovery through the umbilical cable (6) to carry out underwater operations.

Images