A machine learning based offshore engineering data management system and method

Abstract

The application discloses a kind of ocean engineering data management system and method based on machine learning, it is related to marine engineering equipment fault diagnosis technical field, system includes multidimensional sensor array deployment module, noise feature topology library management module, real-time noise separation and feature extraction module, fault rule chain deduction module, multidimensional fault determination module and intelligent operation and maintenance feedback module.Vibration data is collected by deploying sensor in equipment stress concentration position, after phase space reconstruction, noise separation, extract fault feature vector, then utilize rule chain deduction fault type and position, finally generate structured repair report and push.The system solves the problem of multi-source noise interference under marine environment, positioning accuracy is insufficient, etc., realizes that residual noise energy proportion after noise separation is less than 5%, fault position error is less than or equal to 0.1mm, reduces comprehensive misjudgment rate, improves the operation and maintenance efficiency and reliability of marine engineering equipment.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of marine engineering equipment fault diagnosis, and particularly relates to a marine engineering data management system and method based on machine learning. BACKGROUND

[0002] Marine engineering equipment is long-term served in extreme environments such as high salt mist, strong corrosion and complex sea conditions, and its fault diagnosis and operation and maintenance face multiple technical bottlenecks:

[0003] Multi-source noise interference: traditional vibration monitoring methods are difficult to accurately extract equipment fault features from background noise such as sea current impact and tidal fluctuation, resulting in a misjudgment rate of more than 30%.

[0004] Environmental parameter coupling effect: the running state of the equipment is dynamically disturbed by parameters such as sea current speed and tidal phase, for example, when the sea current speed is greater than 2m / s, the bearing vibration amplitude threshold needs to be raised by 20% to effectively identify faults, but the existing system lacks a dynamic threshold correction mechanism.

[0005] Insufficient spatial positioning accuracy: the spatial error between the fault location and the vibration source often exceeds 0.5mm, which cannot meet the precise maintenance needs of marine engineering equipment, for example, if the positioning error of the loose blade tenon is greater than 0.1mm, it may cause chain structure damage.

[0006] Lagging nature of operation and maintenance decision: the traditional process relies on manual analysis reports, and the average time from fault discovery to maintenance execution is more than 24 hours, which is difficult to cope with the high economic loss caused by the shutdown of deep-sea equipment. SUMMARY

[0007] The purpose of the present application is to provide a marine engineering data management system and method based on machine learning to solve the problems in the prior art.

[0008] To achieve the above purpose, the present application provides the following technical scheme:

[0009] The marine engineering data management system based on machine learning comprises a multi-dimensional sensing array deployment module, a noise feature topology library management module, a real-time noise separation and feature extraction module, a fault rule chain deduction module, a multi-dimensional fault judgment module and an intelligent operation and maintenance feedback module; the multi-dimensional sensing array deployment module deploys 9 MEMS micro-vibration sensors at the blade surface, the blade root, the middle of the blade, the blade tip, the contact points of the inner and outer rings of the bearing, the main shaft section, the coupling joint and the bearing support to synchronously collect sub-micron level vibration signals; the noise feature topology library management module constructs a feature library containing all marine environmental noise modes, and each mode stores its topology feature parameters; the real-time noise separation and feature extraction module maps the real-time collected vibration data to a phase space to construct a topology network, separates the noise by calculating the Hamming distance between the vibration data and the modes in the noise library, and extracts a fault feature vector; the fault rule chain deduction module establishes a vibration feature-fault type logical rule chain based on a historical fault case library, which contains a three-layer reasoning mechanism of feature screening, environmental parameter filtering and logical operation; the multi-dimensional fault judgment module inputs the extracted feature vector into the rule chain for deduction, matches through multi-level logical expressions, and outputs the fault type, position and severity; and the intelligent operation and maintenance feedback module generates a structured maintenance report according to the fault judgment result, which contains fault position three-dimensional labeling, maintenance priority sorting, historical similar case solution reference, and is pushed to the operation and maintenance personnel terminal in real time.

[0010] The multi-dimensional sensing array deployment module comprises a sub-micron level sensor deployment unit and a time-space consistency synchronous acquisition unit.

[0011] The sub-micron level sensor deployment unit selects a MEMS capacitive acceleration sensor with a range of ±50g and a resolution of 0.1ug, adopts three-dimensional coordinate positioning technology according to the equipment structure dynamics, and deploys the sensor at 9 stress concentration positions, i.e. the blade surface, the blade root, the middle of the blade, the blade tip, the contact points of the inner and outer rings of the bearing, the main shaft section, the coupling joint and the bearing support, to collect real-time sub-micron level vibration signals;

[0012] The spatio-temporal consistency synchronous acquisition unit uses an intra-equipment laser tracker to acquire three-dimensional coordinates of nine sensor physical installation positions and structural nodes of the sensor installation positions, and establishes a mapping table of sensor number-three-dimensional coordinates-structural nodes; for the vibration data synchronously acquired by the sensor, the Kalman filtering algorithm is used to correct the errors of the channel time stamps, and the multi-channel data is aligned to a unified time axis with the Beidou satellite clock as the reference; then, according to the mapping table, the vibration data at each time point is associated with the three-dimensional coordinates of the corresponding sensor, and a structured data record containing the time stamp, the three-dimensional coordinates and the vibration amplitude is generated, all records are arranged in chronological order, forming a multi-dimensional vibration data set stored in the HDF5 format, and each data block contains the time dimension, the spatial dimension and the vibration amplitude dimension, facilitating the real-time noise separation and feature extraction module to call.

[0013] The noise feature topology library management module includes a historical data import unit, a phase space topology feature reconstruction unit and a noise mode clustering modeling unit.

[0014] The historical data import unit extracts all historical vibration data of marine environmental noise from a device full life cycle fault case library, and the data source needs to be labeled with the power plant name, the acquisition time and the environmental parameters, and then the unstructured data is converted into a standardized format by an ETL tool: each record contains a time sequence, environmental parameters and a noise type label;

[0015] The phase space topology feature reconstruction unit performs five-dimensional phase space reconstruction on the historical imported noise time domain data based on the Takens embedding theorem, specifically: taking the noise time domain sequence x(t) as the input, generating a five-dimensional phase space vector according to the formula X(t) = [x(t), x(t+T), x(t+2T), x(t+3T), x(t+4T)], wherein the embedding dimension m is determined by the false neighbor method, and the time delay T is determined by the first minimum value of the mutual information function; after reconstruction, the Delaunay triangulation algorithm is used to topologically connect the phase space point set, and through the empty circle property, no other nodes are contained in the incircle of any triangle, forming a topological network reflecting the data distribution density; finally, three types of topological indexes of the topological network are calculated in real time, which are node connectivity, shortest path length and betweenness centrality; for node connectivity, the number of connections of each node with adjacent nodes is counted, and the average number of connections of all network nodes is taken as the node connectivity; for the shortest path length, the shortest path distance between all node pairs is calculated, and the average value of the whole network is taken as the global shortest path length; for the betweenness centrality, the proportion of the number of times each node appears in the shortest path of all node pairs is counted, reflecting the key degree of the node in network information transmission; then, the average values and variances of the three types of indexes are combined into a 10-dimensional feature vector, providing structured feature input for noise mode clustering;

[0016] The noise pattern clustering modeling unit adopts a density clustering algorithm to perform unsupervised classification on the 10-dimensional feature vectors, divides them into same-class feature vectors and different-class feature vectors, and aggregates the same-class feature vectors and distinguishes the different-class feature vectors; then, for various noise patterns generated by clustering, the range interval of the topological features is established, and each type of topological feature range is correspondingly associated with the corresponding marine environment type and stored in a data structure to form a noise feature library, thereby providing a benchmark for real-time noise separation.

[0017] The real-time noise separation and feature extraction module includes a topological mapping noise separation unit and a fault feature vector extraction unit.

[0018] The topological mapping noise separation unit maps the real-time vibration data output by the multi-dimensional sensor array deployment module to a five-dimensional phase space based on the Takens embedding theorem and using the embedding dimension and time delay parameters pre-trained by the noise feature topology library management module, and then constructs a real-time topological network using the Delaunay triangulation algorithm; then, by calculating the Hamming distance between the network and each mode feature vector in the noise feature library, when the distance exceeds a preset threshold, it is determined as noise data, and noise separation is realized through a topological edge weight screening mechanism, and the denoised effective vibration data is output to the fault feature vector extraction unit; wherein, the Hamming distance refers to the standardized count of the difference in corresponding dimensional topological indicators; the calculation logic is as follows: let the feature vector of the real-time topological network be X = [x1, x2,,, x n ], and the feature vector of a mode in the noise library be Y = [y1, y2,,, y n ], where n is the feature dimension; the Hamming distance calculation formula is as follows:

[0019] ;

[0020] wherein, δ(x i , y i ) is an indicator function, which takes 1 when |x i − y i | > ε i , and 0 otherwise; ε i is the difference threshold of the i-th feature;

[0021] The Hamming distance reflects the quantitative comparison of topological features through standardized dimensional difference counting, and provides a mathematical basis for noise separation;

[0022] The fault feature vector extraction unit extracts the denoised vibration data output by the topology mapping noise separation unit, reconstructs a five-dimensional phase space topology network based on a three-dimensional CAD model of the equipment and a structural dynamics finite element analysis result, and then identifies a high-weight path through the following steps: first, according to the node stress distribution obtained by the finite element analysis, a weight is given to each connection edge, the weight value = node stress amplitude / global stress average, the energy transfer efficiency of each path is calculated based on the vibration energy flow theory, the efficiency = path energy flux / total energy flux, and an energy proportion threshold is set to screen out connected subgraphs and high-weight paths;

[0023] Then, six types of geometric parameters are quantified by using a morphological algorithm: path curvature, node betweenness, connected subgraph density, topology network diameter, clustering coefficient, and edge weight variance. The path curvature is obtained by calculating the root mean square deviation of the path node coordinates from an ideal straight line. The node betweenness is obtained by counting the frequency of the node in all shortest paths. The connected subgraph density is obtained by calculating the ratio of the actual number of edges to the number of completely connected edges. The topology network diameter is the length of the longest shortest path in the network. The clustering coefficient is obtained by calculating the ratio of the number of connecting edges between neighbors to the maximum possible number of connecting edges. The edge weight variance is obtained by calculating the standard deviation of all connecting edge weights.

[0024] Meanwhile, four types of dynamic parameters are quantified: vibration amplitude distribution entropy, characteristic frequency component proportion, time-frequency domain cross-correlation coefficient, and energy distribution gradient.

[0025] The vibration amplitude distribution entropy is calculated based on the amplitude probability density function of the denoised vibration signal. The characteristic frequency component proportion is determined by the normalized spectral energy value in the sub-hertz frequency band. The time-frequency domain cross-correlation coefficient refers to the correlation index of the time domain signal and the frequency domain feature. The energy distribution gradient is the vibration energy attenuation rate along the high-weight path.

[0026] Finally, the above 10 types of parameters are standardized and combined into a feature vector, which is complementary to the noise features output by the noise feature topology library management module, and is directly input into the fault rule chain deduction module to provide quantitative basis for fault type determination by combining structural geometric features and dynamic characteristics.

[0027] The fault rule chain deduction module includes a rule library construction unit and a multi-layer reasoning execution unit.

[0028] The rule library construction optimization unit extracts more than 1000 groups of fault data verified by actual maintenance from the historical fault case library, establishes a ternary mapping relationship of feature vector-fault type-environment condition through structured processing, decomposes the fault feature vector into geometric features and dynamic features based on the principles of structural mechanics and fault mechanism analysis, and constructs a three-layer rule chain combined with marine environmental parameters.

[0029] Feature screening layer: set the abnormal threshold of each dimension feature;

[0030] Environment filtering layer: establish the environmental parameter-feature validity association table;

[0031] Logic operation layer: combine feature conditions using and / or / not logic gates;

[0032] Among them, the feature validity refers to the quantitative correlation between each dimension feature value in the fault feature vector and the actual fault type of the equipment under the condition of a specific marine environmental parameter;

[0033] The multi-layer reasoning execution unit inputs the 10-dimensional feature vector output by the fault feature vector extraction unit and the real-time environmental parameters collected by the device sensor into three layers of rules for reasoning: first, the feature screening layer compares each dimension parameter in the feature vector with the threshold condition pre-stored in the rule base, and marks the abnormal feature dimensions that exceed the preset threshold;

[0034] Then, according to the current marine environmental parameters, the environmental parameter-feature threshold correction association rule is called through the environment filtering layer. This rule is based on the mapping relationship between environmental parameters and feature thresholds in historical fault cases, and adjusts the threshold condition of the feature screening layer. Specifically, according to the real-time environmental parameter interval, the corresponding feature threshold correction coefficient is retrieved from the association rule base, and then the preset threshold of the feature screening layer is operated with the correction coefficient according to the formula default threshold x correction coefficient to generate a corrected threshold that adapts to the real-time environment;

[0035] Finally, through the logic operation layer, the marked abnormal features are logically combined with the threshold condition corrected by the environment using the forward reasoning algorithm, and the logic expression of the feature combination-fault type in the rule base is matched to deduce the fault type, occurrence location and severity level;

[0036] During the reasoning process, the noise separation results of the noise feature topology library are called in real time to exclude environmental noise interference.

[0037] The multi-dimensional fault judgment module includes a comprehensive reasoning verification unit and a confidence evaluation unit;

[0038] The comprehensive reasoning verification unit carries out multi-source fusion verification on the preliminary fault result output by the fault rule chain deduction module, the denoising data of the real-time noise separation module, and the equipment operation parameters and environmental parameters; specifically: first, calculate the dimensional difference between the fault feature vector and the standard feature in the rule base through the Euclidean distance; then call the running parameter-fault sensitivity association rule in the rule base to verify whether the current speed and load of the equipment trigger the fault sensitive condition; finally, use the sensor number-three-dimensional coordinate-structure node mapping table established by the space-time consistency synchronous acquisition unit to calculate the spatial distance between the fault position coordinates and the sensor collection points through the Euclidean distance formula, and when the error is less than the preset threshold, determine that the spatial correlation is effective; through the three-layer verification mechanism, eliminate the misjudgment caused by environmental interference, and output the final fault judgment result fused with the structure feature, operation state and spatial position;

[0039] The confidence assessment unit quantifies the fault judgment result output by the comprehensive reasoning verification unit in multiple dimensions:

[0040] Rule matching degree calculation: based on the number of rules matched in the logical operation layer and the rule weight given by the rule base construction unit through more than 1000 historical cases, generate a basic confidence coefficient;

[0041] Historical similarity search: query the number of cases with the same feature combination in the historical fault case library, and generate a similarity score according to the consistency of the case repair result;

[0042] Environmental robustness evaluation: based on the mapping relationship between environmental parameters and feature effectiveness in historical cases, call the environmental parameter-confidence correction association rule in the rule base, and correct the confidence according to the current environmental parameters; when the comprehensive confidence is less than 80%, trigger the secondary reasoning mechanism: backtrack the fault feature vector extraction process, check the noise separation result based on the noise feature topology library, call the Hamming distance to calculate the noise matching degree, and verify the effectiveness of the 10-dimensional feature vector, forming a closed-loop optimization of the judgment result.

[0043] The intelligent operation and maintenance feedback module includes a structured report generation unit and a real-time terminal pushing unit;

[0044] The structured report generation unit generates a three-dimensional visual label of the fault position based on the fault type, location, severity and confidence result output by the multi-dimensional fault judgment module, calls the three-dimensional CAD model of the equipment to generate a three-dimensional visual label of the fault position, and distinguishes the fault severity by color coding; combined with the equipment operation parameters and marine environmental parameters to calculate the maintenance priority coefficient; at the same time, retrieve the cases with the same feature combination in the historical fault case library, extract the solutions, maintenance time and spare parts list, generate a structured repair report containing three-dimensional label diagram, maintenance priority ranking and historical case reference, and the report format adopts XML structure conforming to the marine engineering standard;

[0045] The real-time terminal pushing unit establishes an encrypted transmission channel by using the Beidou satellite communication module, and pushes the structured maintenance report to the operation and maintenance personnel terminal in real time, wherein the pushing content includes a fault early warning pop-up window, a maintenance report attachment and a real-time communication interface; the system automatically records the pushing time, the receiving state and the operation and maintenance personnel confirmation information, and forms a closed-loop operation and maintenance record chain.

[0046] A marine engineering data management method based on machine learning comprises the following steps:

[0047] S1, deploying sensors at stress concentration parts of equipment, synchronously collecting sub-micron level vibration data and generating a multi-dimensional data set containing time, space and amplitude dimensions through space-time consistency processing;

[0048] S2, extracting marine environmental noise data from a historical fault case library, clustering to generate a noise feature library through phase space reconstruction and topological feature calculation, and providing a comparison benchmark for real-time noise separation;

[0049] S3, mapping real-time vibration data to phase space to construct a topological network, separating noise by comparison with the noise feature library, and extracting a fault feature vector containing geometric and dynamic characteristics based on an equipment structure model;

[0050] S4, constructing a rule chain based on historical fault cases, inputting the extracted feature vector and real-time environmental parameters into the rule chain to deduce, and deriving a fault type, position and severity through multi-layer logic matching;

[0051] S5, further verifying the deduced result through multi-source fusion, calculating a rule matching degree, a historical similarity and an environmental robustness confidence, and triggering secondary reasoning optimization to determine the result when the confidence is lower than a threshold;

[0052] S6, generating a structured report containing three-dimensional labeling, maintenance priority and historical cases according to the final fault determination, and pushing the report to an operation and maintenance terminal through Beidou communication encryption to form a closed-loop operation and maintenance record.

[0053] Compared with the prior art, the present application has the following advantages:

[0054] 1. Improved noise separation accuracy: the phase space reconstruction technology is used to construct a noise feature library (containing 1000 groups of historical noise data), and the Hamming distance matching algorithm is used to realize that the residual noise energy accounts for less than 5% after noise separation, thereby significantly improving the reliability of fault feature recognition.

[0055] 2. Breakthrough in three-dimensional spatial positioning accuracy: based on a three-dimensional CAD model (accuracy 0.1mm) of equipment and a sensor array mapping table (positioning accuracy 0.05mm), the fault position error is less than or equal to 0.1mm through Euclidean distance calculation, thereby meeting the needs of precise maintenance of deep-sea equipment.

[0056] 3. The false positive rate is significantly reduced: the confidence evaluation unit reduces the comprehensive false positive rate from 30% in the traditional method to less than 8% through three-dimensional verification of rule matching degree (weight coefficient), historical similarity (matching cases >= 30 groups) and environmental robustness (correction coefficient 0.9). BRIEF DESCRIPTION OF DRAWINGS

[0057] Figure 1 The system workflow diagram of the marine engineering data management system based on machine learning. DETAILED DESCRIPTION

[0058] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0059] Embodiment: As shown in the figure, the present application provides a technical solution, Figure 1

[0060] A marine engineering data management system based on machine learning, comprising a multi-dimensional sensing array deployment module, a noise feature topology library management module, a real-time noise separation and feature extraction module, a fault rule chain deduction module, a multi-dimensional fault judgment module and an intelligent operation and maintenance feedback module; the multi-dimensional sensing array deployment module deploys 9 MEMS micro-vibration sensors at the blade surface, blade root, blade middle, blade tip, bearing inner and outer ring contact points, main shaft section, shaft coupling and bearing support for synchronous collection of sub-micron level vibration signals; the noise feature topology library management module constructs a feature library containing all marine environmental noise modes, and each mode stores its topology feature parameters; the real-time noise separation and feature extraction module maps the real-time collected vibration data to a phase space to construct a topology network, separates the noise by calculating the Hamming distance between the vibration data and the modes in the noise library, and extracts a fault feature vector; the fault rule chain deduction module establishes a vibration feature-fault type logical rule chain based on a historical fault case library, including a feature screening, an environmental parameter filtering and a three-layer reasoning mechanism of logical operation; the multi-dimensional fault judgment module inputs the extracted feature vector into the rule chain for deduction, matches through multi-level logical expressions, and outputs the fault type, position and severity; and the intelligent operation and maintenance feedback module generates a structured maintenance report according to the fault judgment result, including three-dimensional labeling of the fault position, maintenance priority sorting, historical similar case solution reference, and real-time pushing to the operation and maintenance personnel terminal.

[0061] The multi-dimensional sensing array deployment module comprises a sub-micron level sensor deployment unit and a time-space consistency synchronous collection unit. ​

[0062] The sub-micron sensor deployment unit selects a MEMS capacitive acceleration sensor with a range of ±50g and a resolution of 0.1μg. According to the device structure dynamics, three-dimensional coordinate positioning technology is used to deploy the sensor at nine stress concentration points, including the blade surface, blade root, blade middle, blade tip, bearing inner and outer ring contact points, main shaft section, coupling and bearing support. Real-time sub-micron vibration signals are collected.

[0063] The space-time consistency synchronous acquisition unit uses an internal laser tracker to collect the three-dimensional coordinates of the physical installation positions of the nine sensors and the structural nodes of the sensor installation positions, and establishes a mapping table of sensor number-three-dimensional coordinates-structural nodes. For the vibration data collected by the sensor, the Kalman filter algorithm is used to correct the error of each channel timestamp, and the multi-channel data is aligned to a unified time axis based on the Beidou satellite clock. Then, according to the mapping table, the vibration data at each time point is associated with the three-dimensional coordinates of the corresponding sensor, and a structured data record containing timestamp, three-dimensional coordinates and vibration amplitude is generated. All records are arranged in chronological order to form a multi-dimensional vibration data set stored in HDF5 format, and each data block contains time dimension, space dimension and vibration amplitude dimension, which is convenient for real-time noise separation and feature extraction module to call.

[0064] The noise feature topology library management module includes a historical data import unit, a phase space topology feature reconstruction unit and a noise mode clustering modeling unit.

[0065] The historical data import unit extracts all historical vibration data of marine environmental noise from the device full life cycle fault case library. The data source needs to be labeled with power plant name, collection time and environmental parameters. Then, through the ETL tool, the unstructured data is converted into a standardized format: each record contains time series, environmental parameters and noise type label.

[0066] The phase space topology feature reconstruction unit performs five-dimensional phase space reconstruction on the historically imported noise time domain data based on the Takens embedding theorem. Specifically, taking the noise time domain sequence x(t) as input, a five-dimensional phase space vector is generated according to the formula X(t) = [x(t), x(t+T), x(t+2T), x(t+3T), x(t+4T)], wherein the embedding dimension m is determined by the false neighbor method, and the time delay T is determined by the first minimum value of the mutual information function; after reconstruction, the Delaunay triangulation algorithm is used to topologically connect the phase space point set, and through the empty circle property, the circumscribed circle of any triangle does not contain other nodes, forming a topology network reflecting the data distribution density; finally, three types of topology indexes of the topology network are calculated in real time, which are node connectivity, shortest path length and betweenness centrality; for node connectivity, the number of connections of each node with adjacent nodes is counted, and the average number of connections of all network nodes is taken as the node connectivity; for the shortest path length, the shortest path distance between all node pairs is calculated, and the global shortest path length is taken as the average value of the whole network; for the betweenness centrality, the proportion of the number of times each node appears in the shortest path of all node pairs is counted, reflecting the key degree of the node in the network information transmission; then the average and variance of the three types of indexes are combined into a 10-dimensional feature vector, providing structured feature input for noise pattern clustering;

[0067] The noise pattern clustering modeling unit uses a density clustering algorithm to unsupervisedly classify the 10-dimensional feature vector into same-class feature vectors and different-class feature vectors, and aggregates the same-class feature vectors and distinguishes the different-class feature vectors; then, for various noise patterns generated by clustering, the range intervals of the topology features are established, and the topology feature ranges of various classes are associated with corresponding marine environment types one by one and stored in a data structure, forming a noise feature library to provide a comparison benchmark for real-time noise separation.

[0068] The real-time noise separation and feature extraction module includes a topology mapping noise separation unit and a fault feature vector extraction unit.

[0069] The topology mapping noise separation unit maps the real-time vibration data output by the multi-dimensional sensor array deployment module to a five-dimensional phase space based on the Takens embedding theorem and using the embedding dimension and time delay parameters pre-trained by the noise feature topology library management module, and then constructs a real-time topology network using the Delaunay triangulation algorithm; then, the Hamming distance between the network and each mode feature vector in the noise feature library is calculated, and when the distance exceeds a preset threshold, the data is determined to be noise data, and noise separation is realized through a topology edge weight screening mechanism, and the effective vibration data after noise removal is output to the fault feature vector extraction unit; wherein the Hamming distance refers to the standardized count of the difference in the corresponding dimension topology index; the calculation logic is as follows: let the feature vector of the real-time topology network be X = [x1, x2,,, x n ], and the feature vector of a mode in the noise library be Y = [y1, y2,,, y n ], wherein n is the feature dimension; the Hamming distance calculation formula is as follows:

[0070] ;

[0071] Wherein, δ(x i , y i ) is an indicator function, which is 1 when |x i −y i |>ε i , and 0 otherwise; ε i is the difference threshold of the i-th feature;

[0072] The Hamming distance reflects the quantitative comparison of topology features through standardized dimension difference counting, and provides a mathematical basis for noise separation;

[0073] The fault feature vector extraction unit reconstructs the five-dimensional phase space topology network based on the three-dimensional CAD model of the device and the structural dynamics finite element analysis results for the denoised vibration data output by the topology mapping noise separation unit; then, the high-weight path is identified through the following steps: first, each connection edge is given a weight according to the node stress distribution obtained from the finite element analysis, and the weight value = node stress amplitude / global stress average; based on the vibration energy flow theory, the energy transfer efficiency of each path is calculated, efficiency = path energy flux / total energy flux; set an energy proportion threshold to screen out connected subgraphs and high-weight paths;

[0074] The morphological algorithm is further used to quantize 6 types of geometric parameters: path curvature, node betweenness, connected subgraph density, topological network diameter, clustering coefficient and edge weight variance. The path curvature is obtained by calculating the root mean square deviation of the path node coordinates from an ideal straight line. The node betweenness is obtained by counting the frequency of the node in all shortest paths. The connected subgraph density is obtained by calculating the ratio of the actual number of edges of the subgraph to the number of completely connected edges. The topological network diameter is the length of the longest shortest path in the network. The clustering coefficient is obtained by calculating the ratio of the number of connected edges between the neighbors of the node to the maximum possible number of connected edges. The edge weight variance is obtained by calculating the standard deviation of all connected edge weights.

[0075] Meanwhile, 4 types of dynamic parameters are quantified: vibration amplitude distribution entropy, characteristic frequency component proportion, time-frequency domain cross-correlation coefficient and energy distribution gradient.

[0076] The vibration amplitude distribution entropy is calculated based on the amplitude probability density function of the denoised vibration signal. The characteristic frequency component proportion is determined by the normalized value of the frequency spectrum energy in the sub-hertz frequency band. The time-frequency domain cross-correlation coefficient refers to the correlation index of the time domain signal and the frequency domain feature. The energy distribution gradient is the vibration energy attenuation rate along the high weight path.

[0077] Finally, the above 10 types of parameters are standardized and combined into a feature vector, which is complementary to the noise feature output by the noise feature topology library management module, and is directly input into the fault rule chain deduction module to provide quantitative basis for fault type judgment by fusing structural geometric features and dynamic characteristics.

[0078] The fault rule chain deduction module includes a rule library construction unit and a multi-layer reasoning execution unit.

[0079] The rule library construction optimization unit extracts more than 1000 groups of fault data verified by actual maintenance from the historical fault case library, and establishes a ternary mapping relationship of feature vector-fault type-environment condition through structured processing. Based on the principles of structural mechanics and fault mechanism analysis, the fault feature vector is decomposed into geometric features and dynamic features, and combined with marine environmental parameters to construct a three-layer rule chain.

[0080] Feature screening layer: set the abnormal threshold of each dimension feature;

[0081] Environment filtering layer: establish an environment parameter-feature effectiveness association table;

[0082] Logic operation layer: combine feature conditions using AND / OR / NOT logic gates;

[0083] Among them, the feature effectiveness refers to the quantitative correlation degree between each dimension feature value in the fault feature vector and the actual fault type of the device under specific marine environmental parameter conditions.

[0084] The multi-layer reasoning execution unit inputs the 10-dimensional feature vector output by the fault feature vector extraction unit and the real-time environmental parameters collected by the device sensor into a three-layer rule chain for reasoning: first, the characteristic vector in each dimension parameter is compared with the threshold condition pre-stored in the rule library through the feature screening layer, and the abnormal feature dimension exceeding the preset threshold is marked;

[0085] Then, according to the current marine environmental parameters, the environmental parameter-feature threshold correction association rule is called through the environmental filtering layer. The rule is based on the mapping relationship between the environmental parameters and the feature threshold in the historical fault cases, and adjusts the threshold condition of the feature screening layer. Specifically, according to the real-time environmental parameter interval, the corresponding feature threshold correction coefficient is retrieved from the association rule library, and then the preset threshold of the feature screening layer is operated with the correction coefficient according to the formula default threshold x correction coefficient to generate a corrected threshold suitable for the real-time environment;

[0086] Finally, through the logic operation layer, the marked abnormal features are logically combined with the threshold conditions corrected by the environment, and the logic expression of the feature combination-fault type in the rule library is matched to deduce the fault type, the occurrence position and the severity level;

[0087] During the reasoning process, the noise separation result of the noise feature topology library is called in real time to exclude environmental noise interference.

[0088] The multi-dimensional fault judgment module includes a comprehensive reasoning verification unit and a confidence assessment unit.

[0089] The comprehensive reasoning verification unit performs multi-source fusion verification on the preliminary fault result output by the fault rule chain reasoning module and the denoising data of the real-time noise separation module, the device operating parameters and the environmental parameters. Specifically: first, calculate the dimensional difference between the fault feature vector and the standard feature in the rule library through the Euclidean distance; then, call the running parameter-fault sensitivity association rule in the rule library to verify whether the current speed and load of the device trigger the fault sensitivity condition; finally, use the sensor number-three-dimensional coordinate-structure node mapping table established by the space-time consistency synchronous acquisition unit to calculate the spatial distance between the fault position coordinates and the sensor collection points through the Euclidean distance formula, and when the error is less than the preset threshold, determine that the spatial correlation is effective; through the three-layer verification mechanism, eliminate the misjudgment caused by environmental interference, and output the final fault judgment result of the fused structure features, operating state and spatial position;

[0090] The confidence assessment unit quantifies the fault judgment result output by the comprehensive reasoning verification unit in multiple dimensions:

[0091] Rule matching degree calculation: based on the number of rules matched in the logical operation layer and the rule weight given by the rule base construction unit through more than 1000 historical cases, the basic confidence coefficient is generated;

[0092] Historical similarity search: query the number of cases with the same feature combination in the historical fault case library, and generate a similarity score according to the consistency of the case repair result;

[0093] Environmental robustness evaluation: based on the mapping relationship between environmental parameters and feature effectiveness in historical cases, call the environmental parameter-confidence correction association rule in the rule base, and modify the confidence according to the current environmental parameters; when the comprehensive confidence is less than 80%, trigger the secondary reasoning mechanism: backtrack the fault feature vector extraction process, check the noise separation result based on the noise feature topology library, call the hamming distance to calculate the noise matching degree, and verify the effectiveness of the 10-dimensional feature vector, forming a closed loop optimization of the judgment result.

[0094] The intelligent operation and maintenance feedback module includes a structured report generation unit and a real-time terminal pushing unit;

[0095] The structured report generation unit generates a three-dimensional visual label of the fault location based on the fault type, location, severity and confidence result output by the multi-dimensional fault judgment module, calls the device three-dimensional CAD model, and distinguishes the fault severity by color coding; combined with the device operating parameters and marine environmental parameters to calculate the maintenance priority coefficient; at the same time, retrieve the cases with the same feature combination in the historical fault case library, extract the solution, maintenance time and spare parts list, generate a structured repair report containing three-dimensional label chart, maintenance priority ranking and historical case reference, and the report format adopts XML structure conforming to marine engineering standards;

[0096] The real-time terminal pushing unit establishes an encrypted transmission channel using the Beidou satellite communication module, and pushes the structured repair report to the operation and maintenance personnel terminal in real time, and the pushing content includes fault warning pop-up window, repair report attachment and real-time communication interface; the system automatically records the pushing time, receiving state and operation and maintenance personnel confirmation information, and forms a closed loop operation and maintenance record chain.

[0097] A marine engineering data management method based on machine learning, comprising the following steps:

[0098] S1, deploy sensors at parts of the device prone to stress concentration, synchronously collect sub-micron level vibration data, and generate a multi-dimensional data set containing time, space and amplitude dimensions through space-time consistency processing;

[0099] S2, extract marine environmental noise data from the historical fault case library, reconstruct the phase space and calculate the topological features, cluster to generate a noise feature library, and provide a comparison benchmark for real-time noise separation;

[0100] S3, mapping real-time vibration data to phase space to construct a topological network, separating noise by comparing with a noise feature library, and extracting a fault feature vector containing geometric and dynamic characteristics based on a device structure model;

[0101] S4, constructing a rule chain based on historical fault cases, inputting the extracted feature vector and real-time environmental parameters into the rule chain to deduce the fault type, location and severity through multi-layer logic matching;

[0102] S5, performing multi-source fusion verification on the deduced results, calculating the rule matching degree, historical similarity and environmental robustness confidence, and triggering secondary reasoning optimization to determine the result when the threshold is below;

[0103] S6, generating a structured report containing three-dimensional labeling, maintenance priority and historical cases according to the final fault determination, and pushing it to the operation and maintenance terminal through Beidou communication encryption to form a closed-loop operation and maintenance record.

[0104] On a 3MW wind turbine on a certain offshore wind power platform in the South China Sea, according to the results of equipment structure dynamics analysis, MEMS capacitive acceleration sensors with a range of ±50g and a resolution of 0.1μg are deployed at 9 stress concentration positions on the blade surface (1 / 3 away from the blade root), blade root, blade middle, blade tip, bearing inner and outer ring contact points, main shaft section (500mm away from the bearing), coupling and bearing support. A laser tracker (accuracy 0.05mm) is used to collect the three-dimensional coordinates of each sensor, and a mapping table is established, such as: sensor S1 corresponds to blade root coordinates (X=12.5m, Y=3.2m, Z=8.7m), and the structure node is the blade-hub connection tenon.

[0105] The sensors are synchronized by a space-time consistency synchronization unit to collect vibration data at a sampling rate of 1024Hz. For example, at a certain time t=10:23:45.678, the vibration amplitude collected by sensor S1 is 12.5μg, the time stamp error is corrected by Kalman filtering (the corrected time stamp and Beidou clock deviation ≤1ms), and the three-dimensional coordinates are associated to generate a structured record: [t=10:23:45.678, X=12.5m, Y=3.2m, Z=8.7m, amplitude=12.5μg]. All data are stored in HDF5 format to form a three-dimensional data set containing time (10000 time points), space (9 sensor coordinates) and amplitude (-50g~50g) for subsequent module calls.

[0106] 1000 sets of marine environmental noise data are extracted from the platform full life cycle case library, such as typhoon period (sea current speed 3.5 m / s), tidal high tide (wave height 2.8 m) and other scene vibration data. Take a set of typhoon period noise time domain sequence x(t) as an example, based on Takens embedding theorem, five-dimensional phase space reconstruction is carried out:

[0107] The embedding dimension m=5 (the false neighbor rate is less than 5% calculated by the false neighbor method), the time delay T=10 sampling points (at the first minimum value of mutual information function), and the phase space vector X(t)=[x(t),x(t+10),x(t+20),x(t+30),x(t+40)] is generated.

[0108] Delaunay triangulation is performed on 1000 phase space points to construct a topological network. The topological indicators are calculated: the average node connection degree is 4.2 (the average connection of each node is 4.2 adjacent nodes), the average global shortest path length is 2.8 (the average distance of all node pairs), and the average betweenness centrality is 0.15 (the average proportion of node appearance in the shortest path). The average and variance of the three indicators are combined into a 10-dimensional feature vector, such as: [4.2,0.5,2.8,0.3,0.15,0.08,...] (the first three dimensions are the average and variance of node connection degree, the middle three dimensions are the average and variance of shortest path length, and the last three dimensions are the average and variance of betweenness centrality).

[0109] The 10-dimensional feature vector is classified by using density clustering algorithm, for example, the typhoon period noise is classified into one class, and the node connection degree range is [3.8,4.6] and the shortest path length range is [2.5,3.1]. The "topological feature range-marine environment type" correlation table is established, such as node connection degree>4.0 and shortest path length<3.0 corresponding to typhoon noise mode, forming a noise feature library stored in a distributed database.

[0110] On July 1, 2025, at 14:00, the system collected real-time vibration data, and based on the noise feature library pre-trained embedding dimension m=5 and time delay T=10, it was mapped to a five-dimensional phase space and constructed a real-time topological network. The Hamming distance between this network and the typhoon noise mode in the noise library is calculated:

[0111] The real-time feature vector X=[4.3,0.6,2.7,0.25,0.16,...], the noise library feature vector Y=[4.0,0.5,2.8,0.3,0.15,...], and the threshold value εi of each dimension is set to 0.2 (such as node connection degree threshold 0.2).

[0112] Calculate δ(xi, yi): when |4.3-4.0|=0.3>0.2, δ=1; |0.6-0.5|=0.1≤0.2, δ=0, and so on. Hamming distance DH=(1+0+1+0+0+...) / 10=0.3 (assuming 3 dimensions are above threshold), exceeds the preset threshold 0.25, and is determined as typhoon noise. Separate the noise by topological edge weight screening (keep edges with weight>0.7), and output the denoised data.

[0113] For the denoised data, the topological network is reconstructed based on the three-dimensional CAD model of the device (mesh accuracy 0.1mm). According to the finite element analysis, the stress amplitude of a certain path node is 150MPa, the global stress average is 100MPa, and the edge weight = 150 / 100=1.5; the path energy flux is 2.3W, the total energy flux is 3.0W, and the energy transfer efficiency = 2.3 / 3.0≈76.7%, which exceeds the 70% threshold, and is determined as a high weight path. Quantify the geometric parameters:

[0114] Path tortuosity: calculate the root mean square deviation of the path node coordinates from the ideal straight line, assuming that the sum of the squared deviations of the 10 node coordinates is 0.01mm², and the root mean square deviation is 0.1mm;

[0115] Node betweenness: count the number of times this node participates in 30 shortest paths, betweenness = 30 / 100=0.3;

[0116] Combine the kinetic parameters such as characteristic frequency component proportion (sub-hertz band 0.3Hz energy proportion 25%), generate 10-dimensional fault feature vector: [0.1, 0.3, 0.65, 5.2, 0.4, 0.2, 0.85, 25%, 0.9, 0.15] (the first six dimensions are geometric parameters, and the last four dimensions are kinetic parameters).

[0117] Input the feature vector into the fault rule chain deduction module:

[0118] Feature selection layer: preset path tortuosity threshold 0.1mm, node betweenness threshold 0.25, current value 0.1mm (not exceeding threshold), 0.3 (exceeding threshold), mark node betweenness as abnormal dimension.

[0119] Environment filtering layer: real-time sea current speed 2.2m / s, call associated rule "when sea current speed>2m / s, node betweenness threshold increases by 20%", modify threshold to 0.25x1.2=0.3, current value 0.3 does not exceed modified threshold, cancel abnormal mark.

[0120] Logical operation layer: combined with other features (such as 0.3Hz energy ratio 25%>20%), match the logical expression "(node betweenness>0.25 ∧ 0.3Hz energy ratio>20%) → blade tenon loose", deduce the fault type as blade tenon loose, location as blade root, severity as medium.

[0121] Multi-dimensional fault judgment module verification:

[0122] Calculate the Euclidean distance between the feature vector and the standard feature vector: assuming the standard vector is [0.1, 0.25, 0.6, 5.0, 0.4, 0.2, 0.8, 20%, 0.9, 0.1], the Euclidean distance =√[(0.1-0.1)²+(0.3-0.25)²+...]=0.07, less than the threshold 0.1, feature consistency is valid;

[0123] Spatial correlation: the distance between the fault location blade root coordinates and the sensor S1 coordinates is 0.08m<0.1mm, and the judgment is valid;

[0124] Confidence evaluation: rule matching degree 0.8 (matching 2 key rules), historical similarity 0.9 (35 similar cases), environmental robustness 0.85 (sea current speed correction coefficient 0.85), comprehensive confidence = 0.8×0.3+0.9×0.4+0.85×0.3=0.86>0.8, no need for secondary reasoning.

[0125] According to the judgment result, the structured report generation unit calls the three-dimensional CAD model, generates a red three-dimensional label at the blade root position (accuracy 0.1mm), calculates the maintenance priority coefficient: severity medium (coefficient 2) × environmental influence factor (sea current speed 2.2m / s corresponds to factor 1.5)=3, priority level 3 (1-5 levels). Retrieve the historical case library, extract the average maintenance time of 35 groups of blade tenon loose cases as 4 hours, spare parts list (screw x2, torque wrench), generate XML format report;

[0126] The real-time terminal push unit converts the report into JSON format and pushes it to the operation and maintenance personnel tablet through the Beidou satellite communication module, with a warning pop-up window prompt "blade root tenon loose, suggest repairing within 4 hours". After receiving, the operation and maintenance personnel confirm the processing, the system records the push time 14:15, the receiving state is read, forms the operation and maintenance record chain, completes the whole process closed loop from data collection to maintenance decision.

[0127] It will be apparent to those skilled in the art that the application is not limited to the details of the above-exemplified embodiments and that the present application can be implemented in other particular forms without departing from the spirit or essential characteristics of the present application. The embodiments should therefore be considered in all respects as illustrative and not restrictive, the scope of the application being indicated by the appended claims rather than by the above description, and all changes which come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein. No reference signs in the claims should be considered as limiting the scope of the claims with respect to the figures of the patent document.

Claims

1. A machine learning based offshore engineering data management system, characterized by: The intelligent operation and maintenance system comprises a multi-dimensional sensing array deployment module, a noise feature topology library management module, a real-time noise separation and feature extraction module, a fault rule chain deduction module, a multi-dimensional fault judgment module and an intelligent operation and maintenance feedback module; the multi-dimensional sensing array deployment module is arranged on the surface of a blade, a blade root, a blade middle, a blade tip, a bearing inner and outer ring contact point, a main shaft section, a shaft coupling and a bearing support, and nine MEMS micro-vibration sensors are arranged to synchronously collect sub-micron level vibration signals; the noise feature topology library management module constructs a feature library containing all ocean environmental noise modes, and each mode stores its topology feature parameters; the real-time noise separation and feature extraction module maps the real-time collected vibration data to a phase space to construct a topology network, separates the noise by calculating the Hamming distance between the vibration data and the noise library, and extracts a fault feature vector; the fault rule chain deduction module is based on a historical fault case library, and establishes a vibration feature-fault type logical rule chain, which comprises a feature screening, an environmental parameter filtering and a three-layer reasoning mechanism; the multi-dimensional fault judgment module inputs the extracted feature vector into the rule chain for deduction, matches through a multi-layer logical expression, and outputs a fault type, a position and a severity; and the intelligent operation and maintenance feedback module generates a structured maintenance report according to the fault judgment result, which comprises a fault position three-dimensional marking, a maintenance priority sorting, a historical similar case solution reference, and is pushed to an operation and maintenance personnel terminal in real time.

2. The machine learning based offshore engineering data management system of claim 1, wherein: The multi-dimensional sensing array deployment module comprises a sub-micron level sensor deployment unit and a space-time consistency synchronous acquisition unit. The sub-micron level sensor deployment unit selects a MEMS capacitive acceleration sensor with a range of ±50g and a resolution of 0.1μg, adopts three-dimensional coordinate positioning technology according to the equipment structure dynamics, and arranges the sensor at nine stress concentration positions, i.e. the surface of a blade, a blade root, a blade middle, a blade tip, a bearing inner and outer ring contact point, a main shaft section, a shaft coupling and a bearing support, to collect sub-micron level vibration signals in real time. The space-time consistency synchronous acquisition unit uses an equipment internal laser tracker to collect three-dimensional coordinates of physical installation positions of the nine sensors and structural nodes of the sensor installation positions, and establishes a mapping table of sensor number-three-dimensional coordinate-structural node. For the vibration data synchronously collected by the sensors, the Kalman filtering algorithm is used to correct the error of the time stamps of each channel, and the multi-channel data is aligned to a unified time axis with the Beidou satellite clock as a reference; then, the vibration data at each time point is associated with the three-dimensional coordinates of the corresponding sensor according to the mapping table, and a structured data record containing a time stamp, three-dimensional coordinates and a vibration amplitude is generated, all records are arranged in chronological order, a multi-dimensional vibration data set is formed in an HDF5 format, and each data block contains a time dimension, a space dimension and a vibration amplitude dimension, so as to facilitate the calling of the real-time noise separation and feature extraction module.

3. The machine learning based offshore engineering data management system of claim 1, wherein: The noise feature topology library management module comprises a historical data import unit, a phase space topology feature reconstruction unit and a noise mode clustering modeling unit. The historical data import unit extracts all historical vibration data of marine environmental noise from a device full life cycle fault case library, and the data source needs to be marked with the power plant name, collection time and environmental parameters, and then the unstructured data is converted into a standardized format by an ETL tool: each record contains a time series, environmental parameters and noise type label; The phase space topology feature reconstruction unit reconstructs the five-dimensional phase space based on the Takens embedding theorem on the historical imported noise time domain data. Specifically, taking the noise time domain sequence x(t) as the input, generating a five-dimensional phase space vector according to the formula X(t) = [x(t), x(t+T), x(t+2T), x(t+3T), x(t+4T)], wherein the embedding dimension m is determined by the false neighbor point method, and the time delay T is determined by the first minimum value of the mutual information function; after reconstruction, the Delaunay triangulation algorithm is used to topologically connect the phase space point set, and through the empty circle property, no other nodes are contained in the incircle of any triangle, forming a topological network reflecting the data distribution density; finally, three types of topological indexes of the topological network are calculated in real time, which are node connectivity, shortest path length and betweenness centrality; for node connectivity, the number of connections of each node with adjacent nodes is counted, and the average number of connections of all network nodes is taken as the node connectivity; for the shortest path length, the shortest path distance between all node pairs is calculated, and the global shortest path length is taken as the average value of the whole network; for the betweenness centrality, the proportion of the number of times each node appears in the shortest path of all node pairs is counted, reflecting the key degree of the node in the network information transmission; then the average and variance of the three types of indexes are combined into a 10-dimensional feature vector, providing a structured feature input for noise pattern clustering; The noise pattern clustering modeling unit uses a density clustering algorithm to perform unsupervised classification on the 10-dimensional feature vector, divides it into same-class feature vectors and different-class feature vectors, and aggregates the same-class feature vectors and distinguishes the different-class feature vectors; then, for various noise patterns generated by clustering, the range interval of the topological features is established, and each type of topological feature range is associated with the corresponding marine environment type one by one and stored in a data structure, forming a noise feature library to provide a benchmark for real-time noise separation.

4. The machine learning based offshore engineering data management system of claim 1, wherein: The real-time noise separation and feature extraction module includes a topological mapping noise separation unit and a fault feature vector extraction unit; The topology mapping noise separation unit maps the real-time vibration data output by the multi-dimensional sensing array deployment module to a five-dimensional phase space based on the Takens embedding theorem and using the embedding dimension and time delay parameters pre-trained by the noise feature topology library management module, and then constructs a real-time topology network using the Delaunay triangulation algorithm; then, by calculating the Hamming distance between the network and each mode feature vector in the noise feature library, when the distance exceeds a preset threshold, it is determined as noise data, and noise separation is realized through a topology edge weight screening mechanism, and the effective vibration data after denoising is output to the fault feature vector extraction unit; wherein the Hamming distance refers to the standardized count of the difference in the corresponding dimension topology index; the calculation logic is as follows: let the feature vector of the real-time topology network be X = [x1, x2,,, x n ], and the feature vector of a mode in the noise library be Y = [y1, y2,,, y n ], wherein n is the feature dimension; the Hamming distance calculation formula is as follows: ； where δ(x i , y i ) is an indicator function that takes the value 1 when |x i − y i | > ε i , and 0 otherwise; and ε i is the difference threshold for the i-th feature. The Hamming distance reflects the quantitative comparison of topological features through the standardized dimension difference count, and provides a mathematical basis for noise separation; The fault feature vector extraction unit reconstructs a five-dimensional phase space topology network based on the three-dimensional CAD model of the equipment and the structural dynamics finite element analysis results of the de-noised vibration data output by the topology mapping noise separation unit; then it identifies high-weight paths through the following steps: first, according to the node stress distribution obtained by finite element analysis, each connection edge is given a weight, the weight value = node stress amplitude / global stress average, the energy transfer efficiency of each path is calculated based on the vibration energy flow theory, efficiency = path energy flux / total energy flux, and an energy proportion threshold is set to screen out connected subgraphs and high-weight paths; Then, six types of geometric parameters are quantified using morphological algorithms: path curvature, node betweenness, connected subgraph density, topology network diameter, clustering coefficient, and edge weight variance. The path curvature is obtained by calculating the root mean square deviation of the path node coordinates from the ideal straight line; the node betweenness is obtained by counting the frequency of the node in all shortest paths; the connected subgraph density is obtained by calculating the ratio of the actual number of edges to the number of fully connected edges; the topology network diameter is the length of the longest shortest path in the network; the clustering coefficient is obtained by calculating the ratio of the number of connecting edges between node neighbors to the maximum possible number of connecting edges; and the edge weight variance is obtained by calculating the standard deviation of all connecting edge weights. Meanwhile, four types of dynamic parameters are quantified: vibration amplitude distribution entropy, characteristic frequency component proportion, time-frequency domain cross-correlation coefficient, and energy distribution gradient. The vibration amplitude distribution entropy is calculated based on the amplitude probability density function of the de-noised vibration signal; the characteristic frequency component proportion is determined by the normalized spectral energy value in the sub-hertz frequency band; the time-frequency domain cross-correlation coefficient refers to the correlation index of the time domain signal and the frequency domain feature; and the energy distribution gradient is the vibration energy attenuation rate along the high-weight path. Finally, the above 10 types of parameters are standardized and combined into a feature vector, which is complementary to the noise feature output by the noise feature topology library management module and directly input into the fault rule chain deduction module, providing quantitative basis for fault type determination by combining structural geometric features and dynamic characteristics.

5. The machine learning based offshore engineering data management system of claim 1, wherein: The fault rule chain deduction module includes a rule library construction unit and a multi-layer reasoning execution unit. The rule library construction optimization unit extracts more than 1000 groups of fault data verified by actual maintenance from the historical fault case library, establishes a ternary mapping relationship of feature vector-fault type-environment condition through structured processing; based on the principles of structural mechanics and fault mechanism analysis, the fault feature vector is decomposed into geometric features and dynamic features, and combined with marine environmental parameters to construct a three-layer rule chain: Feature selection layer: set abnormal threshold for each dimension feature; Environment filtering layer: establish an environment parameter-feature effectiveness association table; Logic operation layer: combine feature conditions using AND / OR / NOT logic gates; Wherein, feature effectiveness refers to the quantitative correlation degree between each dimension feature value in the fault feature vector and the actual fault type of the equipment under specific marine environmental parameter conditions; The multi-layer reasoning execution unit inputs the 10-dimensional feature vector output by the fault feature vector extraction unit and the real-time environmental parameters collected by the device sensor into a three-layer rule chain for reasoning: first, the feature screening layer compares each dimension parameter in the feature vector with the pre-stored threshold condition in the rule base, and marks the abnormal feature dimension that exceeds the preset threshold; Then, the environmental filtering layer calls the environmental parameter-feature threshold correction association rule based on the current marine environmental parameters, which is established based on the mapping relationship between environmental parameters and feature threshold values in historical fault cases, adjusts the threshold condition of the feature screening layer, specifically, according to the real-time environmental parameter interval, retrieves the corresponding feature threshold correction coefficient from the association rule base, and then calculates the preset threshold of the feature screening layer and the correction coefficient according to the formula default threshold × correction coefficient to generate a corrected threshold that adapts to the real-time environment; Finally, the logic operation layer uses a forward reasoning algorithm to logically combine the marked abnormal features and the threshold condition corrected by the environment, matches the logic expression of the feature combination-fault type in the rule base, and deduces the fault type, occurrence location and severity level; During the reasoning process, the noise separation result of the noise feature topology library is called in real time to exclude environmental noise interference.

6. The machine learning based offshore engineering data management system of claim 1, wherein: The multi-dimensional fault judgment module includes a comprehensive reasoning verification unit and a confidence assessment unit; The comprehensive reasoning verification unit performs multi-source fusion verification on the preliminary fault result output by the fault rule chain reasoning module and the denoising data of the real-time noise separation module, the device operating parameters and the environmental parameters; Specifically: first, calculate the dimensional difference between the fault feature vector and the standard feature in the rule base by Euclidean distance; Then, call the running parameter-fault sensitivity association rule in the rule base to verify whether the current speed and load of the device trigger the fault sensitivity condition; Finally, use the sensor number-three-dimensional coordinate-structure node mapping table established by the spatiotemporal consistency synchronous acquisition unit to calculate the spatial distance between the fault location coordinates and the sensor collection points by the Euclidean distance formula, and when the error is less than the preset threshold, determine that the spatial correlation is valid; Through the three-layer verification mechanism, false positives caused by environmental interference are eliminated, and the final fault judgment result of the fused structure feature, operating state and spatial position is output; The confidence assessment unit quantifies the fault judgment result output by the comprehensive reasoning verification unit in multiple dimensions: Rule matching degree calculation: based on the number of rules matched by the logic operation layer and the rule weight given by the rule base construction unit through more than 1000 historical cases, a basic confidence coefficient is generated; Historical similarity retrieval: query the number of cases with the same feature combination in the historical fault case library, and generate a similarity score according to the consistency of the case repair results; Environmental robustness evaluation: Based on the mapping relationship between environmental parameters and feature effectiveness in historical cases, the environmental parameter-confidence correction association rule in the rule library is called to correct the confidence according to the current environmental parameters. When the comprehensive confidence is less than 80%, the secondary reasoning mechanism is triggered: backtracking the fault feature vector extraction process, checking the noise separation result based on the noise feature topology library, calling the Hamming distance to calculate the noise matching degree, and verifying the effectiveness of the 10-dimensional feature vector to form a closed-loop optimization of the judgment result.

7. The machine learning based offshore engineering data management system of claim 1, wherein: The intelligent operation and maintenance feedback module includes a structured report generation unit and a real-time terminal pushing unit; The structured report generation unit generates a three-dimensional visualization label of the fault location based on the fault type, location, severity and confidence results output by the multi-dimensional fault judgment module, calls the device three-dimensional CAD model, and distinguishes the fault severity by color coding; combined with the device operating parameters and marine environmental parameters to calculate the maintenance priority coefficient; at the same time, retrieve the cases with the same feature combination in the historical fault case library, extract the solution, maintenance time and spare parts list, generate a structured maintenance report containing three-dimensional label chart, maintenance priority ranking and historical case reference, and the report format adopts XML structure conforming to marine engineering standards; The real-time terminal pushing unit uses the Beidou satellite communication module to establish an encrypted transmission channel and push the structured maintenance report to the operation and maintenance personnel terminal in real time, and the push content includes fault warning pop-up window, maintenance report attachments and real-time communication interface; the system automatically records the push time, reception status and operation and maintenance personnel confirmation information to form a closed-loop operation and maintenance record chain.

8. A machine learning based offshore engineering data management method applied to the machine learning based offshore engineering data management system of any one of claims 1-7, characterized in that: The method comprises the following steps: S1, deploying sensors at stress concentration parts of the equipment, synchronously collecting sub-micron level vibration data, and generating a multi-dimensional data set containing time, space and amplitude dimensions through space-time consistency processing; S2, extracting marine environmental noise data from the historical fault case library, reconstructing the phase space and calculating the topological features, clustering to generate a noise feature library, and providing a comparison benchmark for real-time noise separation; S3, mapping real-time vibration data to phase space to construct a topological network, separating noise by comparing with the noise feature library, and extracting a fault feature vector containing geometric and dynamic characteristics based on the equipment structure model; S4, constructing a rule chain based on historical fault cases, inputting the extracted feature vector and real-time environmental parameters into the rule chain, and deducing the fault type, location and severity through multi-layer logic matching; S5, further verifying the deduced results through multi-source fusion, calculating the rule matching degree, historical similarity and environmental robustness confidence, and triggering secondary reasoning optimization when the threshold is lower than the threshold; S6, generating a structured report containing three-dimensional labels, maintenance priority and historical cases based on the final fault judgment, and pushing the report to the operation and maintenance terminal through Beidou communication encryption to form a closed-loop operation and maintenance record.

Images