A method and system for identifying structural damage of an offshore jacket platform

Abstract

The application discloses a kind of marine jacket platform structure damage identification method and system, belong to marine platform structure damage identification technical field, by classification, different categories of reference data in reference database are obtained, respectively determine the statistical value of damage sensitive feature corresponding to different categories of reference data and damage threshold value;When the statistical value of damage sensitive feature corresponding to test data is greater than the damage threshold value corresponding to the reference data of current category, it is determined that damage occurs to jacket platform structure, and field detection and maintenance process is started;When actually no damage occurs to field detection, then the test data is taken as new working condition sample, is included in reference database and class label is updated;Otherwise, the test data is taken as normal data, normal data is classified, and is included in the reference data of corresponding category.The method considers the influence of excitation environment change on structure damage identification result, greatly reduces the uncertainty of marine platform structure damage identification result.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of offshore platform structure damage identification, and more particularly to a method and system for offshore jacket platform structure damage identification. BACKGROUND

[0002] The jacket platform is a kind of offshore engineering structure, which provides a place for offshore operation and life for the development and utilization of marine resources. Therefore, it is very meaningful to monitor the health of the jacket platform structure.

[0003] Because the system characteristics of the offshore platform structure will be affected by the changes in the external excitation environment, resulting in non-stationary changes, the existing structure damage identification method cannot guarantee the accuracy and reliability of the system identification results. The operation environment and characteristics of the offshore platform determine that the damage identification of the offshore platform structure must be carried out under the condition of environmental excitation. Therefore, it is the key to solve the problem of damage identification of the offshore platform structure under the condition of environmental excitation to clarify the influence of the excitation environment change on the structure damage identification.

[0004] The traditional damage identification method of the offshore platform structure under the environmental excitation mainly uses the global damage identification method. The output vibration response of the monitored structure is used to reflect the overall condition of the structure. Acceleration sensors are placed on each node of the structure. The measured acceleration signals are converted into indexes reflecting the existence of damage by combining algorithms, and then the structure damage is identified. Real-time data and probability statistical analysis methods are usually used to cope with the changes in the excitation environment, but the influence of the excitation environment change on the system characteristics of the structure is not considered in essence. There is great uncertainty in the identification results, which greatly increases the cost of health monitoring of the offshore platform structure. In view of this, the prior art discloses a subspace Mahalanobis distance damage detection method resistant to excitation covariance change. This method considers the influence of excitation covariance change on damage identification results on the basis of the random subspace method. The influence of excitation environment change is eliminated by standardizing the excitation covariance Hankel matrix, but its research object is mainly simple structures such as simple beams or indoor models, which cannot adapt to the complex working environment of the jacket platform, resulting in inaccurate damage identification results. SUMMARY

[0005] In view of the problems existing in the above-mentioned field, the present application provides a method and system for offshore jacket platform structure damage identification. Since different types of data often correspond to different states of the jacket platform structure, the statistical value of the damage sensitive feature and the damage threshold value of each type of data can be determined after establishing a reference database and classifying it, so that the judgment of the structure damage is more accurate.

[0006] To solve the above technical problems, the present application discloses a method for offshore jacket platform structure damage identification, comprising the following steps:

[0007] acquire historical vibration monitoring signals of the jacket platform structure in the field;

[0008] extract signal components in the historical vibration monitoring signals, classify and label the signal components, and establish a reference database; acquire reference data of different categories in the reference database, and determine statistical values and damage thresholds of damage-sensitive features corresponding to the reference data of different categories;

[0009] acquire the vibration monitoring signals of the jacket platform structure in real time as test data, determine that the jacket platform structure has been damaged when the statistical value of the damage-sensitive features corresponding to the test data is greater than the damage threshold corresponding to the reference data of the current category, and start the on-site detection and maintenance process; when it is detected that the jacket platform structure has not actually been damaged, the test data is taken as a new working condition sample, is included in the reference database, and the category label is updated;

[0010] when the statistical value of the damage-sensitive features corresponding to the test data does not exceed the damage threshold corresponding to the reference data of the current category, the test data is taken as normal data, the normal data is classified, and is included in the reference data of the corresponding category.

[0011] Preferably, the acquisition of the historical vibration monitoring signals of the jacket platform structure in the field specifically comprises:

[0012] acquire the real size of the jacket platform structure to be measured, and construct a finite element model of the jacket platform structure by using ANSYS software;

[0013] perform numerical simulation on the jacket platform structure based on the finite element model of the jacket platform structure, and acquire a numerical simulation result;

[0014] determine a layout scheme of the sensors according to the numerical simulation result by using an effective independence method, a MAC criterion method, and a modal kinetic energy method;

[0015] optimize the determined layout scheme of the sensors by using a multi-objective Liechtenberg algorithm, acquire an optimal layout scheme of the sensors, and the optimal layout scheme comprises an optimal number and a layout position of the sensors;

[0016] build a data acquisition system according to the optimal layout scheme of the sensors;

[0017] acquire the historical vibration monitoring signals of the jacket platform structure in the field by using the data acquisition system.

[0018] Preferably, the extraction of the signal components in the historical vibration monitoring signals, the classification and labeling of the signal components, and the establishment of the reference database specifically comprise:

[0019] The obtained historical vibration monitoring signals are divided into data sets according to preset acquisition time intervals, and the signals in the data sets are sequentially subjected to baseline adjustment, detrending processing and standard deviation standardization processing;

[0020] The processed signals are subjected to low-pass filtering through a filter to remove high-frequency noise and electromagnetic field interference in the signals;

[0021] The number of decomposition and filter parameters of the reduced-order variational modal decomposition method are determined through a particle swarm optimization algorithm, and the signals after low-pass filtering are subjected to signal decomposition to extract signal components;

[0022] The center frequency values corresponding to each signal component and the number thereof are determined, the signal components are classified and labeled using the Mahalanobis distance method, and a reference database is established.

[0023] Preferably, the reference data of different categories in the reference database is obtained, and the statistical values and damage thresholds of the damage-sensitive features corresponding to the reference data of different categories are determined, specifically including:

[0024] Based on the covariance-driven stochastic subspace method, the covariance Hankel matrix of a group of reference data of each category is determined ;

[0025] The groups of reference data of each category are connected head to tail to form a group of long data, and the covariance Hankel matrix of the overall data of the category is obtained ;

[0026] Wherein, the dimension of the Hankel matrix and the system order are determined by the number corresponding to each signal component and the number of sensors;

[0027] The damage-sensitive feature is established as:

[0028] ;

[0029] The reference database is divided into K groups, and the data length of each group is K b The covariance matrix of the reference data of the category is estimated as:

[0030] ;

[0031] In the formula, Hankel matrix of the reference data of the k th group;

[0032] According to the damage-sensitive feature and the covariance matrix , the statistical value of the damage sensitive feature is established is:

[0033] ;

[0034] wherein T is a transpose symbol;

[0035] Based on the central limit theorem and the law of large numbers, it is determined that the statistical value of the damage sensitive feature satisfies the Gaussian normal distribution;

[0036] Based on the statistical value of the damage sensitive feature of the jacket platform structure when there is no damage, the probability distribution is fitted, and the damage threshold is determined by using the 3δ principle.

[0037] Preferably, the vibration monitoring signal of the jacket platform structure acquired in real time is taken as test data, when the statistical value of the damage sensitive feature corresponding to the test data is greater than the damage threshold corresponding to the reference data of the current category, it is determined that the jacket platform structure is damaged, and the on-site detection and maintenance process is started; when it is detected on site that the jacket platform structure is actually not damaged, the test data is taken as a new working condition sample, and is included in the reference database and the category label is updated, which specifically includes:

[0038] The vibration monitoring signal of the jacket platform structure is acquired in real time by the same data acquisition system as the historical vibration monitoring signal, and is taken as test data;

[0039] The test data is classified by the Mahalanobis distance method, and test data of different categories is obtained;

[0040] Taking any selected category of reference data as a reference, the statistical value of the damage sensitive feature corresponding to the test data of the current category is obtained;

[0041] The greater the statistical value, the higher the damage degree of the jacket platform structure;

[0042] When the statistical value of the damage sensitive feature corresponding to the test data is greater than the damage threshold corresponding to the reference data of the current category, it is determined that the jacket platform structure is damaged;

[0043] When it is determined that the jacket platform structure is damaged, the on-site detection and maintenance process is started; when it is detected on site that the jacket platform structure is actually not damaged, the test data is taken as a new working condition sample, is added to the reference database, and the category label of the reference database is updated, and the reference data of different categories in the updated reference database is obtained;

[0044] According to the reference data of different categories in the updated reference database, the statistical value of the damage sensitive feature corresponding to the reference data and the damage threshold are updated;

[0045] Also included is taking temporary regulatory measures on the potential risk area corresponding to the new working condition sample until the risk is excluded.

[0046] Preferably, when the statistical value of the damage sensitive feature corresponding to the test data does not exceed the damage threshold value corresponding to the reference data of the current category, the test data is regarded as normal data, the normal data is classified, and the normal data is included in the reference data of the corresponding category, and specifically includes:

[0047] When the statistical value of the damage sensitive feature corresponding to the test data does not exceed the damage threshold value corresponding to the reference data of the current category, the test data is regarded as normal data, the normal data is classified by the Mahalanobis distance method, each data in the normal data is assigned to the category corresponding to the reference data with the smallest Mahalanobis distance by determining the Mahalanobis distance between each data in the normal data and each reference data in each category in the reference database, and the statistical value and the damage threshold value of the damage sensitive feature corresponding to the reference data are updated.

[0048] Preferably, the application further includes a marine jacket platform structure damage identification system, comprising:

[0049] The data acquisition module is used to acquire historical vibration monitoring signals of the field jacket platform structure.

[0050] The reference data damage threshold value determination module is used to extract signal components in the historical vibration monitoring signals, classify and label the signal components, and establish a reference database; the reference data of different categories in the reference database is acquired, and the statistical value and the damage threshold value of the damage sensitive feature corresponding to the reference data of different categories are determined respectively.

[0051] The damage detection module is used to acquire the vibration monitoring signals of the jacket platform structure in real time as test data, determine that the jacket platform structure has damage when the statistical value of the damage sensitive feature corresponding to the test data is greater than the damage threshold value corresponding to the reference data of the current category, and start the field detection and maintenance process; when it is detected in the field that the jacket platform structure has not actually occurred damage, the test data is regarded as a new working condition sample, is included in the reference database, and the category label is updated; when the statistical value of the damage sensitive feature corresponding to the test data does not exceed the damage threshold value corresponding to the reference data of the current category, the test data is regarded as normal data, the normal data is classified, and the normal data is included in the reference data of the corresponding category.

[0052] Compared with the prior art, the application has the following beneficial effects:

[0053] The proposed method for identifying structural damage in offshore jacket platforms addresses this issue by establishing a reference database and classifying the data. This allows for the determination of statistical values ​​for damage-sensitive features and damage thresholds for each data category, leading to more accurate damage assessment. By comparing the statistical values ​​of damage-sensitive features corresponding to real-time acquired test data with the damage thresholds corresponding to the current category of reference data, the method quickly determines whether the test data has caused damage to the jacket platform structure. This enables the timely detection of early structural damage, saving valuable time for subsequent maintenance and reinforcement work and preventing further damage and more serious consequences. Simultaneously, the real-time database update mechanism not only increases the amount of reference data, improving the reliability of damage detection results, but also allows for the timely addition of reference data under new operating conditions, expanding the applicability of the historical database. Attached Figure Description

[0054] Figure 1 This is a flowchart of the marine jacket platform structural damage identification method proposed in this invention;

[0055] Figure 2 Candidate sensor locations and final optimized positions on the marine jacket platform structure provided in this embodiment of the invention;

[0056] Figure 3 This invention provides a set of multi-channel acquired vibration acceleration signals and their corresponding power spectral density maps.

[0057] Figure 4 The spectrum diagram of the signal components and corresponding center frequencies extracted after processing by the RVMD method is provided in the embodiment of the present invention;

[0058] Figure 5 This is the damage identification result of a set of monitoring data of a marine jacket platform structure under different excitation environments and in a damaged state, provided in an embodiment of the present invention. Detailed Implementation

[0059] The following will refer to the appendices in the embodiments of the present invention. Figures 1-5 The technical solutions in the embodiments of the present invention will be clearly and completely described. It should be understood that the terminology used in the present invention is only for describing particular implementation methods and is not intended to limit the present invention.

[0060] Example

[0061] like Figure 1 As shown, this invention proposes a method for identifying structural damage to a marine jacket platform, comprising the following steps:

[0062] S1: Acquire historical vibration monitoring signals of the on-site jacket platform structure;

[0063] S2: extracting signal components in the historical vibration monitoring signal, classifying and labeling the signal components, and establishing a reference database; obtaining reference data of different categories in the reference database, and respectively determining statistical values of damage sensitive features corresponding to the reference data of different categories and damage thresholds;

[0064] S3: taking the vibration monitoring signal of the jacket platform structure obtained in real time as test data, determining that the jacket platform structure is damaged when the statistical value of the damage sensitive feature corresponding to the test data is greater than the damage threshold corresponding to the reference data of the current category, and starting the on-site detection and maintenance process; when it is detected on site that the jacket platform structure has not actually been damaged, the test data is taken as a new working condition sample, is included in the reference database, and the category label is updated;

[0065] When the statistical value of the damage sensitive feature corresponding to the test data does not exceed the damage threshold corresponding to the reference data of the current category, the test data is taken as normal data, the normal data is classified, and the normal data is included in the reference data of the corresponding category.

[0066] The method can significantly improve the adaptability of the monitoring system of the offshore jacket platform in a complex environment by optimizing the sensor layout scheme, upgrading the hardware of the collection system, and combining dynamic maintenance of the reference database and robustness design of the algorithm, thereby improving the reliability of damage identification.

[0067] Specifically, in step S1, the vibration monitoring signal of the field jacket platform structure is obtained, specifically including:

[0068] The real size of the jacket platform structure to be measured is obtained, and an ANSYS software is used to construct a finite element model of the jacket platform structure.

[0069] Based on the finite element model of the jacket platform structure, modal analysis is performed, the jacket platform structure is numerically simulated, and a numerical simulation result is obtained.

[0070] According to the numerical simulation result, the effective independent method, the MAC criterion method and the modal kinetic energy method are used to preliminarily determine the layout scheme of the sensor, and the initial arrangement point of the acceleration sensor of the upper platform of the jacket is determined.

[0071] A multi-objective Lichtenberg algorithm (MOLA) is used to optimize the preliminarily determined layout scheme of the sensor, with the maximum value of the characteristic value vector product of the initial arrangement point and the minimum value of the number of sensors as the objective function, to obtain an optimal layout scheme of the sensor, including the optimal number and layout position of the sensor, such as Figure 2As shown, the optimal sensor has 4, respectively, layout in the upper platform of the jacket top.

[0072] According to the layout scheme of the optimal sensor, the low-frequency acceleration sensor is arranged in sequence, connected to the multi-channel data acquisition instrument by cable, transmitted to the computer through the network cable, and the data acquisition system is built.

[0073] Through the data acquisition system, the historical vibration monitoring signals of the field jacket platform structure are collected.

[0074] By constructing the finite element model of the jacket platform structure and optimizing the sensor layout scheme, the adaptability of the signal jacket platform monitoring system in complex environment can be significantly improved, thereby improving the reliability of damage identification.

[0075] In step S2, the collected historical vibration monitoring signals of the field jacket platform structure are divided into data sets according to the preset collection time interval (10 minutes), and the signals in the data sets are sequentially subjected to baseline adjustment, detrending and standard deviation standardization, as shown in Figure 3 The processed signal includes time domain waveform (left) and frequency spectrum (right).

[0076] In the time domain waveform, the horizontal coordinate represents time t , unit: seconds (s). It describes the change of the signal with time, showing the value of the signal at different times, and the vertical coordinates "Leg1#-X / g", "Leg1#-Y / g", etc. represent the acceleration signal in the corresponding direction (X direction or Y direction), unit: gravity acceleration g, which reflects the vibration acceleration of the structure in the corresponding direction at a certain time.

[0077] In the frequency spectrum, the horizontal coordinate represents frequency f , unit: hertz (Hz), which shows the different frequency components contained in the signal, and the vertical coordinate is power spectral density (PSD), unit: (g 2 / Hz). PSD describes the distribution of signal power at different frequencies, i.e. the contribution of each frequency component to the total power of the signal. The larger the PSD value, the higher the energy proportion of the frequency component in the signal.

[0078] The processed signal is low-pass filtered by a filter to remove high-frequency noise and electromagnetic field interference in the signal.

[0079] The decomposition number and filter parameters of the reduced-order variational mode decomposition (RVMD) are determined by a particle swarm optimization algorithm, and the signal decomposition is performed on the low-pass filtered signal. Effective signal components are extracted from historical vibration monitoring signals of multiple sensors, as shown in Figure 4 The number corresponding to each effective signal component is determined, and the center frequency value of the effective signal component is obtained by recording.

[0080] The decomposition number and filter parameters of the RVMD are determined by a particle swarm optimization algorithm, and the search process of the two parameters is divided into two steps, including:

[0081] First, fix the initial value of the filter parameter, search for the optimal decomposition number, perform RVMD decomposition on the signal components under different decomposition numbers, calculate the maximum value of the correlation coefficient between the signal components, and let it be less than the threshold value Calculate the energy proportion of each signal component, and let the minimum value of the energy proportion be greater than the threshold value Take the minimum value of the decomposition number that meets the condition as the optimal value;

[0082] Second, fix the value of the decomposition number, perform RVMD decomposition on the signal components under different filter parameter settings, calculate the total energy entropy of each signal component, take the filter parameter with the minimum signal energy entropy as the optimal value, and calculate the energy proportion of each signal component. Set the signal component with an energy proportion greater than the threshold value as effective, and record the center frequency value of the effective signal component.

[0083] Using the Mahalanobis distance method, the center frequency value and the number of each effective signal component are used as a sample data to construct a two-dimensional feature vector of the sample data. By determining the mean and covariance matrix of the feature vector of each category, the center frequency value and the number of effective signal components are classified and labeled, and a reference database is constructed.

[0084] When the reference database increases new data samples, each new data sample is assigned to the category of the reference data with the minimum Mahalanobis distance.

[0085] According to the assignment result, a category label is added to the new data sample, and the reference database is updated.

[0086] As shown in Figure 4 Two signal components (IMF1 and IMF2) and their corresponding frequency spectra extracted after RVMD decomposition are analyzed Figure 4The left graph in Figure 1 shows the waveform changes of two signal components (IMF1 and IMF2) in the time domain (the horizontal axis represents time in seconds s). t The fluctuations in the amplitude (vertical axis, unit: g) of IMF1 and IMF2 over time reflect the dynamic characteristics of the respective signal components.

[0087] Figure 4 The horizontal axis of the right graph (spectrum graph) in Figure 1 represents frequency f in hertz Hz, and the vertical axis represents power spectral density PSD in g 2 / Hz. The frequency corresponding to the peak in each spectrum graph is the center frequency of the signal component. As can be seen from the graph, the center frequencies of IMF1 and IMF2 are both around 5 Hz, but there may be some slight differences. The center frequency reflects the main vibration frequency of the signal component, and different center frequencies correspond to different vibration modes of the structure.

[0088] As can be seen from Figure 1, Figure 4 the number of center frequencies of the signals in this category n ref = 2.

[0089] In step S3, the excitation covariance Hankel matrix of the reference data of different categories is standardized to determine the statistical value and damage threshold of the damage sensitive feature corresponding to the reference data of different categories, specifically including:

[0090] Based on the covariance-driven random subspace method, the covariance Hankel matrix of a group of reference data of each category is determined;

[0091] The groups of reference data of each category are connected end to end to form a group of long data, and the covariance Hankel matrix of the overall data of this category is obtained;

[0092] wherein the dimension of the Hankel matrix and the system order are determined by the number of each effective signal component and the number of sensors;

[0093] The dimension of the Hankel matrix p is (2 n ref +1)×number of sensors, and the system order is set to 2 n ref The present application p is 30, and the system order is set to 4.

[0094] The damage sensitive feature is established as:

[0095] ; (1)

[0096] The reference database is divided into K groups, and the data length of each group is K b The covariance matrix of the reference data of the category is estimated :

[0097] ; (2)

[0098] In the formula, Hankel matrix of the reference data of the first k group;

[0099] According to formula (1) and formula (2), the statistical value of the damage sensitive feature is established :

[0100] ; (3)

[0101] Wherein, T is the transpose symbol;

[0102] Based on the central limit theorem and the law of large numbers, it is determined that the statistical value of the damage sensitive feature satisfies the Gaussian distribution;

[0103] Based on the statistical value of the damage sensitive feature of the jacket platform structure when there is no damage, the probability distribution is fitted, and the damage threshold is determined by using the 3δ principle, including:

[0104] When the significance level of the statistical value exceeds 95%, a warning is started; when the significance level of the statistical value reaches 99%, a damage report is sent.

[0105] The warning threshold value established by the reference data of different categories in the reference database is 10.9, and the damage threshold value is 11.8, as shown in Figure 5 .

[0106] In step S3, the vibration monitoring signal of the jacket platform structure is collected in real time by the same data acquisition system of the historical vibration monitoring signal, and is taken as test data;

[0107] The test data is classified by the Mahalanobis distance method, and different categories of test data are obtained;

[0108] In different categories of test data, any category of test data is selected, and the selected any category of test data is taken as a reference to match the classification result of the reference data, to obtain the difference value of the center frequency value of the current category of test data and the reference data of the category; when the difference value meets the preset error requirement, the matching is successful, and the Hankel matrix of the test data is constructed, the excitation covariance Hankel matrix is normalized, and the Hankel matrix of each category of test data is obtained .

[0109] The obtained Hankel matrix and are singular value decomposed, and the expression of the decomposition process is as follows:

[0110] ; (4)

[0111] In the formula, E 1 is the first 2 n ref columns of the left singular value matrix, E 2 is the remaining columns of the left singular value matrix, F 1 is the first 2 n ref singular values, F 2 is the remaining singular values, is the first 2 n ref rows of the right singular value matrix related to the reference data, is the first 2 n ref rows of the right singular value matrix related to the test data, is the remaining rows of the right singular value matrix related to the reference data, is the remaining rows of the right singular value matrix related to the test data. Let the excitation environment , the excitation environment . In order to make the two data in the same excitation environment, replace in with , to obtain the excitation environment standardized . By the obtained Hankel matrix and , the damage sensitive feature and the statistical value are established, and compared with the damage threshold value, when the statistical value is greater than the damage threshold value, it is considered that the jacket platform structure has damage, and the greater the value of the statistical value , the higher the damage degree. As shown in Figure 5 , the damage identification results of two groups of damage working condition data (excitation environment 1 and excitation environment 2), wherein, the excitation environment 1 is to disassemble a horizontal support of the jacket platform structure, and the excitation environment 2 is to disassemble two diagonal supports on this basis. The excitation environment 1 is to guide the jacket platform structure in a relatively stable environmental excitation, and the excitation environment 2 is to guide the jacket platform structure in a sinusoidal excitation environment with a load of 1.5kN and a load frequency of 0.2Hz.

[0112] According to any selected category of test data, the damage sensitive feature statistical value corresponding to the current category of test data is obtained;

[0113] The greater the statistical value, the higher the damage degree of the jacket platform structure;

[0114] When the damage sensitive feature statistical value corresponding to the test data is greater than the damage threshold value corresponding to the reference data of the current category, it is determined that the jacket platform structure is damaged;

[0115] When it is determined that the jacket platform structure is damaged, the on-site detection and maintenance process is started; when it is detected that the jacket platform structure is not actually damaged, the test data is added as a new working condition data sample to the reference database, and the category label of the reference database is updated, and the reference data of different categories in the updated reference database is obtained.

[0116] According to the reference data of different categories in the updated reference database, the statistical value and the damage threshold value of the damage sensitive feature corresponding to the reference data are updated;

[0117] It also includes taking temporary regulatory measures on the potential risk area corresponding to the new working condition sample until the risk is eliminated.

[0118] When the damage sensitive feature statistical value corresponding to the test data does not exceed the damage threshold value corresponding to the reference data of the current category, the test data is regarded as normal data, the normal data is classified by the Mahalanobis distance method, each data in the normal data is assigned to the category corresponding to the minimum Mahalanobis distance of each data in the normal data and each reference data in each category in the reference database, and the statistical value and the damage threshold value of the damage sensitive feature corresponding to the reference data are updated.

[0119] As shown in FIG. Figure 5 The damage identification results of the two groups of damage working condition data of the excitation environment 1 and the excitation environment 2 are shown.

[0120] The present application also provides a jacket platform structure damage identification system for offshore platform, comprising:

[0121] The data acquisition module is used for acquiring the real size of the jacket platform structure to be measured, constructing a finite element model of the jacket platform structure, determining a layout scheme of the sensor, building a data acquisition system, and collecting historical vibration monitoring signals of the jacket platform structure on site.

[0122] The reference data damage threshold determination module is used for extracting effective signal components in historical vibration monitoring signals, classifying and labeling the signal components, and establishing a reference database; different categories of reference data in the reference database are obtained, and statistical values of damage sensitive features corresponding to the different categories of reference data and damage thresholds are determined respectively;

[0123] The damage detection module is used for taking the vibration monitoring signals of the jacket platform structure acquired in real time as test data, determining that the jacket platform structure is damaged when the statistical value of the damage sensitive features corresponding to the test data is greater than the damage threshold corresponding to the reference data of the current category, and starting an on-site detection and maintenance process; when it is detected on site that the jacket platform structure is not actually damaged, the test data is taken as a new working condition sample, is included in the reference database, and the category label is updated; when the statistical value of the damage sensitive features corresponding to the test data does not exceed the damage threshold corresponding to the reference data of the current category, the test data is taken as normal data, the normal data is classified, and is included in the reference data of the corresponding category.

[0124] The system proposed in the application transmits the output of the data acquisition module to the reference data damage threshold determination module, and then transmits the reference information output by the latter to the damage detection module, so as to realize continuous detection and evaluation of the state of the jacket platform structure. Based on the detection and evaluation results and the actual investigation situation, the system dynamically completes the update of the reference database, thereby forming a complete and self-learning marine jacket platform structure damage identification system.

[0125] Compared with the traditional marine jacket platform structure damage identification method, the application not only considers the influence of the change of the excitation environment on the identification result, but also improves the accuracy and adaptability of damage identification under different working conditions through sensor optimization arrangement, statistical analysis method application and dynamic update of the reference database, thereby providing a new and effective scheme for realizing reliable offshore platform structure health monitoring.

[0126] The above is only a preferred specific embodiment of the application, but the protection scope of the application is not limited thereto, any person skilled in the art can make equivalent replacement or change according to the technical scheme and the inventive concept of the application within the technical range disclosed by the application, which should be covered in the protection scope of the application.

[0127] In addition, all technical and scientific terms used in the application have the same meaning as generally understood by those skilled in the art to which the application belongs, unless otherwise specified. All documents mentioned in the specification are incorporated by reference to disclose and describe the methods related to the documents. In the case of conflict with any incorporated document, the content of the specification prevails.

Claims

1. A method for identifying structural damage to a marine jacket platform, characterized in that, Includes the following steps: Acquire historical vibration monitoring signals of the on-site jacket platform structure; Extract signal components from historical vibration monitoring signals, classify and label the signal components, and establish a reference database; Obtain reference data of different categories from the reference database, and determine the statistical values ​​and damage thresholds of the damage sensitivity features corresponding to the different categories of reference data; The vibration monitoring signal of the jacket platform structure acquired in real time is used as test data. When the statistical value of the damage sensitivity feature corresponding to the test data is greater than the damage threshold corresponding to the reference data of the current category, it is determined that the jacket platform structure has been damaged, and the field inspection and maintenance process is initiated. When the field inspection shows that the jacket platform structure has not actually been damaged, the test data is used as a new working condition sample, included in the reference database, and the category label is updated. When the statistical value of the damage sensitivity feature corresponding to the test data does not exceed the damage threshold corresponding to the reference data of the current category, the test data is regarded as normal data, the normal data is classified, and it is included in the reference data of the corresponding category. Specifically, the step of acquiring reference data of different categories from the reference database and determining the statistical values ​​and damage thresholds of the damage sensitivity features corresponding to different categories of reference data includes: Based on a covariance-driven stochastic subspace method, the covariance Hankel matrix of a set of reference data for each class is determined. ; By concatenating the reference data sets for each category, a long dataset is formed, yielding the covariance Hankel matrix for the overall data of that category. ; The dimensions of the Hankel matrix and the system order are determined by the number of each signal component and the number of sensors. Establish damage sensitivity characteristics for: ； Divide the reference database into K Groups, each group of data has a length of [length missing]. K b Estimate the covariance matrix of the reference data for this category. for: ； In the formula, Indicates the first k The Hankel matrix of the reference data set; Based on damage sensitivity characteristics Covariance Matrix Establish statistical values ​​for damage sensitivity characteristics. for: ； Where T is the transpose symbol; Based on the central limit theorem and the law of large numbers, the statistical values ​​of the damage sensitivity characteristics are determined to satisfy a Gaussian distribution. Based on the statistical values ​​of the damage sensitivity characteristics of the catheter platform structure when it is undamaged, its probability distribution is fitted, and the damage threshold is determined using the 3δ principle.

2. The method for identifying structural damage to a marine jacket platform according to claim 1, characterized in that, The acquisition of historical vibration monitoring signals of the on-site jacket platform structure specifically includes: Obtain the actual dimensions of the jacket platform structure to be tested, and construct a finite element model of the jacket platform structure using ANSYS software; Numerical simulation of the jacket platform structure was performed based on the finite element model of the jacket platform structure, and the numerical simulation results were obtained. Based on the numerical simulation results, the effective independence method, MAC criterion method and modal kinetic energy method are used to determine the sensor layout scheme; The multi-objective Lichtenberg algorithm is used to optimize the given sensor layout scheme and obtain the optimal sensor layout scheme, including the optimal number of sensors and their placement positions. Based on the optimal sensor layout scheme, a data acquisition system is built; Historical vibration monitoring signals of the on-site jacket platform structure were collected using a data acquisition system.

3. The method for identifying structural damage to offshore jacket platforms according to claim 1, characterized in that, The process of extracting signal components from historical vibration monitoring signals, classifying and labeling these signal components, and establishing a reference database specifically includes: After the acquired historical vibration monitoring signals are divided into datasets according to the preset acquisition time interval, the signals in the datasets are sequentially subjected to baseline adjustment, detrending processing, and standard deviation standardization. The processed signal is low-pass filtered to remove high-frequency noise and electromagnetic interference. The number of decompositions and filtering parameters of the reduced-order variational mode decomposition method are determined by the particle swarm optimization algorithm. The low-pass filtered signal is then decomposed to extract signal components. By determining the center frequency value and the number of each signal component, the signal components are classified and labeled using the Mahalanobis distance method, and a reference database is established.

4. The method for identifying structural damage to a marine jacket platform according to claim 1, characterized in that, The vibration monitoring signal of the jacket platform structure acquired in real time is used as test data. When the statistical value of the damage sensitivity feature corresponding to the test data is greater than the damage threshold corresponding to the reference data of the current category, it is determined that the jacket platform structure has been damaged, and the on-site inspection and maintenance process is initiated. When on-site testing reveals that the jacket platform structure has not actually suffered damage, the test data is used as a new working condition sample, included in the reference database, and the category label is updated, specifically including: By using the same data acquisition system that acquires historical vibration monitoring signals, the vibration monitoring signals of the jacket platform structure are acquired in real time and used as test data. The test data is classified using Mahalanobis distance to obtain different categories of test data; Using any selected category of reference data as a benchmark, obtain the statistical value of the damage sensitivity feature corresponding to the test data of the current category; The larger the statistical value, the higher the degree of damage to the jacket platform structure; When the statistical value of the damage sensitivity feature corresponding to the test data is greater than the damage threshold corresponding to the reference data of the current category, it is determined that the duct platform structure has been damaged. Once damage to the jacket platform structure is determined, the on-site inspection and maintenance process is initiated. When the on-site inspection reveals that the jacket platform structure is not actually damaged, the test data is used as a new working condition sample, added to the reference database, and the category labels of the reference database are updated to obtain reference data of different categories in the updated reference database. Based on the updated reference data of different categories in the updated reference database, update the statistical values ​​and damage thresholds of the damage sensitivity features corresponding to the reference data; This also includes taking temporary control measures for potential risk areas corresponding to new operating conditions until the risks are eliminated.

5. The method for identifying structural damage to a marine jacket platform according to claim 1, characterized in that, When the statistical value of the damage sensitivity feature corresponding to the test data does not exceed the damage threshold corresponding to the reference data of the current category, the test data is regarded as normal data, the normal data is classified, and it is included in the reference data of the corresponding category, specifically including: When the statistical value of the damage sensitivity feature corresponding to the test data does not exceed the damage threshold corresponding to the reference data of the current category, the test data is regarded as normal data. The normal data is classified by Mahalanobis distance. By determining the Mahalanobis distance between each data in the normal data and each reference data in each category in the reference database, each data in the normal data is assigned to the category of the reference data with the smallest Mahalanobis distance. The statistical value of the damage sensitivity feature and the damage threshold corresponding to the reference data are updated.

6. A damage identification system for marine jacket platform structures, characterized in that, include: The data acquisition module is used to acquire historical vibration monitoring signals of the on-site jacket platform structure; The reference data damage threshold determination module is used to extract signal components from historical vibration monitoring signals, classify and label the signal components, and establish a reference database. Obtain reference data of different categories from the reference database, and determine the statistical values ​​and damage thresholds of the damage sensitivity features corresponding to the different categories of reference data; The damage detection module uses the vibration monitoring signals of the jacket platform structure acquired in real time as test data. When the statistical value of the damage sensitivity feature corresponding to the test data is greater than the damage threshold corresponding to the reference data of the current category, it determines that the jacket platform structure has been damaged and initiates the field inspection and maintenance process. When the field inspection reveals that the jacket platform structure has not actually been damaged, the test data is used as a new working condition sample, included in the reference database, and the category label is updated. When the statistical value of the damage sensitivity feature corresponding to the test data does not exceed the damage threshold corresponding to the reference data of the current category, the test data is treated as normal data, classified, and included in the reference data of the corresponding category. Specifically, the step of acquiring reference data of different categories from the reference database and determining the statistical values ​​and damage thresholds of the damage sensitivity features corresponding to different categories of reference data includes: Based on a covariance-driven stochastic subspace method, the covariance Hankel matrix of a set of reference data for each class is determined. ; By concatenating the reference data sets for each category, a long dataset is formed, yielding the covariance Hankel matrix for the overall data of that category. ; The dimensions of the Hankel matrix and the system order are determined by the number of each signal component and the number of sensors. Establish damage sensitivity characteristics for: ； Divide the reference database into K Groups, each group of data has a length of [length missing]. K b Estimate the covariance matrix of the reference data for this category. for: ； In the formula, Indicates the first k The Hankel matrix of the reference data set; Based on damage sensitivity characteristics Covariance Matrix Establish statistical values ​​for damage sensitivity characteristics. for: ； Where T is the transpose symbol; Based on the central limit theorem and the law of large numbers, the statistical values ​​of the damage sensitivity characteristics are determined to satisfy a Gaussian distribution. Based on the statistical values ​​of the damage sensitivity characteristics of the catheter platform structure when it is undamaged, its probability distribution is fitted, and the damage threshold is determined using the 3δ principle.

Images