Marine hydrogen system full life cycle detection and evaluation method and system and storage medium

Abstract

This invention discloses a method, system, and storage medium for the full lifecycle detection and evaluation of marine hydrogen systems. The method includes: collecting static prior data and real-time dynamic monitoring data of the marine hydrogen system to build a multi-parameter dataset for the entire lifecycle; identifying key leakage areas based on hydrogen diffusion patterns; completing spatiotemporal alignment and noise reduction of monitoring data; dynamically configuring the weights of each feature parameter based on failure identification data; and generating preliminary feature characterization signals through adaptive preprocessing. The signals are then input into a multi-dimensional risk model for coupled analysis to identify closed-loop scenarios during ship passage through locks and to calculate dynamic risk thresholds based on safe zone theory. Combined with fuel cell operating condition time-series prediction of component degradation life, an evaluation result combining spatial risk and lifespan prediction is formed. Based on the results, tiered measures are implemented, and closed-loop management instructions for the entire lifecycle are output. This provides reliable technical support for the navigation safety of hydrogen-powered ships.

Description

Technical Field

[0001] This invention belongs to the field of marine hydrogen energy safety monitoring technology, and more specifically, relates to a method, system and storage medium for full life cycle detection and evaluation of marine hydrogen systems. Background Technology

[0002] Under the industry trend of green and low-carbon transformation of inland waterway shipping, hydrogen energy, as a zero-pollution alternative energy source, is gradually being applied to the field of inland waterway vessel power equipment. With the continuous increase in the number of hydrogen-powered vessels put into use in the Three Gorges navigation area, routine lock passage operations for hydrogen-powered vessels have become an important part of navigation business. Hydrogen has special physical and chemical properties and a wide explosion limit range. Once a leak occurs, it can easily accumulate in confined spaces, creating safety hazards. In addition, the continuous impact of ship navigation on factors such as hull vibration, changes in ambient temperature and humidity, and fluctuations in cabin ventilation conditions means that core components such as hydrogen storage devices, hydrogen supply pipelines, and fuel cells are under complex alternating operating conditions for a long time, continuously increasing the pressure of safety management throughout the entire life cycle.

[0003] Current marine hydrogen system safety monitoring technologies still have many shortcomings. Most monitoring methods focus only on collecting data from single points of equipment operation, failing to comprehensively integrate basic parameters from the equipment design phase with dynamic operational data throughout the ship's voyage, and lacking a data linkage management system covering the entire lifecycle. Traditional data processing methods treat all sensor data uniformly, failing to differentiate the reliability of data from key monitoring areas based on hydrogen diffusion patterns, making it difficult to effectively eliminate invalid interference signals caused by hull vibration, resulting in insufficient reliability of monitoring data.

[0004] Current risk assessments generally use fixed threshold criteria, severing the correlation between cabin space leakage risk and fuel cell performance aging. This makes it impossible to dynamically adjust risk assessment benchmarks for the special environment of sealed cabins during ship lock passage, and makes it difficult to quantify the combined risks arising from equipment aging and hydrogen leakage. Supporting control mechanisms mostly adopt a passive over-limit alarm mode, which can only trigger emergency shutdown operations after a failure occurs, and cannot develop maintenance plans in advance based on equipment degradation trends.

[0005] In waterways with dense vessel traffic and fixed lock passage sequences, sudden malfunctions in the hydrogen system can not only cause losses due to vessel downtime but also easily lead to major safety accidents such as navigation blockages, explosions, and fires. The limitations of existing control technologies have hindered the large-scale promotion and application of hydrogen-powered vessels, necessitating a new, comprehensive lifecycle testing and evaluation solution to fill existing technological gaps and solidify the foundation for safe navigation of hydrogen-powered vessels on inland waterways. Summary of the Invention

[0006] This invention aims to overcome the shortcomings of existing marine hydrogen system monitoring and evaluation technologies, such as fragmented data, one-sided risk assessment, and fixed thresholds. It relies on multi-source data fusion and adaptive feature processing, combined with the characteristics of the closed space during lock passage, to achieve coupled assessment of space leakage and battery aging. Based on risk level classification and control, it constructs a closed-loop detection and evaluation system for the entire life cycle, thereby improving the operational safety of hydrogen-powered ships in the Three Gorges navigation waters.

[0007] To address the aforementioned deficiencies or improvement needs of existing technologies, as a first aspect of this invention, the present invention provides a method for the full life-cycle testing and evaluation of marine hydrogen systems, comprising: S1. Obtain static prior data and real-time dynamic monitoring data of marine hydrogen systems, and construct a full life cycle multi-parameter input dataset; S2. Adaptive feature extraction and preprocessing are performed on the full life cycle multi-parameter input dataset to obtain a preliminary feature characterization signal; wherein, the key area of ​​hydrogen leakage and diffusion is determined based on the hydrogen diffusion evolution law data, the real-time dynamic monitoring data is spatiotemporally aligned and noise suppressed, and the feature parameters in the full life cycle multi-parameter input dataset are dynamically weighted based on the failure mode identification data, and the preliminary feature characterization signal is output. S3. Input the preliminary feature characterization signal into a multi-dimensional risk assessment model for dynamic coupling analysis to obtain the full life cycle risk assessment result; wherein, the dynamic coupling analysis includes: identifying the lock-passing confined space scenario based on the ship's scenario state data, and calculating the dynamic risk threshold using the safety zone theory model; performing time-series prediction of the fuel cell lifespan degradation trend based on the fuel cell operating condition data, and outputting the full life cycle risk assessment result including spatial risk distribution and time lifespan prediction; S4. Based on the full life cycle risk assessment results, execute hierarchical response and adaptive decision-making, and output full life cycle closed-loop management instructions.

[0008] Furthermore, the static prior data in S1 includes hydrogen diffusion evolution data, failure mode identification data, and component initial performance baseline data determined during the design verification phase. The real-time dynamic monitoring data includes the spatiotemporal distribution data of hydrogen concentration in the cabin, the stress and pressure data of hydrogen-related pipelines, the operating condition data of fuel cells, and the status data of the ship's environment during the ship's operation.

[0009] Furthermore, in step S2, the key areas for hydrogen leakage and diffusion are determined based on hydrogen diffusion evolution data, and the real-time dynamic monitoring data undergoes spatiotemporal alignment and noise suppression, specifically including: Computational fluid dynamics simulation is used to simulate the leakage and diffusion process of hydrogen in the fuel cell site and hydrogen storage chamber. The three-dimensional spatial coordinates of the hydrogen accumulation concentration exceeding the preset threshold are extracted and identified as the key area of ​​hydrogen leakage and diffusion. The key area of ​​hydrogen leakage and diffusion includes at least the four corners of the top of the chamber and the flange connection of the hydrogen pipeline. A spatial confidence matrix is ​​constructed based on the three-dimensional spatial coordinates of the key area of ​​hydrogen leakage and diffusion. High confidence weights are assigned to hydrogen concentration data collected by sensors arranged in the key area of ​​hydrogen leakage and diffusion, and low confidence weights are assigned to sensor data in non-key areas. The Kalman filter algorithm, combined with the spatial confidence matrix, is used to suppress noise in the real-time dynamic monitoring data, eliminating abrupt interference signals caused by ship vibration, and outputting the spatiotemporally aligned dynamic monitoring data.

[0010] Furthermore, in step S2, dynamic weight allocation is performed on the feature parameters in the full lifecycle multi-parameter input dataset based on the failure mode identification data, specifically including: Failure Mode, Effects and Criticality Analysis (FEMA) is used to identify the risks of hydrogen storage, hydrogen supply and fuel cell modules in marine hydrogen systems, and to calculate the hazard priority number of each component's failure mode in order to update the failure mode identification data. A nonlinear mapping function is constructed between the hazard priority number and the weights of the feature parameters. When the hazard priority number of a certain component exceeds a set threshold, the feature weights of the stress and pressure data of the hydrogen-related pipeline and the spatiotemporal distribution data of the hydrogen concentration in the cabin corresponding to that component are dynamically increased. During ship operation, the hazard priority number is updated in real time based on the cumulative operating time of the components and historical fault records, thereby completing the dynamic adaptive adjustment of the weight of the feature parameters.

[0011] Furthermore, in step S3, the process of identifying the enclosed space scenario for lock passage based on the ship's current situation data and calculating the dynamic risk threshold using a safety zone theoretical model specifically includes: When a ship is detected to be entering a lock passage confined space based on the ship's current scene status data, the estimated time of the current lock passage process and the spatial environment parameters of the lock passage confined space are obtained. The hydrogen leakage rate in the preliminary feature characterization signal vector and the space environment parameters are input into a pre-constructed safe zone theoretical model to calculate the critical time required for the hydrogen concentration in the closed space to reach the lower explosive limit. Using the critical time being equal to the expected time as a safety boundary condition, the maximum allowable leakage rate threshold under the current scenario is calculated by reverse calculation. Based on the maximum allowable leakage rate threshold and leakage source parameters, calculate the diffusion distance when the hydrogen concentration decays to the lower explosive limit concentration, and use it as the safe distance threshold. The maximum allowable leakage rate threshold and the safe distance threshold are used together as the dynamic risk threshold.

[0012] Furthermore, in step S3, the time-series prediction of the fuel cell lifespan degradation trend based on the fuel cell operating condition data specifically includes: The output voltage and current fluctuation data of the fuel cell stack are obtained, and the actual output load spectrum of the fuel cell stack under actual operating conditions is extracted. The actual output load spectrum is divided into working condition segments, and a multidimensional aging stress sequence containing start-stop cycle characteristics, dynamic load characteristics, and steady-state operation characteristics is extracted. The multidimensional aging stress sequence is input into a pre-trained fuel cell performance degradation time-series prediction model, which outputs the health status degradation sequence of the fuel cell stack within a future preset period. Based on the health state decay sequence, the initial polarization curve in the initial performance baseline data of the component is decayed and reconstructed to obtain the predicted polarization curve. Based on the predicted polarization curve and the preset failure voltage threshold, the voltage decay rate and remaining service life are calculated as the time life prediction.

[0013] Furthermore, in S3, the specific calculation method for the full life-cycle risk assessment result is as follows: in, This is a comprehensive risk assessment value covering the entire life cycle. To construct a space risk index, This is a time risk index. Indicates the basic weight of spatial risk. Indicates the basic weight of time risk. This represents the spatiotemporal coupling penalty factor, used to amplify the nonlinear risks associated with the superposition of spatial leakage and time aging.

[0014] Furthermore, in step S4, based on the full lifecycle risk assessment results, hierarchical response and adaptive decision-making are executed, and full lifecycle closed-loop management instructions are output, specifically including: When the spatial risk distribution shows that the local hydrogen concentration exceeds the first-level warning threshold, a linkage ventilation and explosion-proof command is output to start the explosion-proof fan for forced ventilation. When the local hydrogen concentration exceeds the level 2 alarm threshold or when the stress and pressure data of the hydrogen-related pipeline show abnormal changes, an emergency cut-off and inerting command is output, the main valve of the hydrogen supply pipeline is automatically closed, and the automatic inerting and hydrogen replacement program is started to inject inert gas into the hydrogen-related pipeline and the compartment. When the time-life prediction results show that the performance degradation of the fuel cell stack or hydrogen storage container reaches the maintenance threshold, a preventive maintenance instruction is output. The preventive maintenance instruction includes generating a hydrogen embrittlement-resistant non-destructive testing plan for hydrogen-related pipelines, as well as a calibration strategy for refueling communication connectors and robotic arms.

[0015] As a second aspect of the present invention, a system for the full life cycle testing and evaluation of marine hydrogen systems is also provided, comprising: The multi-parameter dataset construction unit is used to acquire static prior data and real-time dynamic monitoring data of marine hydrogen systems and construct a full life cycle multi-parameter input dataset. An adaptive feature extraction and processing unit is used to perform adaptive feature extraction and preprocessing on the full life cycle multi-parameter input dataset to obtain a preliminary feature characterization signal; wherein, based on the hydrogen diffusion evolution law data, key areas of hydrogen leakage diffusion are determined, the real-time dynamic monitoring data is spatiotemporally aligned and noise suppressed, and the feature parameters in the full life cycle multi-parameter input dataset are dynamically weighted based on failure mode identification data, and the preliminary feature characterization signal is output. A multi-dimensional risk coupling assessment unit is used to input the preliminary feature characterization signal into a multi-dimensional risk assessment model for dynamic coupling analysis to obtain the full life cycle risk assessment result. The dynamic coupling analysis includes: identifying the confined space scenario of the lock passage based on the ship's current scenario status data, and calculating the dynamic risk threshold using a safe zone theory model; and performing time-series prediction of the fuel cell lifespan degradation trend based on the fuel cell operating condition data, outputting the full life cycle risk assessment result that includes spatial risk distribution and temporal lifespan prediction. The hierarchical response closed-loop management unit is used to execute hierarchical response and adaptive decision-making based on the full life cycle risk assessment results, and output full life cycle closed-loop management instructions.

[0016] As a third aspect of the invention, a computer-readable storage medium is also provided, on which a computer program is stored, which is executed by a processor as described in any one of the claims: a method for full life cycle detection and evaluation of a marine hydrogen system.

[0017] In summary, compared with the prior art, the above-described technical solutions conceived by this invention can achieve the following beneficial effects: 1. The marine hydrogen system full life cycle detection and evaluation method of this invention integrates static prior data and real-time dynamic monitoring data of the ship to construct a full life cycle multi-parameter dataset. It identifies key leakage areas based on hydrogen diffusion evolution laws, uses spatial confidence matrix and Kalman filtering to achieve spatiotemporal alignment and noise suppression of monitoring data, calculates the component hazard priority number based on failure mode identification results, dynamically adjusts the weights of each characteristic parameter using nonlinear mapping relationships, and generates preliminary characteristic signals through weighted calculations. Static data retains system design, component failure, and factory performance benchmark information, while dynamic data synchronously collects concentration, pressure, operating conditions, and environmental information during navigation. The two types of data complement each other to improve the full life cycle data source. Key area division and filtering can filter out interference data generated by hull vibration, enhance the reference value of sensor data at high-risk points, and continuously adjust the dynamic weights according to component operating wear and fault records, amplifying the proportion of high-risk indicators in the characteristic signal, reducing interference from redundant and invalid data, ensuring the data validity of the pre-processed signal, and providing data support for subsequent risk analysis.

[0018] 2. The marine hydrogen system full life cycle detection and evaluation method of the present invention imports the pre-processed preliminary characteristic signal into a multi-dimensional risk assessment model. Based on ship scenario information, it determines the operating conditions of the confined space during lock passage. Utilizing the gas accumulation law of the confined space and the safe zone theory, it calculates the dynamic leakage threshold and safe distance threshold. It extracts multiple types of aging stress data by splitting the actual operating load spectrum of the fuel cell. A time-series prediction model is used to deduce the battery health degradation trend and remaining service life. Finally, relying on a spatiotemporal coupling calculation formula, it integrates spatial and temporal risks to obtain a comprehensive evaluation value. Addressing the actual characteristics of the confined space and limited ventilation conditions in lock passage compartments, it dynamically adjusts the risk judgment criteria, overcoming the limitation of fixed thresholds being unsuitable for special navigation scenarios. Battery aging time-series prediction can quantify the performance degradation caused by long-term operation. The coupled calculation mode objectively reflects the superimposed risks generated by the combined effects of leakage hazards and equipment aging. The final evaluation result simultaneously includes two categories: spatial risk distribution and device lifespan changes, realizing the transformation of risk assessment from point-based monitoring to comprehensive quantitative assessment.

[0019] 3. The marine hydrogen system full life cycle detection and assessment method of the present invention uses the comprehensive risk assessment results of the entire life cycle as the basis for decision-making, classifies response levels according to the severity of risks, and outputs corresponding closed-loop management and control commands for different scenarios. When the hydrogen concentration in the cabin slightly exceeds the standard, a ventilation and explosion-proof command is issued. When the concentration exceeds the standard or the pipeline parameters are abnormal, the hydrogen supply pipeline is cut off and the cabin is inerted. When the equipment life index is close to the preset maintenance limit, a component preventive maintenance plan is generated. The graded disposal mode can match fault scenarios with different levels of danger. Low-level hazards rely on ventilation to improve the gas environment in the cabin. High-risk faults quickly isolate the hydrogen source to block the spread of the leakage source. The maintenance plan generated based on life prediction can complete non-destructive testing and component calibration before component failure. The entire disposal process relies on the quantitative assessment data in the early stage for implementation, opens up the connection between risk calculation and on-site control, and forms a coherent full life cycle closed-loop management mode of monitoring, assessment, disposal and maintenance, steadily reducing the probability of safety accidents of hydrogen-powered ships in navigation operations. Attached Figure Description

[0020] Figure 1 This is a flowchart of a method for full life cycle testing and evaluation of a marine hydrogen system according to an embodiment of the present invention; Figure 2 This is a schematic diagram of a marine hydrogen fuel cell system according to an embodiment of the present invention; Figure 3 This is a schematic diagram of the system units in an embodiment of the present invention. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. Furthermore, the technical features involved in the various embodiments of this invention described below can be combined with each other as long as they do not conflict with each other.

[0022] Example 1 Please refer to Figure 1 This embodiment 1 provides a method for the full life cycle testing and evaluation of a marine hydrogen system, including: S1. Obtain static prior data and real-time dynamic monitoring data of marine hydrogen systems, and construct a full life cycle multi-parameter input dataset; S2. Adaptive feature extraction and preprocessing are performed on the full life cycle multi-parameter input dataset to obtain a preliminary feature characterization signal; wherein, the key area of ​​hydrogen leakage and diffusion is determined based on the hydrogen diffusion evolution law data, the real-time dynamic monitoring data is spatiotemporally aligned and noise suppressed, and the feature parameters in the full life cycle multi-parameter input dataset are dynamically weighted based on the failure mode identification data, and the preliminary feature characterization signal is output. S3. Input the preliminary feature characterization signal into a multi-dimensional risk assessment model for dynamic coupling analysis to obtain the full life cycle risk assessment result; wherein, the dynamic coupling analysis includes: identifying the lock-passing confined space scenario based on the ship's scenario state data, and calculating the dynamic risk threshold using the safety zone theory model; performing time-series prediction of the fuel cell lifespan degradation trend based on the fuel cell operating condition data, and outputting the full life cycle risk assessment result including spatial risk distribution and time lifespan prediction; S4. Based on the full life cycle risk assessment results, execute hierarchical response and adaptive decision-making, and output full life cycle closed-loop management instructions.

[0023] Please refer to Figure 2 Marine hydrogen fuel cell systems mainly include high-pressure hydrogen cylinders, power generation devices, DC / DC devices, loads, air compressors, humidifiers, heat exchangers, batteries, and monitoring devices.

[0024] Regarding the material flow path and electrical connections, the high-pressure hydrogen cylinder is connected to the hydrogen inlet of the power generation device via a hydrogen-related pipeline to supply the hydrogen fuel required for the reaction; the air compressor is connected to an air pipeline to provide the power generation device with oxidant (air); the humidifier is installed in the gas pipeline to regulate the humidity of the reaction gas; and the heat exchanger is used to manage the system's thermal balance. The power output terminal of the power generation device is electrically connected to the input terminal of the DC / DC device, and the output terminal of the DC / DC device is electrically connected to the load and the battery respectively to realize the conversion, distribution, and storage of electrical energy; the exhaust port of the power generation device is used to discharge the water and exhaust gas generated by the electrochemical reaction.

[0025] In terms of data acquisition and control, the monitoring device serves as the control and data processing center of the system, and is communicatively connected to the valves and sensors of the high-pressure hydrogen cylinder, the power generation device, the DC / DC device, the air compressor, and the ship condition monitoring equipment. The monitoring device is used to collect multi-source heterogeneous data (including but not limited to pipeline pressure, hydrogen concentration, fuel cell voltage / current, and temperature) in real time during system operation, and to receive scenario status data of the ship (such as location information in the confined space of the lock).

[0026] Based on the aforementioned hardware architecture, the monitoring device has a built-in or communication-connected data processing and evaluation module for executing the full lifecycle detection and evaluation method for marine hydrogen systems described in this embodiment of the invention. Specifically, the monitoring device constructs a multi-parameter input dataset by aggregating static prior data and real-time dynamic monitoring data from each node. After adaptive feature extraction and preprocessing, the dataset is input into a multi-dimensional risk assessment model. Finally, based on the output full lifecycle risk assessment results, the device issues graded response and closed-loop management commands to actuators such as the shut-off valve of the high-pressure hydrogen cylinder and the air compressor, thereby achieving full lifecycle safety monitoring and control of the marine hydrogen fuel cell system.

[0027] Based on the above framework, this embodiment 1 will further elaborate on the specific steps of the method it employs.

[0028] (1) Construction of multi-parameter dataset First, static prior data and real-time dynamic monitoring data of the marine hydrogen system are acquired. The static prior data are derived from the design verification phase: first, hydrogen diffusion evolution data, which is obtained by simulating the leakage and diffusion process of hydrogen in the compartment through computational fluid dynamics simulation; second, failure mode identification data, which is obtained by analyzing the failure modes of key components and calculating the risk priority number using the FMECA method; and third, initial performance baseline data of components, which is extracted from factory bench tests and used as a benchmark for performance degradation assessment.

[0029] Real-time dynamic monitoring data is collected during ship operation: First, the spatiotemporal distribution data of hydrogen concentration in the cabin is collected in seconds by a sensor array in the easy accumulation area; second, the stress and pressure data of hydrogen-related pipelines are recorded in real time, including pressure fluctuations and stress changes; third, the operating condition data of fuel cells are read, including operating parameters such as voltage and temperature; and fourth, the ship scene status data are obtained, including location, speed and whether it has entered the lock.

[0030] Finally, preprocessing is performed to construct the dataset: outliers and missing values ​​in dynamic data are removed and imputed by interpolation; static data are treated as constant features and dynamic data as time-series variable features, which are then aligned and stitched together with a unified timestamp, and finally converted into a multi-dimensional feature matrix to complete the construction of a full lifecycle multi-parameter input dataset.

[0031] (2) Adaptive feature extraction processing After constructing the full lifecycle multi-parameter input dataset, adaptive feature extraction and preprocessing are performed on the full lifecycle multi-parameter input dataset to obtain preliminary feature representation signals.

[0032] First, the key areas for hydrogen leakage and diffusion are determined based on data on the evolution of hydrogen diffusion patterns. Then, the real-time dynamic monitoring data is spatiotemporally aligned and noise suppressed. The specific process is as follows: A three-dimensional fluid domain model of the fuel cell site and hydrogen storage chamber was constructed using computational fluid dynamics simulation software. Hydrogen leakage sources and ventilation boundary conditions were set to simulate the leakage and diffusion process of hydrogen within the chamber. In the simulation results, the three-dimensional spatial coordinates of hydrogen accumulation concentrations exceeding a preset safety threshold were extracted and identified as critical areas for hydrogen leakage and diffusion. Based on the physical characteristics of hydrogen—low density and tendency to accumulate upwards—and the leakage probability at pipe connections, the identified critical areas include at least the four corners of the chamber ceiling and the flange connections of hydrogen-related pipelines.

[0033] After identifying the critical areas, a spatial confidence matrix is ​​constructed based on the three-dimensional spatial coordinates of these areas. The dimension of this matrix corresponds to the number of nodes in the hydrogen sensor array deployed within the cabin. Sensors located within the critical areas of hydrogen leakage and diffusion are assigned high confidence weights in their corresponding matrix elements; sensors located in non-critical areas are assigned low confidence weights. Subsequently, a Kalman filter algorithm is used to suppress noise in the real-time dynamic monitoring data. In the observation equation of the Kalman filter, the spatial confidence matrix is ​​used as an adjustment factor for the observation noise covariance matrix. Sensor data with higher confidence weights have smaller observation noise covariances, thus achieving greater Kalman gain in state estimation. This method eliminates high-frequency abrupt interference signals caused by vibrations due to ship engine operation or wave impact, and interpolates multi-source sensor data to the same timestamp, achieving spatiotemporal alignment and outputting processed dynamic monitoring data.

[0034] Based on the preceding processing, the feature parameters in the full lifecycle multi-parameter input dataset are then dynamically weighted according to the failure mode identification data. The specific process is as follows: First, initial risk identification and basic parameter settings are conducted. For each component in the marine hydrogen system involved in the assessment, failure mode, effects, and hazard analysis (FMEA) methods are used for risk identification. The total number of components involved in the assessment is set to [number missing]. And set the index of the component being evaluated to Regarding the first Each component is assessed for its failure mode based on the severity of the hydrogen leak explosion consequences assessment, denoted as... The initial occurrence frequency based on the factory-set failure rate is denoted as . And evaluate its initial undetectability based on sensor monitoring coverage, denoted as Based on the above three dimensions, calculate the first... Initial hazard priority number of each component Its calculation formula is At the same time, the maximum allowable hazard priority number is set to... and for the first Each feature parameter is assigned an initial base weight. .

[0035] After initial risk identification, a nonlinear mapping model between hazard priority numbers and feature parameter weights is constructed to calculate the feature weights in the initial state. A weight amplification adjustment coefficient is introduced. And set the traversal index variable in the normalization summation process as . No. Dynamic weights of each feature parameter The calculation is performed using an exponential normalization model, and the specific formula is as follows: This formula allows for the non-linear amplification of the weights of characteristic parameters corresponding to high-risk components based on the initial hazard priority of each component, thereby completing the initial weight allocation of characteristic parameters.

[0036] As the ship continues to operate, it enters a dynamic adaptive adjustment phase, utilizing the cumulative operating time of components and historical fault records to correct risk parameters in real time. The cumulative operating time of each component is recorded as follows: Furthermore, the Weibull distribution function is introduced to characterize the aging and degradation properties of the component. In the Weibull distribution function, the shape parameter is set as... Set the scale parameter as Utilizing cumulative running time The initial occurrence frequency is corrected to obtain the first occurrence frequency. Frequency of correction of each component during the current operating time The calculation formula is as follows: This formula quantifies the increasing failure probability of a component as operating time increases by using the cumulative failure probability characteristics of the Weibull distribution.

[0037] At the same time, the system retrieves the first... The historical fault records of each component are used to correct its undetectability. Statistics are presented in the first... The total number of failures of each component during its historical operating cycle is denoted as: The number of missed reports that were not captured by the sensors in time is recorded as follows. The initial undetectability is corrected using the ratio of missed detections to total failures, resulting in the... The undetectability of the correction of each component at the current runtime. The calculation formula is as follows: This formula reflects the blind spots or performance degradation that may exist in the actual operation of the sensor monitoring system. The higher the false negative rate, the greater the undetectability after correction.

[0038] Finally, based on the corrected occurrence frequency and undetectability, the dynamic weights of the feature parameters are recalculated and updated. The corrected occurrence frequency... With correction of undetectability Substituting into the hazard priority number calculation formula, we get the first... The hazard priority of each component after its update at the current runtime. Its calculation formula is Subsequently, the updated hazard priority number will be... Substituting back into the aforementioned exponential normalization model, we obtain the... The updated dynamic weights of each feature parameter at the current runtime. Through the aforementioned closed-loop calculation mechanism, the system can continuously and quantitatively adaptively adjust the weights of each feature parameter based on the actual aging state of the components and the actual performance of the monitoring system, ensuring the accuracy and timeliness of the input features of the full life cycle risk assessment model.

[0039] After obtaining the updated dynamic weights, the preliminary feature representation signal is further calculated through multi-dimensional feature weighted mapping. The first parameter in the full lifecycle multi-parameter input dataset is then obtained. The standardized input values ​​of the feature parameters at the current runtime are denoted as follows: The standardized input values ​​of all feature parameters are combined into a multidimensional input feature vector. Simultaneously, the updated dynamic weights of all feature parameters are combined into a dynamic weight vector. To highlight the characteristic responses of high-risk components while preserving information across multiple parameter spatial dimensions, the system processes the multi-dimensional input feature vector. With dynamic weight vector Performing the Hadamard product operation, i.e., element-wise multiplication, yields the preliminary feature representation signal vector at the current runtime. The preliminary feature characterization signal vector. The Middle element values ​​of each dimension The specific calculation formula is as follows: Through this element-wise weighted mapping process, the amplitude of high-weight feature parameters with greater risk impact is amplified in their corresponding signal dimensions, while the amplitude of low-weight feature parameters in their corresponding signal dimensions is suppressed. The final output is the preliminary feature representation signal vector that retains the multi-parameter spatial structure throughout the entire lifecycle and integrates the real-time health status and risk evolution characteristics of the equipment. This provides high signal-to-noise ratio multidimensional quantitative data input for subsequent in-depth evaluation and lifetime prediction models.

[0040] (3) Multidimensional risk coupling assessment After obtaining the preliminary feature representation signal, the signal is input into a multi-dimensional risk assessment model for dynamic coupling analysis to obtain the full life-cycle risk assessment result. First, the system identifies the lock passage confined space scenario based on the ship's current situation status data and calculates the dynamic risk threshold using a safety zone theoretical model. When the system detects that a ship has entered the lock passage confined space based on the ship's current situation status data, it obtains the estimated time for the current lock passage process. and the spatial environment parameters of the gate-passing sealed space, including the real-time internal net volume of the gate-passing sealed space. Total equivalent ventilation flow Subsequently, the hydrogen leakage rate in the preliminary feature characterization signal vector is used. The spatial environment parameters are input into a pre-constructed theoretical model of the safe zone. Based on the principle of mass conservation in a confined space, the concentration accumulation calculation formula is used. The hydrogen concentration in the enclosed space of the gate was calculated to reach the lower explosive limit. Required critical time ,in express The hydrogen concentration at time [time]. Represents the natural constant.

[0041] Next, at the critical time Equal to the estimated time For safety boundary conditions, soon and Substituting the values ​​into the concentration accumulation calculation formula and performing an inverse algebraic solution, the maximum allowable leakage rate threshold under the current scenario can be calculated. Its calculation formula is Furthermore, based on the stated maximum allowable leakage rate threshold... Based on the leakage source parameters, calculate the hydrogen concentration decay to the lower explosive limit. The diffusion distance at that time is used as a safety distance threshold. The leakage source parameters include the effective nozzle area of ​​the hydrogen leakage source. and ambient wind speed The safe distance threshold Based on the jet diffusion model, the formula is: ,in This represents the preset jet diffusion correction coefficient. Finally, the maximum allowable leakage rate threshold is... With the aforementioned safe distance threshold Together they serve as the dynamic risk threshold.

[0042] Simultaneously, an operation is performed to predict the lifespan degradation trend of the fuel cell based on the fuel cell operating condition data. Output voltage and current fluctuation data of the fuel cell stack are acquired, and the actual output load spectrum of the fuel cell stack under actual operating conditions is extracted. The actual output load spectrum is divided into operating condition segments, and a multidimensional aging stress sequence including start-stop cycle characteristics, dynamic load variation characteristics, and steady-state operation characteristics is extracted. The multidimensional aging stress sequence is input into a pre-trained fuel cell performance degradation time-series prediction model, which outputs the health state degradation sequence of the fuel cell stack within a preset future period. ,in This represents a time variable within a predetermined future period. express Health status value at any given time.

[0043] Subsequently, based on the health status decay sequence The initial polarization curve in the component's initial performance baseline data is attenuated and reconstructed to obtain the predicted polarization curve. The initial polarization curve represents the initial output voltage. With operating current The functional relationship is given by the formula for reconstructing the predicted polarization curve. ,in Indicates time and given operating current The predicted output voltage is calculated based on the predicted polarization curve and a preset failure voltage threshold. At rated operating current Calculate the voltage attenuation rate Its calculation formula is At the same time, solve for conditions. Failure time point and the failure time point As of remaining useful life The voltage attenuation rate With the remaining service life Together they serve as time-life predictions.

[0044] Finally, the operation outputs the full life-cycle risk assessment results, which include spatial risk distribution and temporal lifetime prediction. First, a spatial risk index is constructed. Its calculation formula is ,in This represents the risk weighting coefficient for leakage rate. This represents the risk weighting coefficient for diffusion distance. This represents the preset reference safety distance. Next, a time risk index is constructed. Its calculation formula is ,in This represents the voltage attenuation risk weighting coefficient. This represents the life loss risk weighting coefficient. This represents the preset design service life baseline value. Subsequently, the comprehensive life-cycle risk assessment value is calculated. Its calculation formula is ,in Indicates the basic weight of spatial risk. Indicates the basic weight of time risk. This represents the spatiotemporal coupling penalty factor, used to amplify the nonlinear risks arising from the superposition of spatial leakage and time aging. When generating the spatial risk distribution, the hydrogen leakage source is taken as the origin, and the aforementioned safety distance threshold is used... As the core warning radius, a three-dimensional risk contour surface is generated by combining the ship's three-dimensional digital model, and within the three-dimensional risk contour surface, a distance attenuation function is applied. Render the concentration gradient distribution, where This indicates the actual spatial distance from the leak source. Indicates distance The estimated hydrogen concentration at the location. Finally, the comprehensive life-cycle risk assessment value... The three-dimensional risk isosurface and the voltage attenuation rate included and the remaining service life The time-series decay curves are encapsulated to output the full life-cycle risk assessment results, which include spatial risk distribution and time lifetime prediction.

[0045] (4) Hierarchical response closed-loop management After completing the multidimensional risk coupling assessment, hierarchical response and adaptive decision-making are executed based on the full life cycle risk assessment results, and full life cycle closed-loop management instructions are output.

[0046] First, by receiving real-time assessment data containing spatial risk distribution and time-life prediction results, the data is compared with preset safety and maintenance thresholds. Then, based on the comparison results, closed-loop management actions of corresponding levels, such as ventilation and explosion disposal, emergency shut-off and inerting, or preventive maintenance, are triggered through the ship's integrated control platform and human-machine interface.

[0047] To address the hydrogen leakage risk in the spatial risk distribution, a primary warning threshold and a secondary alarm threshold were established. When the spatial risk distribution indicates that the local hydrogen concentration in the compartment exceeds the primary warning threshold, an initial leakage risk is identified, and a coordinated ventilation and explosion-proof command is immediately issued. This command is transmitted through the ship's control system to the compartment environment control unit, which automatically activates the explosion-proof fans located at the top and bottom of the compartment for forced ventilation. Simultaneously, the compartment's exhaust louvers and air inlet dampers are opened, and the fan speed is adjusted to its rated maximum power, creating forced air convection from bottom to top within the compartment to dilute and expel the accumulated hydrogen. When the sensor array detects that the local hydrogen concentration has continuously decreased and stabilized below the primary warning threshold, the command is automatically deactivated, and the explosion-proof fans return to their normal monitoring speed.

[0048] When the local hydrogen concentration continues to rise and exceeds the secondary alarm threshold, or when abnormal changes occur in the stress and pressure data of the hydrogen-related pipelines, a serious leak or pipeline rupture risk is identified, and an emergency shut-off and inerting command is immediately issued. For the judgment of abnormal changes in pipeline data, the rate of pressure drop and stress change are calculated in real time. When the pressure drop rate exceeds the set safety threshold or the stress value exceeds the safe allowable proportion of the material's yield strength, the command is triggered. Upon execution, an emergency shut-off signal is first sent to the cylinder valve of the hydrogen storage tank group and the main shut-off valve of the hydrogen supply pipeline, physically cutting off the hydrogen supply source. Subsequently, the automatic inerting and hydrogen replacement program is initiated, opening the electromagnetic release valves of the onboard nitrogen or argon inert gas storage tanks, and injecting inert gas into the hydrogen-related pipelines and cabin space through a dedicated replacement pipeline. During the injection process, the oxygen and hydrogen concentrations in the cabin are continuously monitored until the hydrogen concentration drops below the safety threshold and the oxygen concentration decreases to a set value that cannot support combustion, completing the inerting protection.

[0049] To address the component aging risks reflected in the time-life prediction results, corresponding performance degradation maintenance thresholds are set. When the time-life prediction results show that the voltage degradation rate of the fuel cell stack exceeds a set percentage, or the cumulative charge-discharge cycles and fatigue damage accumulation of the hydrogen storage container reach the design life limit, the component performance degradation is determined to have reached the maintenance threshold, and a preventive maintenance command is immediately issued. This command is transmitted to the ship's shore-based maintenance management system or shipboard maintenance terminal. The preventive maintenance command contains two parts: First, based on the material, operating pressure, and cumulative service time of the hydrogen-related pipelines, an automatic hydrogen embrittlement-resistant non-destructive testing plan is generated for the hydrogen-related pipelines, specifying the testing cycle, testing locations, and recommended ultrasonic or phased array flaw detection parameters; Second, for the communication connectors and robotic arms during the hydrogen refueling process, a calibration strategy is generated, specifying the calibration frequency, interface sealing test standards, and communication protocol handshake verification procedures to ensure the reliability of physical connections and data transmission during the refueling operation.

[0050] Example 2 Please refer to Figure 3 This embodiment 2 provides a marine hydrogen system full life cycle testing and evaluation system, including: The multi-parameter dataset construction unit is used to acquire static prior data and real-time dynamic monitoring data of marine hydrogen systems and construct a full life cycle multi-parameter input dataset. An adaptive feature extraction and processing unit is used to perform adaptive feature extraction and preprocessing on the full life cycle multi-parameter input dataset to obtain a preliminary feature characterization signal; wherein, based on the hydrogen diffusion evolution law data, key areas of hydrogen leakage diffusion are determined, the real-time dynamic monitoring data is spatiotemporally aligned and noise suppressed, and the feature parameters in the full life cycle multi-parameter input dataset are dynamically weighted based on failure mode identification data, and the preliminary feature characterization signal is output. A multi-dimensional risk coupling assessment unit is used to input the preliminary feature characterization signal into a multi-dimensional risk assessment model for dynamic coupling analysis to obtain the full life cycle risk assessment result. The dynamic coupling analysis includes: identifying the confined space scenario of the lock passage based on the ship's current scenario status data, and calculating the dynamic risk threshold using a safe zone theory model; and performing time-series prediction of the fuel cell lifespan degradation trend based on the fuel cell operating condition data, outputting the full life cycle risk assessment result that includes spatial risk distribution and temporal lifespan prediction. The hierarchical response closed-loop management unit is used to execute hierarchical response and adaptive decision-making based on the full life cycle risk assessment results, and output full life cycle closed-loop management instructions.

[0051] Example 3 This embodiment 3 also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, can implement any step of a method for full life cycle detection and evaluation of a marine hydrogen system.

[0052] The computer-readable storage medium may include various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0053] For a description of the computer-readable storage medium provided in this application, please refer to the above method embodiments; further details will not be repeated here.

[0054] Those skilled in the art will readily understand that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for the whole life cycle testing and evaluation of a marine hydrogen system, characterized in that, include: S1. Obtain static prior data and real-time dynamic monitoring data of marine hydrogen systems, and construct a full life cycle multi-parameter input dataset; S2. Adaptive feature extraction and preprocessing are performed on the full life cycle multi-parameter input dataset to obtain a preliminary feature characterization signal; wherein, the key area of ​​hydrogen leakage and diffusion is determined based on the hydrogen diffusion evolution law data, the real-time dynamic monitoring data is spatiotemporally aligned and noise suppressed, and the feature parameters in the full life cycle multi-parameter input dataset are dynamically weighted based on the failure mode identification data, and the preliminary feature characterization signal is output. S3. Input the preliminary feature characterization signal into a multi-dimensional risk assessment model for dynamic coupling analysis to obtain the full life cycle risk assessment result; wherein, the dynamic coupling analysis includes: identifying the lock-passing confined space scenario based on the ship's scenario state data, and calculating the dynamic risk threshold using the safety zone theory model; performing time-series prediction of the fuel cell lifespan degradation trend based on the fuel cell operating condition data, and outputting the full life cycle risk assessment result including spatial risk distribution and time lifespan prediction; S4. Based on the full life cycle risk assessment results, execute hierarchical response and adaptive decision-making, and output full life cycle closed-loop management instructions.

2. The method for full life-cycle testing and evaluation of a marine hydrogen system according to claim 1, characterized in that, The static prior data in S1 includes hydrogen diffusion evolution data, failure mode identification data, and component initial performance baseline data determined during the design verification phase. The real-time dynamic monitoring data includes the spatiotemporal distribution data of hydrogen concentration in the cabin, the stress and pressure data of hydrogen-related pipelines, the operating condition data of fuel cells, and the status data of the ship's environment during the ship's operation.

3. The method for full life-cycle testing and evaluation of a marine hydrogen system according to claim 1, characterized in that, In step S2, the key area for hydrogen leakage and diffusion is determined based on hydrogen diffusion evolution data, and the real-time dynamic monitoring data is spatiotemporally aligned and noise suppressed, specifically including: Computational fluid dynamics simulation is used to simulate the leakage and diffusion process of hydrogen in the fuel cell site and hydrogen storage chamber. The three-dimensional spatial coordinates of the hydrogen accumulation concentration exceeding the preset threshold are extracted and identified as the key area of ​​hydrogen leakage and diffusion. The key area of ​​hydrogen leakage and diffusion includes at least the four corners of the top of the chamber and the flange connection of the hydrogen pipeline. A spatial confidence matrix is ​​constructed based on the three-dimensional spatial coordinates of the key area of ​​hydrogen leakage and diffusion. High confidence weights are assigned to hydrogen concentration data collected by sensors arranged in the key area of ​​hydrogen leakage and diffusion, and low confidence weights are assigned to sensor data in non-key areas. The Kalman filter algorithm, combined with the spatial confidence matrix, is used to suppress noise in the real-time dynamic monitoring data, eliminating abrupt interference signals caused by ship vibration, and outputting the spatiotemporally aligned dynamic monitoring data.

4. The method for full life-cycle testing and evaluation of a marine hydrogen system according to claim 1, characterized in that, In step S2, dynamic weight allocation is performed on the feature parameters in the full lifecycle multi-parameter input dataset based on failure mode identification data, specifically including: Failure Mode, Effects and Criticality Analysis (FEMA) is used to identify the risks of hydrogen storage, hydrogen supply and fuel cell modules in marine hydrogen systems, and to calculate the hazard priority number of each component's failure mode in order to update the failure mode identification data. A nonlinear mapping function is constructed between the hazard priority number and the weights of the feature parameters. When the hazard priority number of a certain component exceeds a set threshold, the feature weights of the stress and pressure data of the hydrogen-related pipeline and the spatiotemporal distribution data of the hydrogen concentration in the cabin corresponding to that component are dynamically increased. During ship operation, the hazard priority number is updated in real time based on the cumulative operating time of the components and historical fault records, thereby completing the dynamic adaptive adjustment of the weight of the feature parameters.

5. The method for full life-cycle testing and evaluation of a marine hydrogen system according to claim 1, characterized in that, In step S3, the closed-space scenario of the lock passage is identified based on the scene status data of the ship, and a dynamic risk threshold is calculated using a safety zone theoretical model, specifically including: When a ship is detected to be entering a lock passage confined space based on the ship's current scene status data, the estimated time of the current lock passage process and the spatial environment parameters of the lock passage confined space are obtained. The hydrogen leakage rate in the preliminary feature characterization signal vector and the space environment parameters are input into a pre-constructed safe zone theoretical model to calculate the critical time required for the hydrogen concentration in the closed space to reach the lower explosive limit. Using the critical time being equal to the expected time as a safety boundary condition, the maximum allowable leakage rate threshold under the current scenario is calculated by reverse calculation. Based on the maximum allowable leakage rate threshold and leakage source parameters, calculate the diffusion distance when the hydrogen concentration decays to the lower explosive limit concentration, and use it as the safe distance threshold. The maximum allowable leakage rate threshold and the safe distance threshold are used together as the dynamic risk threshold.

6. The method for full life-cycle testing and evaluation of a marine hydrogen system according to claim 1, characterized in that, In step S3, the time-series prediction of the fuel cell lifespan degradation trend based on the fuel cell operating condition data specifically includes: The output voltage and current fluctuation data of the fuel cell stack are obtained, and the actual output load spectrum of the fuel cell stack under actual operating conditions is extracted. The actual output load spectrum is divided into working condition segments, and a multidimensional aging stress sequence containing start-stop cycle characteristics, dynamic load characteristics, and steady-state operation characteristics is extracted. The multidimensional aging stress sequence is input into a pre-trained fuel cell performance degradation time-series prediction model, which outputs the health status degradation sequence of the fuel cell stack within a future preset period. Based on the health state decay sequence, the initial polarization curve in the initial performance baseline data of the component is decayed and reconstructed to obtain the predicted polarization curve. Based on the predicted polarization curve and the preset failure voltage threshold, the voltage decay rate and remaining service life are calculated as the time life prediction.

7. The method for full life-cycle testing and evaluation of a marine hydrogen system according to claim 1, characterized in that, In S3, the specific calculation method for the full life cycle risk assessment result is as follows: in, This is a comprehensive risk assessment value covering the entire life cycle. To construct a space risk index, This is a time risk index. Indicates the basic weight of spatial risk. Indicates the basic weight of time risk. This represents the spatiotemporal coupling penalty factor, used to amplify the nonlinear risks associated with the superposition of spatial leakage and time aging.

8. The method for full life-cycle testing and evaluation of a marine hydrogen system according to claim 1, characterized in that, In step S4, based on the full lifecycle risk assessment results, tiered response and adaptive decision-making are executed, and full lifecycle closed-loop management instructions are output, specifically including: When the spatial risk distribution shows that the local hydrogen concentration exceeds the first-level warning threshold, a linkage ventilation and explosion-proof command is output to start the explosion-proof fan for forced ventilation. When the local hydrogen concentration exceeds the level 2 alarm threshold or when the stress and pressure data of the hydrogen-related pipeline show abnormal changes, an emergency cut-off and inerting command is output, the main valve of the hydrogen supply pipeline is automatically closed, and the automatic inerting and hydrogen replacement program is started to inject inert gas into the hydrogen-related pipeline and the compartment. When the time-life prediction results show that the performance degradation of the fuel cell stack or hydrogen storage container reaches the maintenance threshold, a preventive maintenance instruction is output. The preventive maintenance instruction includes generating a hydrogen embrittlement-resistant non-destructive testing plan for hydrogen-related pipelines, as well as a calibration strategy for refueling communication connectors and robotic arms.

9. A system for the full life-cycle testing and evaluation of marine hydrogen systems, characterized in that, include: The multi-parameter dataset construction unit is used to acquire static prior data and real-time dynamic monitoring data of marine hydrogen systems and construct a full life cycle multi-parameter input dataset. An adaptive feature extraction and processing unit is used to perform adaptive feature extraction and preprocessing on the full life cycle multi-parameter input dataset to obtain a preliminary feature characterization signal; wherein, based on the hydrogen diffusion evolution law data, key areas of hydrogen leakage diffusion are determined, the real-time dynamic monitoring data is spatiotemporally aligned and noise suppressed, and the feature parameters in the full life cycle multi-parameter input dataset are dynamically weighted based on failure mode identification data, and the preliminary feature characterization signal is output. A multi-dimensional risk coupling assessment unit is used to input the preliminary feature characterization signal into a multi-dimensional risk assessment model for dynamic coupling analysis to obtain the full life cycle risk assessment result. The dynamic coupling analysis includes: identifying the confined space scenario of the lock passage based on the ship's current scenario status data, and calculating the dynamic risk threshold using a safe zone theory model; and performing time-series prediction of the fuel cell lifespan degradation trend based on the fuel cell operating condition data, outputting the full life cycle risk assessment result that includes spatial risk distribution and temporal lifespan prediction. The hierarchical response closed-loop management unit is used to execute hierarchical response and adaptive decision-making based on the full life cycle risk assessment results, and output full life cycle closed-loop management instructions.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, The computer program is executed by a processor as described in any one of claims 1-8: a method for full life-cycle testing and evaluation of a marine hydrogen system.

Images