Fault assessment method and system based on ship maintenance

Abstract

The invention relates to the technical field of fault assessment, and discloses a fault assessment method and system based on ship maintenance, and the method comprises the steps: carrying out the fault assessment of nodes through a knowledge graph, and constructing the distribution of health risks according to a fault assessment result; through analysis of ship maintenance records, distribution of health risks between nodes is migrated, and a risk distribution model adapted to ships is obtained. According to the risk distribution model, actively generating a risk probability of a fault combination about a node space and a risk probability of a single node fault; and generating fault hypotheses according to the risk probability of the fault combination about the node space and the risk probability of the single node fault, evaluating the coverage rate of each fault hypothesis to the risk distribution model, and finally outputting a fault evaluation result. The problem that a traditional method is insufficient in adaptability to a single ship type and specific working conditions is effectively solved, and the individuation and generalization ability of a risk model is enhanced.

Description

Technical Field

[0001] The present invention relates to the technical field of fault assessment, and specifically to a fault assessment method and system based on ship maintenance. Background Art

[0002] With the continuous development of the global shipping industry, modern ships have been continuously improved in terms of function integration, structural complexity, and automation level; as large-scale engineering equipment, ships involve many subsystems such as power systems, propulsion systems, navigation, electro-electronics, hydraulics, pipelines, and outfitting. Each part operates under long-term high loads and complex marine environments and is extremely vulnerable to the influence of various physical and chemical factors, resulting in problems such as wear, corrosion, loosening, aging, and faults; equipment failures may not only cause sailing delays and increased operating costs but also pose serious potential hazards to personal safety and environmental pollution; At present, most ship maintenance management is mainly based on regular inspections, planned maintenance, and experience judgment; traditional maintenance modes rely on manual inspections, regular disassembly inspections, or simple threshold alarm methods; these means often have a lagging response and are difficult to detect complex faults caused by concealment, gradualness, or multi-point coupling, and cannot meet the high requirements of modern ships for safety and economy; in recent years, with the development of sensor technology, ship automation, and intelligent monitoring systems, the ability to collect ship operation data has been greatly improved, providing more basis for systematic health management and scientific decision-making; however, the effective integration of massive data, the accurate extraction of abnormal features, and the fault propagation and causal correlation analysis of multi-node complex systems still face huge challenges; how to make full use of new-generation information technology and data analysis tools to improve the scientific and intelligent levels of early ship fault identification and risk warning has become an important technical development direction in the fields of ship engineering and intelligent operation and maintenance. Summary of the Invention

[0003] In view of the above existing problems, the present invention is proposed.

[0004] To solve the above technical problems, the present invention provides the following technical solutions: A fault assessment method based on ship maintenance, including: Collect multi-modal data of each monitored part during the normal operation of the ship; Use a knowledge graph to perform fault assessment of nodes, and construct a distribution of health risks according to the results of the fault assessment; Through the analysis of ship maintenance records, migrate the distribution of health risks between nodes to obtain a risk distribution model adapted to the ship; According to the risk distribution model, actively generate the risk probability of a fault combination in the node space and the risk probability of a single-node fault; Generate fault hypotheses based on the risk probabilities of fault combinations in the node space and the risk probabilities of single-node faults, and evaluate the coverage rate of each fault hypothesis for the risk distribution model, and finally output the fault evaluation results.

[0005] As a preferred solution of the fault evaluation method based on ship maintenance according to the present invention, wherein: the monitoring part includes mechanical components and subsystems in the ship structure that can be independently identified and are suitable for installing sensors; The multi-modal data includes data for describing the operating characteristics of the monitoring part obtained according to the sensor types in each monitoring part.

[0006] As a preferred solution of the fault evaluation method based on ship maintenance according to the present invention, wherein: the knowledge graph includes taking each component in the ship as a node, based on actual connections, functional dependencies, and relationships between entity nodes extracted from the historical maintenance data of the samples; constructing one-way or two-way causal relationships; among them, the causal relationships between nodes are learned through Bayesian networks; If there are multiple nodes in the same monitoring part, generalize the multi-modal data of the nodes in the monitoring part so that each node in the monitoring part synchronizes data according to the data distribution law; The data distribution law includes, for each monitoring part, through the pre-trained mapping relationship and the measured values of the sensors, mapping each node in the monitoring part to the corresponding monitoring data value.

[0007] As a preferred solution of the fault evaluation method based on ship maintenance according to the present invention, wherein: the distribution of the health risks includes, according to the monitoring data of each node, taking the health status of each node as the target variable, and combining the causal relationships obtained from the Bayesian network in the knowledge graph; Output the probability distribution of each node for different fault states, and form the distribution D of the health risks.

[0008] As a preferred solution of the fault evaluation method based on ship maintenance according to the present invention, wherein: the migration of the distribution of health risks between nodes includes using each maintenance record and the maintenance records of the same type of ships to generate an adjusted distribution of the health risks; combining all the maintenance records of the current ship to comprehensively generate the result of the distribution migration of the health risks; The specific generation steps of the adjusted distribution are as follows: Step 1: Generate the distribution of the health risks of the same type of ships according to the maintenance records of the same type of ships ; Step 2: Generate the weights and fault distributions according to each maintenance record of the current ship; Denote the probability distribution of failures during the \(i\)-th maintenance of the current ship; the weight during the \(i\)-th maintenance of the current ship is defined as: , Denote a preset adjustment coefficient; Denote the time interval between the \(i\)-th maintenance and the previous maintenance; Denote a mapping function that, through a preset functional relationship, obtains a reference value from the corresponding value; the weight for the health risk distribution of ships of the same type during the \(i\)-th maintenance of the current ship is defined as: ; Step 3: Assume that the current ship has undergone \(I\) maintenances, and combine the records of the \(I\) maintenances: ; where Denote the combined probability distribution; \(G()\) denotes a normalization function; is always equal to ; The risk distribution model includes generating the attention weight of each node according to the adjusted distribution; performing weighted adjustment and normalization on the distribution \(D\) of the health risk according to the attention weight at each node to realize the migration of the health risk distribution.

[0009] As a preferred solution of the fault assessment method based on ship maintenance according to the present invention, wherein: the risk probability of a single-node fault includes calculating the probability for each node \(i\) in the risk distribution model: the fault probability of node \(i\) in the risk distribution model; obtaining a risk probability sequence of single-node faults; The risk probability of the fault combination includes calculating the probability of failure for each group of multiple nodes with an associated relationship by using the risk distribution model: using the average value of the fault probabilities of the nodes with an associated relationship; obtaining a risk probability sequence of the fault combination; The fault hypothesis includes randomly combining the risk probability sequence of single-node faults and the risk probability sequence of the fault combination; Evaluating the coverage rate of each combination for the risk distribution model; The coverage rate includes reassigning each node in the fault hypothesis, recalculating the value distribution of the node space according to the assignment result; calculating the similarity between the value distribution and the risk distribution model, and taking the similarity as the coverage rate of each fault hypothesis; If a node in the fault hypothesis belongs to the node fault in the fault combination, the node is assigned the risk probability of the corresponding fault combination; if a node in the fault hypothesis belongs to a single-node fault, it is directly assigned according to the risk probability of the corresponding single-node fault.

[0010] As a preferred solution of the fault assessment method based on ship maintenance according to the present invention, the fault hypotheses are sequentially output in descending order of coverage rate as the fault assessment result.

[0011] A fault assessment system based on ship maintenance adopting the method according to the present invention, wherein: An acquisition unit that acquires multimodal data of each monitoring part during the normal operation of the ship; A distribution unit that uses a knowledge graph to perform fault assessment of nodes and constructs a distribution of health risks according to the results of the fault assessment; A migration unit that analyzes the ship maintenance records to migrate the distribution of health risks between nodes to obtain a risk distribution model adapted to the ship; according to the risk distribution model, actively generates the risk probability of a fault combination regarding the node space and the risk probability of a single-node fault; An output unit that generates fault hypotheses according to the risk probability of the fault combination regarding the node space and the risk probability of a single-node fault, and evaluates the coverage rate of each fault hypothesis for the risk distribution model, and finally outputs the fault assessment result.

[0012] A computer device, comprising: a memory and a processor; the memory stores a computer program, wherein: when the processor executes the computer program, the steps of the method according to any one of the present invention are implemented.

[0013] A computer-readable storage medium, on which a computer program is stored, wherein: when the computer program is executed by a processor, the steps of the method according to any one of the present invention are implemented.

[0014] Advantages of the present invention: The fault assessment method based on ship maintenance provided by the present invention realizes accurate assessment of the health status of each key component of the ship and dynamic risk distribution modeling by integrating multimodal monitoring data, knowledge graph modeling, and Bayesian causal reasoning, and significantly improves the early identification ability of systematic faults. By combining maintenance history and empirical data of the same type of ship and adopting an adaptive migration and attention weighting mechanism, the problem of insufficient adaptability of traditional methods to a single ship type and specific working conditions is effectively overcome, and the personalization and generalization ability of the risk model are enhanced. The present invention can actively infer the probabilities of single-point and combined faults, introduce a probability distribution coverage rate evaluation mechanism, and systematically output highly representative fault hypotheses and assessment results, thereby providing a scientific basis for ship intelligent operation and maintenance, precise maintenance, and safety decision-making, and greatly improving the safety and economy of ship operation. Description of the Drawings

[0015] To more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the accompanying drawings required for the description of the embodiments. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other accompanying drawings can be obtained based on these drawings.

[0016] Figure 1 It is the overall flowchart of a fault assessment method based on ship maintenance provided for the first embodiment of the present invention. Specific embodiments

[0017] To make the above objects, features, and advantages of the present invention more obvious and understandable, the following will make a detailed description of the specific embodiments of the present invention in conjunction with the accompanying drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of them. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0018] Embodiment 1, referring to Figure 1 , which is an embodiment of the present invention, provides a fault assessment method based on ship maintenance, including: S1: During the normal operation of the ship, collect multimodal data of each monitored part.

[0019] The monitored parts include mechanical components and subsystems in the ship structure that can be independently identified and are suitable for installing sensors; including but not limited to: each main cylinder block, bearing, connecting rod, cylinder head, exhaust pipe section of the main engine (diesel engine / main propeller). Bearings, couplings, sealing chambers, tailshaft sleeves, etc. of the propulsion shafting. Rotating or power units such as gearboxes, speed reducers, pumps, fans, hydraulic cylinders, and oil circuit valve groups. Electrical components such as electrical control cabinets, generators, and motors. Key cabin sections (such as engine rooms, pump rooms, cooling water tanks) or special structural nodes. And local functional units of the above components (such as lubricating oil pump seats, cooling water pipe outlets, local pressure nodes, etc.).

[0020] Further, the multimodal data includes data for describing the operating characteristics of the monitored part obtained according to the sensor type in each monitored part.

[0021] By deploying multiple types of sensors in key mechanical components and subsystems of the ship structure, it is possible to continuously collect and obtain multimodal operation parameters including vibration, temperature, pressure, current, flow rate, sound, etc. These multimodal data can fully reflect the true operation status and dynamic change characteristics of each monitored part under different working conditions. In this way, it not only lays a solid data foundation for subsequent intelligent analysis links such as fault identification, health assessment, and risk modeling, but also significantly improves the system's perception ability of early hidden dangers, complex coupling faults, and atypical anomalies, helping to realize the transformation of ship operation and maintenance from experience-driven to data-driven, and from passive response to active prediction, ensuring the safety and efficiency of ship operation.

[0022] S2: Use the knowledge graph to perform fault assessment on the nodes, and construct the distribution of health risks according to the results of the fault assessment.

[0023] Furthermore, the knowledge graph includes taking each component in the ship as a node, based on actual connections, functional dependencies, and the relationships between entity nodes extracted from historical maintenance data of samples (samples collected from multiple ships); constructing one-way or two-way causal relationships; among which, the causal relationships between nodes are learned through Bayesian networks; in fact, this is essentially transfer learning of the relationships between nodes. Through learning from multiple ship samples, the obtained knowledge graph can reflect the association relationships between nodes in the general category of "ship". Then, through "transfer using the maintenance records of the current ship" in the following text, the unique characteristics of the current ship are transferred.

[0024] By abstracting each component of the ship as a node in the knowledge graph, and based on structural connections, functional dependencies, and large-sample historical maintenance data, systematically extracting and modeling the multi-level and multi-type causal relationships between nodes, it can comprehensively reflect the linkage and influence mechanism between various elements in the complex ship system. With the help of Bayesian networks for learning and expressing causal relationships, the knowledge graph is equipped with dynamic reasoning and uncertainty modeling capabilities, and through transfer learning of multi-ship type sample data, the general laws of the industry-level "ship system" are summarized and refined. On this basis, combined with the maintenance history of the current ship itself, transfer adaptability for individual differences is realized, enhancing the accuracy and interpretability of the model in different ships and different operating scenarios.

[0025] If there are multiple nodes in the same monitoring section, generalize the multi-modal data of the nodes in the monitoring section so that each node in the monitoring section synchronizes data according to the data distribution law. The data distribution law includes, for each monitoring section, through the pre-trained mapping relationship and the measured values of the sensors, each node in the monitoring section is mapped to the corresponding monitoring data value. Generalize and synchronize the data distribution of multiple nodes within the monitoring section to ensure the consistency of feature expression and the standardization of model input, providing reliable data support and knowledge guarantee for subsequent health risk inference and fault identification.

[0026] Based on the monitoring data of each node, with the health status of each node as the target variable, combine the causal relationship obtained from the Bayesian network in the knowledge graph. Output the probability distribution of each node under different fault states to form the distribution D of health risks.

[0027] It should be noted that the Bayesian network is a directed acyclic graph structure that can effectively represent and calculate the conditional dependence relationships between variables in the system. In the context of ship health management, the Bayesian network can use each monitoring node as a random variable node in the network, and embed the structural, functional, and maintenance history dependence relationships between components into the network structure in the form of edges. By inputting the node monitoring data, the Bayesian network can use conditional probability inference technology to calculate the posterior probability distribution of each node under different health / fault states. This inference method can not only process multi-source, incomplete, and uncertain data, but also reflect the complex causal linkage and abnormal propagation mechanisms between nodes. By combining the knowledge graph with the Bayesian network, it not only ensures the global expression of the ship system structure and operation and maintenance knowledge, but also improves the ability to depict the common laws and ontology semantics of the industry in the inference process. In this way, the health risk distribution D output by the model not only has a strict probability meaning, but also has good physical interpretability and industry adaptability.

[0028] S3: Through the analysis of the ship maintenance records, migrate the distribution of health risks between nodes to obtain a risk distribution model adapted to the ship.

[0029] Among them, migrating the distribution of health risks between nodes includes using each maintenance record and the maintenance records of the same type of ships to generate an adjusted distribution of health risks; combining all the maintenance records of the current ship to comprehensively generate the result of the distribution migration of health risks.

[0030] The adjusted distribution, by introducing the health risk distribution obtained from the maintenance data of the same type of ships as the industry baseline or group knowledge, can provide a strong empirical reference for the risk modeling of the current ship. At the same time, by combining each maintenance record of the current ship and assigning dynamic weights to each maintenance interval (frequency), it can reflect the unique fault evolution and maintenance response rules of the target ship during actual operation. The specific generation steps are as follows: Step 1: Generate the distribution of the health risks of the same type of ships based on the maintenance records of the same type of ships .

[0031] Step 2: Generate the weights and fault distributions according to each maintenance record of the current ship; Let represent the probability distribution of faults during the i-th maintenance of the current ship; the weight during the i-th maintenance of the current ship is defined as: where represents a preset adjustment coefficient; represents the time interval between the i-th maintenance and the previous maintenance (this is actually the frequency factor, which represents the concentration of the current maintenance in time in the entire maintenance history); represents a mapping function that obtains a reference value through a preset functional relationship (which can be non-linear or linear, but this result value must be negatively correlated with the input value) for the corresponding value of .

[0032] Step 3: Assume that the current ship has undergone I maintenances, and combine the I maintenance records: ; where represents the combined probability distribution; G() represents the normalization function; is always equal to .

[0033] The risk distribution model includes generating the attention weights of each node according to the adjusted distribution; performing weighted adjustment and normalization of the probability distribution of the health risk distribution D according to the attention weights at each node to achieve the migration of the health risk distribution (for the personalized configuration of the ship's subtle features, that is, this step is to perform learning migration on the current faults through the maintenance records of the same type and the current ship).

[0034] ; where represents the attention weight of each node. Represents a mapping relationship, which is obtained without using a preset linear transformation in this embodiment. In other implementable embodiments, it can be output through a pre-trained neural network. Represents the risk distribution model.

[0035] First, by introducing the maintenance data of the same type of ships, an industry baseline of the health risk distribution is formed, enabling the model to inherit and utilize the group experience and having a good generalization basis. Secondly, combined with all the maintenance records of the current target ship itself, for the actual occurrence frequency and fault manifestations of each maintenance, weights are dynamically assigned to reflect the unique risk evolution characteristics of the target ship in the actual operation, management, and maintenance processes. This weight design based on the maintenance interval (frequency) and fault distribution can highlight the impact of recent, frequent, or abnormal maintenance behaviors on the risk distribution, fully reflecting the core idea of "personalized migration".

[0036] Through multiple fusions of the above weights and distributions, and combined with normalization processing, a risk distribution migration result that can accurately describe the health status of the target ship and adaptively reflect its evolution process is finally obtained. On this basis, the attention weights of each node are generated by adjusting the distribution, and the risk distribution model in the node space is subjected to probability weighting and normalization, realizing the dynamic learning and efficient adaptation of the model to the ship structure, working conditions, and historical behaviors, providing a solid data and knowledge basis for intelligent risk assessment and proactive operation and maintenance decision-making.

[0037] S4: According to the risk distribution model, actively generate the risk probability of the fault combination in the node space and the risk probability of a single node fault.

[0038] Specifically, the risk probability of a single node fault includes calculating the probability for each node i in the risk distribution model: the fault probability of node i in the risk distribution model; obtaining a risk probability sequence of a single node fault.

[0039] The risk probability of the fault combination includes calculating the probability of a fault occurring for each group of multiple nodes with an associated relationship by using the risk distribution model: using the average value of the fault probabilities of the nodes with an associated relationship; obtaining a risk probability sequence of the fault combination.

[0040] S5: Generate fault hypotheses according to the risk probability of the fault combination in the node space and the risk probability of a single node fault, and evaluate the coverage rate of each fault hypothesis for the risk distribution model, and finally output a fault assessment result.

[0041] Furthermore, perform a random combination of the risk probability sequences of the single-node failures and the risk probability sequences of the failure combinations. By randomly combining the risk probability sequences of single-node failures with those of failure combinations, a large number of failure hypotheses with different levels and structures can be actively generated. This method can simulate various single-point failures, multi-point linkage failures, and rare high-risk scenarios that a ship may encounter during actual operation, avoiding the evaluation results being limited to a finite number of prior-specified risk scenarios.

[0042] This randomized hypothesis generation mechanism can, on the one hand, expand the coverage of risk analysis and enhance the early warning ability for abnormal and boundary events; on the other hand, it also provides a diverse input basis for subsequent distribution coverage calculation and priority ranking, thus supporting the scientific screening and intelligent decision-making of failure modes. Ultimately, it helps to improve the systematicness, completeness, and robustness of failure assessment in practical applications.

[0043] Evaluate the coverage rate of each combination for the risk distribution model. The coverage rate includes re-assigning values to each node in the failure hypothesis, recalculating the value distribution of the node space based on the assignment results; calculating the similarity between the value distribution and the risk distribution model, and using the similarity as the coverage rate of each failure hypothesis. By re-assigning values to each failure hypothesis, constructing its corresponding node space probability distribution, and calculating the similarity with the risk distribution model, the representativeness of the hypothesis in the global risk pattern can be effectively reflected. As an evaluation criterion, the coverage rate can not only quantitatively measure the "matching degree" of a failure hypothesis to the overall risk distribution, but also be used to compare the influence and importance of different failure modes in the global risk. It can effectively screen out those failure modes with high representativeness or typicality, making the risk assessment results more focused on the risk scenarios with greater actual threats, providing more targeted reference basis for intelligent operation and maintenance, resource allocation, and emergency decision-making. At the same time, it also avoids excessive attention to low-correlation and marginal hypotheses, improving the efficiency and scientific nature of risk management.

[0044] If a node is a node failure in a fault combination in the fault hypothesis, the node is assigned the risk probability of the corresponding fault combination; if a node is a single-node failure in the fault hypothesis, it is directly assigned according to the risk probability of the corresponding single-node failure. By assigning nodes belonging to a fault combination the risk probability of the corresponding combined fault, the risks of linkage, amplification, or special impacts faced by the system when multiple nodes fail simultaneously can be accurately expressed, reflecting the systemic complexity brought by combined faults; while directly using the single-point risk probability assignment for nodes belonging to single-node failures highlights the true role and weight of independent failures in the overall risk pattern. This classification and context-based assignment strategy not only avoids the repeated superposition of risk probabilities but also prevents the dilution of risk information in space, making the spatial distribution of nodes generated by each hypothesis more in line with the actual engineering semantics and operation and maintenance requirements.

[0045] Finally, this assignment method can provide a scientific probability basis for subsequent distribution similarity calculation and coverage evaluation, improving the physical interpretability of fault hypothesis assessment and the accuracy of intelligent decision-making.

[0046] Output the fault hypotheses in descending order of coverage as the fault assessment results.

[0047] Embodiment 2. This embodiment also provides a fault assessment system based on ship maintenance, which includes: An acquisition unit that acquires multi-modal data of each monitored part during the normal operation of the ship.

[0048] A distribution unit that uses a knowledge graph to perform fault assessment of nodes and constructs a distribution of health risks according to the results of the fault assessment.

[0049] A migration unit that analyzes the ship maintenance records to migrate the distribution of health risks between nodes to obtain a risk distribution model adapted to the ship; according to the risk distribution model, actively generates the risk probability of a fault combination and the risk probability of a single-node failure for the node space.

[0050] An output unit that generates fault hypotheses according to the risk probability of a fault combination and the risk probability of a single-node failure for the node space, and evaluates the coverage rate of each fault hypothesis for the risk distribution model, and finally outputs the fault assessment results.

[0051] If the above functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes: USB flash drives, mobile hard disks, read-only memories (ROMs), random access memories (RAMs), magnetic disks, or optical discs, etc., which can store program codes of various kinds.

[0052] The logic and / or steps represented in the flowchart or described in other ways herein, for example, can be considered as a predefined sequence list of executable instructions for implementing logical functions, and can be specifically implemented in any computer-readable medium for use by an instruction execution system, apparatus, or device (such as a computer-based system, a system including a processor, or other systems that can fetch and execute instructions from the instruction execution system, apparatus, or device), or in combination with these instruction execution systems, apparatuses, or devices. For the purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transmit a program for use by or in combination with an instruction execution system, apparatus, or device.

[0053] More specific examples (non-exhaustive list) of computer-readable media include the following: electrical connection parts with one or more wirings (electronic devices), portable computer disk cartridges (magnetic devices), random access memories (RAMs), read-only memories (ROMs), erasable programmable read-only memories (EPROMs or flash memories), fiber optic devices, and portable compact disc read-only memories (CDROMs). Additionally, a computer-readable medium can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other media, followed by editing, interpretation, or, if necessary, other appropriate processing, and then storing it in a computer memory.

[0054] It should be understood that each part of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, any one or combination of the following techniques well known in the art can be used: discrete logic circuits having logic gate circuits for implementing logic functions on data signals, application specific integrated circuits having appropriate combinational logic gate circuits, programmable gate arrays (PGAs), field programmable gate arrays (FPGAs), etc.

[0055] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered by the scope of the claims of the present invention.

Claims

1. A fault assessment method based on ship maintenance, characterized in that, Including: During the normal operation of the ship, multi-modal data of each monitored part is collected; Using a knowledge graph for fault assessment of nodes, and constructing a distribution of health risks according to the results of the fault assessment; By analyzing the ship's maintenance records, migrating the distribution of health risks between nodes to obtain a risk distribution model adapted to the ship; According to the risk distribution model, actively generate the risk probability of a fault combination in the node space and the risk probability of a single-node fault; According to the risk probability of the fault combination in the node space and the risk probability of the single-node fault, generate a fault hypothesis, and evaluate the coverage rate of each fault hypothesis for the risk distribution model, and finally output the fault assessment result; The migration of the distribution of health risks between nodes includes using each maintenance record and the maintenance records of the same type of ships to generate an adjusted distribution of health risks; combining all the maintenance records of the current ship to comprehensively generate the result of the distribution migration of health risks; The specific generation steps of the adjusted distribution are as follows: Step 1: Generate the distribution of the health risks of ships of the same type based on the maintenance records of ships of the same type ; Step 2: Generate the weights and failure distributions based on the maintenance records of the current ship for each time; Denote the probability distribution of failures during the $i$-th maintenance of the current ship; the weight during the $i$-th maintenance of the current ship is defined as: , Denote the preset adjustment coefficient; Denote the time interval between the $i$-th maintenance and the previous maintenance; Denote the mapping function that obtains the reference value through a preset functional relationship for the corresponding value; the weight for the health risk distribution of ships of the same type during the $i$-th maintenance of the current ship is defined as: ; Step 3: Assume that the current ship has undergone I times of maintenance, and combine the I times of maintenance records; ； Among them, represents the combined probability distribution; G() represents the normalization function; is identically equal to ; The risk distribution model includes generating the attention weight of each node according to the adjusted distribution; according to the attention weight at each node, performing weighted adjustment of the probability distribution of the distribution D of health risks and then normalizing to achieve the migration of the distribution of health risks.

2. The fault assessment method based on ship maintenance according to claim 1, wherein: The monitored parts include mechanical components and subsystems in the ship structure that can be independently identified and are suitable for installing sensors; The multi-modal data includes data obtained according to the sensor type in each monitored part for describing the operating characteristics of the monitored part.

3. The fault assessment method based on ship maintenance according to claim 2, characterized in that: The knowledge graph includes taking each component in the ship as a node, based on actual connections, functional dependencies, and the relationships between entity nodes extracted from the historical maintenance data of samples; constructing one-way or two-way causal relationships; among them, the causal relationships between nodes are learned through a Bayesian network; If there are multiple nodes in the same monitored part, generalize the multi-modal data of the nodes in the monitored part so that each node in the monitored part synchronizes data according to the data distribution law; The data distribution law includes, for each monitored part, through a pre-trained mapping relationship and the measured value of the sensor, mapping each node in the monitored part to the corresponding monitored data value.

4. The fault assessment method based on ship maintenance according to claim 3, characterized in that: The distribution of health risks includes, according to the monitoring data of each node, taking the health state of each node as the target variable, and combining the causal relationships obtained from the Bayesian network in the knowledge graph; Output the probability distribution of each node under different fault states to form the distribution D of health risks.

5. The fault assessment method based on ship maintenance according to claim 4, characterized in that: The risk probability of a single-node fault includes, in the risk distribution model, calculating the probability for each node i: the fault probability of node i in the risk distribution model; The risk probability of the fault combination includes using the risk distribution model to calculate the probability of a fault occurring for each group of multiple nodes with an associated relationship: the average value of the fault probabilities of the nodes with an associated relationship; Obtain the risk probability sequence of the fault combination; The fault hypothesis includes randomly combining the risk probability sequences of the single-node faults and the risk probability sequences of the fault combinations; Evaluating the coverage rate of each combination for the risk distribution model; The coverage rate includes reassigning each node in the fault hypothesis and recalculating the value distribution of the node space according to the assignment results; Calculating the similarity between the value distribution and the risk distribution model, and using the similarity as the coverage rate of each fault hypothesis; If a node belongs to a node fault in a fault combination in the fault hypothesis, the node is assigned the risk probability of the corresponding fault combination; If a node belongs to a single-node fault in the fault hypothesis, it is directly assigned according to the risk probability of the corresponding single-node fault.

6. The fault assessment method based on ship maintenance according to claim 5, wherein: Output the fault hypotheses in descending order of the coverage rate as the fault assessment result.

7. A fault assessment system based on ship maintenance using the method according to any one of claims 1-6, characterized in that: A collection unit that collects multimodal data of each monitored part during the normal operation of the ship; A distribution unit that uses a knowledge graph to perform fault assessment of nodes and constructs a distribution of health risks according to the results of the fault assessment; A migration unit that migrates the distribution of health risks between nodes through the analysis of ship maintenance records to obtain a risk distribution model adapted to the ship; According to the risk distribution model, actively generate the risk probability of the fault combination and the risk probability of the single-node fault for the node space; An output unit that generates fault hypotheses based on the risk probability of the fault combination and the risk probability of the single-node fault for the node space, and evaluates the coverage rate of each fault hypothesis for the risk distribution model, and finally outputs the fault assessment result.

8. A computer device, comprising: A memory and a processor; The memory stores a computer program, characterized in that: when the processor executes the computer program, the steps of the method according to any one of claims 1-6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, the steps of the method according to any one of claims 1-6 are implemented.

Images