Offshore unmanned platform vitality online identification method and system

Abstract

The invention discloses a maritime unmanned platform vitality online identification method and system. The method comprises the following steps: 1) equipment-level vitality evaluation; characteristic parameters of equipment performance changes are collected, and a correlation model with equipment vitality is established for equipment vitality evaluation; 2) obtaining a device-level vitality evaluation result of each device of the system according to the method in the step 1), and comprehensively considering the current vitality of each device according to a fusion method for empowering the importance of the system by each device to obtain a system-level vitality evaluation result; and 3) considering a vitality association relationship between the equipment vitality and each system, establishing an equipment-level vitality evaluation model for expressing a complex association relationship by using a Bayesian network, and performing equipment-level vitality evaluation based on the Bayesian network. According to the method, the operation data is acquired through the sensor arranged on the unmanned platform, so that unmanned platform vitality evaluation is realized, and subjectivity is reduced.

Description

Technical Field

[0001] This invention relates to unmanned system evaluation technology, and more particularly to an online method and system for identifying the survivability of unmanned maritime platforms. Background Technology

[0002] When unmanned maritime platforms malfunction or degrade, it can lead to reduced or lost system functionality, affecting the execution of the entire mission and even causing catastrophic economic losses and safety accidents. Therefore, establishing survivability assessment criteria for unmanned maritime platforms and conducting research on key technologies for online survivability identification can provide strong support for the autonomous assessment of availability and survivability of unmanned maritime platforms in real-time throughout their normal service life and under adversarial conditions, as well as for autonomous intelligent task allocation and mission feasibility decisions at both the single-equipment and cross-domain cluster levels.

[0003] Traditional vitality assessment methods require human intervention and heavily rely on expert knowledge and experience, which conflicts with the requirement to reduce human involvement in unmanned platforms. However, unmanned maritime platforms are equipped with numerous sensors, providing massive amounts of operational data. Effectively utilizing this vast amount of operational data from unmanned equipment at sea is key to achieving vitality assessment for these platforms. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a method and system for online identification of the vitality of unmanned marine platforms, addressing the deficiencies in the existing technology.

[0005] The technical solution adopted by this invention to solve its technical problem is: an online survivability identification method for unmanned marine platforms, comprising the following steps:

[0006] 1) Equipment-level lifespan assessment;

[0007] Collect characteristic parameters of equipment performance changes and establish a correlation model between these parameters and equipment lifespan to assess equipment lifespan; specifically as follows:

[0008] 1.1) Acquire monitoring signals from the equipment, including vibration signals, voltage signals, and current signals;

[0009] 1.2) Data preprocessing;

[0010] 1.3) Extract features from the data in the monitoring signals and use the evaluation fusion results of each signal to evaluate the equipment's viability;

[0011] 2) Obtain the device-level lifespan assessment results of each device in the system using the method in step 1), and use a fusion method that assigns weights to each device based on its importance to the system to comprehensively consider the current lifespan of each device and obtain the system-level lifespan assessment results.

[0012] 3) Considering the relationship between equipment vitality and the vitality of various systems, a Bayesian network is used to establish an equipment-level vitality assessment model that expresses complex relationships, and equipment-level vitality assessment is carried out based on the Bayesian network.

[0013] According to the above scheme, in step 1.2), data preprocessing includes data cleaning, which improves data quality by filling missing values, smoothing noisy data, and removing outliers.

[0014] According to the above scheme, in step 1.3), a time-domain feature extraction method is selected to extract features from the vibration signal data in the monitoring signal; the following time-domain features are extracted to form its feature vector, namely: root mean square, mean, skewness factor, impulse value, kurtosis factor, and waveform factor;

[0015] Using combined feature vectors as training samples, multiple Gaussian Mixture Models (GMMs) are constructed for vitality evaluation.

[0016] According to the above scheme, in step 1.3), the voltage time-frequency and current time-frequency features are extracted by a preset neural network to complete the extraction of voltage signal and current signal data features in the monitoring signal and to perform a vitality assessment.

[0017] According to the above scheme, in step 2), each device is weighted according to its importance to the system. The importance weight of the device is a combination of the expert weight and the objective weight of the device. A coefficient of variation is introduced to maximize the weighted sum of the expert weight and the objective weight of the device, so as to obtain a comprehensive weighting of each indicator based on the coefficient of variation.

[0018] According to the above scheme, step 3) is specifically as follows:

[0019] 3.1) Considering the characteristics of multiple system components and the interrelationships between different systems, Bayesian networks are used to describe the polymorphic and uncertain logical relationships between system nodes, construct a complex relationship model of each system, and form an equipment-level life assessment model.

[0020] The process is as follows:

[0021] First, the network structure is constructed, and the variables and connections of each node are determined through expert experience and data learning. Then, the edge probability distribution and conditional probability distribution of each node are given according to the system type. Finally, the equipment-level life assessment model is obtained by reasoning using known information.

[0022] 3.2) Extend the Bayesian network by introducing a ring-based Bayesian network, determine the topology and conditional probability distribution of all nodes in the network, and conduct equipment-level life assessment based on the Bayesian network.

[0023] According to the above scheme, in step 3.2), each node of the evaluation model is defined as a four-state state, that is, the state space is {A,B,C,D}, which are: A, completely lost vitality; B, basically lost vitality; C, partially lost vitality; and D, basically not lost vitality.

[0024] An online vitality identification system for unmanned maritime platforms includes:

[0025] The equipment-level lifespan assessment module is used to collect characteristic parameters of equipment performance changes and establish a correlation model between these parameters and equipment lifespan for equipment lifespan assessment; it includes:

[0026] The data acquisition submodule is used to acquire monitoring signals from the device, including vibration signals, voltage signals, and current signals.

[0027] The data preprocessing submodule is used to improve data quality by methods such as filling missing values, smoothing noisy data, and removing outliers.

[0028] The equipment lifespan assessment submodule is used for extracting monitoring signal features and assessing equipment lifespan; details are as follows:

[0029] The vibration signal data features in the monitoring signal were extracted using a time-domain feature extraction method. The following time-domain features were extracted to form its feature vector: root mean square, mean, skewness factor, impulse value, kurtosis factor, and waveform factor.

[0030] Using combined feature vectors as training samples, multiple Gaussian Mixture Models (GMMs) are constructed for vitality evaluation.

[0031] By using a pre-set neural network to extract features from voltage and current time-frequency signals, the data features of voltage and current signals in the monitoring signals are extracted, and a vitality assessment is performed.

[0032] The fusion results are used to evaluate the equipment's lifespan data, and the monitoring data is quantitatively evaluated using a lifespan index between 0 and 1.

[0033] The system-level lifespan assessment module is used to assign weights to each device according to its importance to the system based on the device-level lifespan assessment results obtained from the device-level lifespan assessment module, and to comprehensively consider the current lifespan of each device to obtain the system-level lifespan assessment result.

[0034] The equipment-level lifespan assessment module is used to assess equipment lifespan by establishing an equipment-level lifespan assessment model that expresses complex relationships based on the lifespan correlation between equipment lifespan and the lifespan correlation between various systems using Bayesian networks.

[0035] Specifically as follows:

[0036] 1) In view of the characteristics of multiple system components and the interrelationship between different systems, Bayesian networks are used to describe the polymorphic and uncertain logical relationships between system nodes, construct a complex relationship model of each system, and form an equipment-level life assessment model.

[0037] The process is as follows:

[0038] First, the network structure is constructed, and the variables and connections of each node are determined through expert experience and data learning. Then, the edge probability distribution and conditional probability distribution of each node are given according to the system type. Finally, the equipment-level life assessment model is obtained by reasoning using known information.

[0039] 2) Extend the Bayesian network by introducing a ring-based Bayesian network, determine the topology and conditional probability distribution of all nodes in the network, and conduct equipment-level lifespan assessment based on the Bayesian network.

[0040] In the evaluation model, each node is defined as a four-state state, i.e., the state space is {A,B,C,D}, where A represents complete loss of vitality, B represents basic loss of vitality, C represents partial loss of vitality, and D represents basic non-loss of vitality.

[0041] The beneficial effects of this invention are:

[0042] 1) The method of the present invention uses sensors equipped on the unmanned platform to collect operational data to realize the life assessment of the unmanned platform, thereby reducing subjectivity.

[0043] 2) Establishing a lifespan evaluation system for unmanned platforms and conducting research on key technologies for online identification of the lifespan of unmanned platforms can provide strong support for the autonomous assessment of the availability and lifespan of unmanned platforms in real time during their normal service life and under combat conditions, as well as for autonomous and intelligent task allocation and task executability decision-making at the single equipment level and cross-domain cluster level. Attached Figure Description

[0044] The present invention will be further described below with reference to the accompanying drawings and embodiments. In the accompanying drawings:

[0045] Figure 1 This is a schematic diagram of the online vitality identification process of an unmanned platform according to an embodiment of the present invention;

[0046] Figure 2 This is a schematic diagram of the device-level lifespan assessment process according to an embodiment of the present invention;

[0047] Figure 3 This is a bearing life index curve diagram according to an embodiment of the present invention;

[0048] Figure 4 This is a lifespan curve diagram of the power system according to an embodiment of the present invention;

[0049] Figure 5 This is a schematic diagram of the equipment-level lifespan assessment model according to an embodiment of the present invention;

[0050] Figure 6 This is a schematic diagram of the equipment survivability assessment results under four missions according to an embodiment of the present invention. Detailed Implementation

[0051] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0052] like Figure 1 As shown, an online survivability assessment method for unmanned maritime platforms is proposed. Unmanned platform survivability refers to the ability to survive while fulfilling operational missions, including the ability to complete missions and return safely. The method includes the following steps:

[0053] 1) Equipment-level lifespan assessment;

[0054] Collect characteristic parameters of equipment performance changes and establish a correlation model between these parameters and equipment lifespan to assess equipment lifespan; specifically as follows:

[0055] 1.1) Acquire monitoring signals from the equipment, including vibration signals, voltage signals, and current signals;

[0056] 1.2) Data preprocessing; Data preprocessing includes data cleaning, which improves data quality by filling missing values, smoothing noisy data, and removing outliers;

[0057] 1.3) Extract features from the monitoring signals to assess equipment lifespan;

[0058] A time-domain feature extraction method was selected to extract features from the vibration signal data in the monitoring signal; the following time-domain features were extracted to form its feature vector, namely: root mean square, mean, skewness factor, impulse value, kurtosis factor, and waveform factor;

[0059] Using combined feature vectors as training samples, multiple Gaussian Mixture Models (GMMs) are constructed for vitality evaluation.

[0060] By using a pre-set neural network to extract features from voltage and current time-frequency signals, the data features of voltage and current signals in the monitoring signals are extracted, and a vitality assessment is performed.

[0061] The lifespan of the equipment is assessed based on the evaluation and fusion results of each signal.

[0062] 2) Obtain the device-level lifespan assessment results of each device in the system using the method in step 1), and use a fusion method that assigns weights to each device based on its importance to the system to comprehensively consider the current lifespan of each device and obtain the system-level lifespan assessment results.

[0063] In this embodiment, each device is weighted according to its importance to the system. The importance weight of each device is a combination of expert weight and objective weight. Then, a coefficient of variation is introduced to maximize the weighted sum of the expert weight and the objective weight, thus obtaining a comprehensive weighting of the importance of each device based on the coefficient of variation.

[0064] 3) Considering the relationship between equipment vitality and the vitality of various systems, a Bayesian network is used to establish an equipment-level vitality assessment model that expresses complex relationships, and equipment-level vitality assessment is carried out based on the Bayesian network.

[0065] 3.1) Considering the characteristics of multiple system components and the interrelationships between different systems, Bayesian networks are used to describe the polymorphic and uncertain logical relationships between system nodes, construct a complex relationship model of each system, and form an equipment-level life assessment model.

[0066] The process is as follows:

[0067] First, the network structure is constructed, and the variables and connections of each node are determined through expert experience and data learning. Then, the edge probability distribution and conditional probability distribution of each node are given according to the system type. Finally, the equipment-level life assessment model is obtained by reasoning using known information.

[0068] 3.2) Extend the Bayesian network by introducing a ring-based Bayesian network, determine the topology and conditional probability distribution of all nodes in the network, and conduct equipment-level life assessment based on the Bayesian network.

[0069] In the evaluation model, each node is defined as a four-state state, i.e., the state space is {A,B,C,D}, where A represents complete loss of vitality, B represents basic loss of vitality, C represents partial loss of vitality, and D represents basic non-loss of vitality.

[0070] A specific example:

[0071] Taking a certain unmanned surface vessel as an example, it includes bearing equipment. The bearing, gearbox and pump together form the power system of the unmanned surface vessel. The power system and other systems together form the equipment of the unmanned surface vessel.

[0072] Taking the full lifecycle axis system dataset of a certain maritime unmanned equipment as an example, the specific process of this step is as follows:

[0073] 1. Equipment-level lifespan assessment;

[0074] 1.1) It is necessary to extract characteristic parameters that can accurately reflect changes in equipment performance from multiple monitoring parameters, and establish a correlation model between characteristic parameters and equipment vitality, so as to realize rapid estimation of equipment vitality based on different monitoring signals.

[0075] The shaft system platform includes an AC motor (2000 RPM), four bearings (Rexnord ZA-2115 double-row bearings), and vibration sensors (X-axis and Y-axis). The shaft and motor are connected via a belt conveyor, with the four bearings evenly mounted on the shaft. A 6000-pound radial load is applied to the shaft and bearings via a spring mechanism, maintaining the shaft system at a constant speed of 2000 rpm. All bearings are lubricated. Bearing vibration signals are acquired using PCB 353B33 high-sensitivity quartz ICP accelerometers mounted on the bearing housings. Bearings failed after exceeding their designed lifespan, and the entire operation represents a full life-cycle degradation process. The dataset includes three datasets under different operating conditions. Each dataset describes a test-to-failure experiment, containing vibration data from the four bearings. Each dataset consists of a separate file containing 1-second snapshots of the vibration signals recorded at specific time intervals. Each file contains 20480 points, with a sampling rate set to 20 kHz. The filename indicates the time of data collection. Each record (row) in the data file represents a data point. The selected bearing data is from Set No. 2. In this dataset, the bearing ran continuously for approximately 168 hours, with data collected every 10 minutes for 1 second each time, at a sampling rate of 20kHz. The experiment was stopped after the bearing failed; ultimately, the bearing exhibited outer ring damage.

[0076] 1.2) Data cleaning methods are used to improve data quality by filling in missing values, smoothing noisy data, and removing outliers;

[0077] Outlier removal requires setting parameters including window length and removal threshold. The window length is selected based on the test results to achieve better performance, and the removal threshold is selected as 3σ according to the 3σ principle. The final selected window length and removal threshold are 100 and 3, respectively.

[0078] The adjustable parameters for locally weighted regression include the length `frac`, which indicates how much data should be truncated for local processing each time; `frac` is a proportion of the original data volume. The number of iterations `it` indicates whether a second regression is needed after one local regression. Based on specific data analysis, `frac` and `it` were ultimately found to be 0.1 and 1, respectively.

[0079] 1.3) A time-domain feature extraction method is selected to extract the data features of the monitoring signal; the following time-domain features are extracted to form its feature space, namely: root mean square, mean, skewness factor, impulse value, kurtosis factor, waveform factor, and margin; and feature vectors are formed.

[0080] To handle large amounts of unlabeled equipment data, clustering methods, such as K-means clustering and self-organizing map networks, can automatically cluster degraded data. By comparing the status of health data, the relative condition of bearing operation can be determined. In reality, the sensing signals of equipment (such as vibration signals) exhibit multi-peak (or non-linear) distribution characteristics. The advantage of Gaussian mixture models (GMMs) is that they can approximate arbitrarily complex data distributions. This characteristic enables GMMs to identify the health status of equipment.

[0081] Multiple bearing performance model (GMM) were constructed using features from bearings under normal operating conditions (during the period following the start of recording) as training samples. Based on these GMM models, bearing lifespan data was evaluated, with lifespan indices ranging from 0 to 1 used to quantify the monitoring data. Adaptive GMM fitting was applied to the feature data to be evaluated, using multidimensional mean and covariance vectors to characterize the distribution characteristics of this batch of feature data. Simultaneously, Euclidean distance was used to measure the differences between the mean and covariance vectors of the normal GMM and those of the tested GMM. A Min-Max normalization algorithm was employed to transform the obtained difference distances into more physically meaningful lifespan indices. The evaluation results of the bearing feature data are visualized as follows: Figure 4 As shown.

[0082] 2. Based on the above equipment-level lifespan assessment results, and combining expert experience with a fusion method of subjective and objective weighting, the current lifespan of each piece of equipment is comprehensively considered to obtain the final system-level lifespan assessment results. Taking the power system as an example, the specific process of this step is as follows:

[0083] 2.1) Once the lifespan of each device under the system level is obtained, a weight is assigned to each device based on its importance to the system. Analyzing the data using analytical algorithms, combined with expert experience, greatly improves the efficiency of lifespan modeling and makes the evaluation results more accurate.

[0084] The subjective weighted values ​​for bearings, gearboxes, and pumps in the power system are 0.6, 0.3, and 0.1, respectively; the objective weighted values ​​are 0.421, 0.381, and 0.198, respectively.

[0085] 2.2) By comprehensively considering both subjective and objective weights, the final weight W of the integrated subjective and objective information is obtained. The coefficient of variation is introduced to maximize the weighted sum of subjective and objective factors, thereby solving for the values ​​of the undetermined coefficients. The coefficient of variation is the quotient of the standard deviation and the mean of a set of data, reflecting the degree of data dispersion. The subjective and objective weights W for each indicator based on the coefficient of variation are then obtained. Based on the final integrated subjective and objective weights and the equipment-level lifespan assessment model, the final system-level lifespan assessment result can be obtained.

[0086] The lifespan of a dynamic system over time is shown in the curve. Figure 3 As shown.

[0087] 3) Taking into account the lifespan of the equipment and the lifespan relationships between various systems, a Bayesian network is used to establish an equipment-level lifespan assessment model that expresses complex relationships. Taking unmanned maritime vessels as an example, the specific process of this step is as follows:

[0088] 3.1) In view of the characteristics of multiple system components and the interrelationship between different systems, Bayesian networks are used to describe the polymorphic and uncertain logical relationships between systems, construct complex relationship models of each system, and finally form an equipment-level life assessment model.

[0089] by Figure 5 For example, a tentatively defined unmanned maritime platform consists of systems such as an electric system, a propulsion system, a ship structure and equipment, a control system, and basic payloads. Its main missions are patrol, combat, transportation, and reconnaissance. First, a network structure is constructed, and the variables and connections of each node are determined using expert experience and data learning methods. Then, the marginal probability distribution and conditional probability distribution of each node are given according to the system type. Finally, an equipment-level lifespan assessment model is obtained by reasoning using known information.

[0090] 3.2) Equipment-level lifespan assessment algorithms are primarily based on Bayesian networks. The main tasks of Bayesian network modeling include determining the network topology and the conditional probability distribution of each node in the network. The conditional probability distribution of all nodes in the network is collectively referred to as the network's probability parameters. Bayesian network modeling includes a qualitative process: determining the topology; and a quantitative stage: determining the probability parameters.

[0091] While classic Bayesian networks possess strong modeling and analytical capabilities, their acyclicity assumption makes them difficult to apply to feedback control systems. Cyclic Bayesian networks, however, abandon this assumption and can effectively address the problem of events recurring within their respective accident scenarios. In vitality assessment, different systems interact, forming a loop; therefore, Bayesian networks must be extended, leading us to introduce cyclic Bayesian networks.

[0092] The inference tasks of Bayesian networks mainly include the marginal probability of a single variable, the joint probability distribution of a set of variables, the conditional probability of nodes, the most probable interpretation of the model, the maximum a posteriori probability, sensitivity analysis, and information value.

[0093] Each node in the evaluation model is defined as a four-state system, with a state space of {A, B, C, D}: A) completely lost survivability, B) basically lost survivability, C) partially lost survivability, and D) basically not lost survivability. Finally, the equipment survivability performance under four mission conditions is obtained as follows: Figure 6 As shown.

[0094] Based on the above method, the present invention also provides an online vitality identification system for unmanned marine platforms, comprising:

[0095] The equipment-level lifespan assessment module is used to collect characteristic parameters of equipment performance changes and establish a correlation model between these parameters and equipment lifespan for equipment lifespan assessment; it includes:

[0096] The data acquisition submodule is used to acquire monitoring signals from the device, including vibration signals, voltage signals, and current signals.

[0097] The data preprocessing submodule is used to improve data quality by methods such as filling missing values, smoothing noisy data, and removing outliers.

[0098] The equipment lifespan assessment submodule is used for extracting monitoring signal features and assessing equipment lifespan; details are as follows:

[0099] The vibration signal data features in the monitoring signal were extracted using a time-domain feature extraction method. The following time-domain features were extracted to form its feature vector: root mean square, mean, skewness factor, impulse value, kurtosis factor, and waveform factor.

[0100] Using combined feature vectors as training samples, multiple Gaussian Mixture Models (GMMs) are constructed for vitality evaluation.

[0101] By using a pre-set neural network to extract features from voltage and current time-frequency signals, the data features of voltage and current signals in the monitoring signals are extracted, and a vitality assessment is performed.

[0102] The fusion results are used to evaluate the equipment's lifespan data, and the monitoring data is quantitatively evaluated using a lifespan index between 0 and 1.

[0103] The system-level lifespan assessment module is used to assign weights to each device according to its importance to the system based on the device-level lifespan assessment results obtained from the device-level lifespan assessment module, and to comprehensively consider the current lifespan of each device to obtain the system-level lifespan assessment result.

[0104] The equipment-level lifespan assessment module is used to assess equipment lifespan by establishing an equipment-level lifespan assessment model that expresses complex relationships based on the lifespan correlation between equipment lifespan and the lifespan correlation between various systems using Bayesian networks.

[0105] Specifically as follows:

[0106] 1) In view of the characteristics of multiple system components and the interrelationship between different systems, Bayesian networks are used to describe the polymorphic and uncertain logical relationships between system nodes, construct a complex relationship model of each system, and form an equipment-level life assessment model.

[0107] The process is as follows:

[0108] First, the network structure is constructed, and the variables and connections of each node are determined through expert experience and data learning. Then, the edge probability distribution and conditional probability distribution of each node are given according to the system type. Finally, the equipment-level life assessment model is obtained by reasoning using known information.

[0109] 2) Extend the Bayesian network by introducing a ring-based Bayesian network, determine the topology and conditional probability distribution of all nodes in the network, and conduct equipment-level lifespan assessment based on the Bayesian network.

[0110] In the evaluation model, each node is defined as a four-state state, i.e., the state space is {A,B,C,D}, where A represents complete loss of vitality, B represents basic loss of vitality, C represents partial loss of vitality, and D represents basic non-loss of vitality.

[0111] It should be understood that those skilled in the art can make improvements or modifications based on the above description, and all such improvements and modifications should fall within the protection scope of the appended claims.

Claims

1. A method for online identification of the vitality of an unmanned offshore platform, characterized in that: The following steps are involved: 1) Equipment-level vitality assessment; Collect characteristic parameters of equipment performance changes and establish a correlation model with equipment vitality to evaluate equipment vitality; the details are as follows: 1.1) Collecting monitoring signals of equipment, wherein the monitoring signals include vibration signals; 1.2) Data preprocessing; 1.3) Extract features from the data in the monitoring signal and evaluate the vitality of the device based on the evaluation fusion results of each signal; 2) Obtain the device-level vitality evaluation results of each device of the system according to the method in step 1), and obtain the system-level vitality evaluation results by using a fusion method that weights each device according to its importance to the system and comprehensively considering the current vitality of each device; 3) Considering the relationship between equipment vitality and the vitality of each system, the Bayesian network is used to establish an equipment-level vitality evaluation model that expresses complex correlations, and equipment-level vitality evaluation is performed based on the Bayesian network.

2. The method for online identification of the vitality of an unmanned offshore platform according to claim 1, characterized in that: In the step 1.2), data preprocessing includes data cleaning, which improves data quality by filling missing values, smoothing noisy data, and removing outliers.

3. The method for online identification of the vitality of an unmanned offshore platform according to claim 1, characterized in that: In the step 1.3), a time domain feature extraction method is used to extract features from the vibration signal data in the monitoring signal; the following time domain features are extracted to form its feature vector: They are: root mean square, mean, skewness factor, pulse value, kurtosis factor, and waveform factor; The combined feature vectors are used as training samples to construct multiple Gaussian mixture models (GMMs) for vitality assessment.

4. The method for online identification of the vitality of an unmanned offshore platform according to claim 1, characterized in that: In the step 1), the monitoring signal of the equipment also includes a voltage signal and a current signal. The voltage time-frequency and current time-frequency are feature extracted by a preset neural network to complete the data feature extraction of the voltage signal and current signal in the monitoring signal, and perform vitality evaluation of the corresponding signal.

5. The method for online identification of the vitality of an unmanned offshore platform according to claim 1, characterized in that: In the step 2), each device is weighted according to its importance to the system. The importance weight of the device is a combination of the expert weight and the objective weight of the device. The coefficient of variation is introduced to maximize the weighted sum of the expert weight and the objective weight of the device, thereby obtaining a comprehensive weighting of each indicator based on the coefficient of variation.

6. The method for online identification of the vitality of an unmanned offshore platform according to claim 1, characterized in that: The specific steps of step 3) are as follows: 3.1) In view of the characteristics of multi-system composition and the interrelationship between different systems, the Bayesian network is used to describe the polymorphism and uncertainty logical relationship between the nodes of each system, and a complex correlation model of each system is constructed to form an equipment-level vitality assessment model; The process is as follows: First, the network structure is constructed, and the variables and connection relationships of each node are determined through expert experience and data learning methods. Then, the edge probability distribution and conditional probability distribution of each node are given according to the system type. Finally, the equipment-level vitality evaluation model is obtained by reasoning with known information. 3.2) Expand the Bayesian network and introduce a cyclic Bayesian network to determine the topological structure and the conditional probability distribution of all nodes in the network, and perform equipment-level vitality evaluation based on the Bayesian network.

7. The method for online identification of the vitality of an unmanned offshore platform according to claim 6, characterized in that: In the step 3.2), each node of the evaluation model is defined as four states, that is, the state space is {A, B, C, D}: A completely lost vitality, B basically lost vitality, C partially lost vitality, D basically no loss of vitality.

8. An online identification system for the vitality of an unmanned offshore platform, characterized in that: include: The equipment-level vitality assessment module is used to collect characteristic parameters of equipment performance changes and establish a correlation model with equipment vitality to evaluate equipment vitality; it includes: A data acquisition submodule, used to collect monitoring signals of the equipment, wherein the monitoring signals include vibration signals, voltage signals and current signals; The data preprocessing submodule is used to improve data quality by filling missing values, smoothing noisy data, and removing outliers, including data cleaning; The equipment vitality assessment submodule is used for equipment monitoring signal feature extraction and equipment vitality assessment; the details are as follows: The time domain feature extraction method is used to extract the vibration signal data features in the monitoring signal; the following time domain features are extracted to form its feature vector, namely: root mean square, mean, skewness factor, pulse value, kurtosis factor, and waveform factor; The combined feature vectors are used as training samples to construct multiple Gaussian mixture models (GMMs) for vitality assessment. The voltage time-frequency and current time-frequency are extracted by using a preset neural network, the voltage signal and current signal data feature extraction in the monitoring signal is completed, and vitality assessment is performed; Evaluate the vitality data of the equipment with the fusion results, and quantify the monitoring data with a vitality index between 0 and 1; The system-level vitality evaluation module is used to weight the device according to its importance to the system based on the device-level vitality evaluation results of each device obtained in the device-level vitality evaluation module, and comprehensively consider the current vitality of each device to obtain the system-level vitality evaluation result; The equipment-level vitality evaluation module is used to establish an equipment-level vitality evaluation model that expresses complex correlations based on the correlation between equipment vitality and the vitality of various systems, and to evaluate equipment vitality using a Bayesian network; The details are as follows: 1) In view of the characteristics of multi-system composition and the interrelationship between different systems, the Bayesian network is used to describe the polymorphism and uncertainty logical relationship between the nodes of each system, and a complex correlation model of each system is constructed to form an equipment-level vitality assessment model; The process is as follows: First, the network structure is constructed, and the variables and connection relationships of each node are determined through expert experience and data learning methods. Then, the edge probability distribution and conditional probability distribution of each node are given according to the system type. Finally, the equipment-level vitality evaluation model is obtained by reasoning with known information. 2) Expand the Bayesian network and introduce a cyclic Bayesian network to determine the topological structure and the conditional probability distribution of all nodes in the network, and perform equipment-level vitality evaluation based on the Bayesian network.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images