Hydrogen leakage concentration comprehensive dynamic evaluation method and device based on subjective and objective weights

Abstract

The application discloses a hydrogen leakage concentration comprehensive dynamic evaluation method and equipment based on subjective and objective weights, and the method comprises the following steps: collecting and time-aligning hydrogen concentration data of multiple monitoring points, pre-processing concentration signals and constraining duration, constructing and dimensionless processing an evaluation index matrix, defining a subjective time weight vector, establishing a time preference coefficient, solving the subjective time weight based on the maximum entropy principle, establishing an objective weight, fusing the subjective and objective weights, calculating a comprehensive weighted concentration, and establishing and outputting a grading early warning criterion; the technical problems that the evaluation method mainly uses a fixed threshold in hydrogen leakage monitoring and safety evaluation of a hydrogenation area of an existing hydrogenation station and is difficult to reflect dynamic evolution characteristics of a leakage process are solved, meanwhile, the technical problem that it is difficult to comprehensively quantify concentration information and time variation characteristics of multiple monitoring points is solved, a quantifiable, reproducible and suitable engineering application method path is provided for dynamic evaluation and early warning of hydrogen leakage risks of the hydrogenation area.

Description

Technical Field

[0001] This invention relates to the field of hydrogen energy infrastructure safety monitoring and risk assessment technology, specifically to a comprehensive dynamic assessment method and equipment for hydrogen leakage concentration based on subjective and objective weights. Background Technology

[0002] Against the backdrop of global efforts to address climate change and promote energy structure transformation, hydrogen energy is seen as a crucial support for achieving carbon neutrality. Hydrogen refueling stations serve as vital infrastructure for the large-scale application of hydrogen fuel cell vehicles, and their operational safety is critical to the development of the hydrogen energy industry. However, hydrogen's rapid diffusion, low minimum ignition energy, and wide flammability mean that leaks during storage, transportation, and refueling can easily create flammable cloud formations in localized areas, posing a significant threat to the safety of hydrogen refueling stations.

[0003] Current hydrogen refueling station hydrogen leakage monitoring mainly adopts a fixed threshold strategy, triggering an alarm when the concentration at a certain monitoring point reaches a preset value. However, the existing technology has the following drawbacks: (1) Dynamic evolution characteristics are difficult to capture. Fixed threshold methods use static concentration values ​​as criteria, which cannot reflect the dynamic characteristics of the leakage process over time and are difficult to capture the risk accumulation trend in the early stage of leakage, resulting in early warning response lagging behind risk development.

[0004] (2) Information from multiple monitoring points is difficult to quantify comprehensively. Existing methods usually make independent judgments based on the concentration values ​​of a single monitoring point, and lack a mechanism to quantify the concentration information from multiple monitoring points in conjunction with the characteristics of time changes. The extreme values ​​of a single monitoring point can easily dominate the overall risk judgment, leading to the one-sidedness of the early warning and an increase in the false alarm rate.

[0005] (3) Insufficient objectivity and robustness in assessment. Existing subjective weighting methods mostly rely on expert scoring or human setting, lacking objective constraints, and the subjective arbitrariness of the weight distribution form is relatively strong; although existing objective weighting methods can reflect the degree of data dispersion, they cannot reflect the characteristics of leakage risk evolution over time, and it is difficult to eliminate local extreme value interference, resulting in insufficient robustness of the early warning triggering strategy.

[0006] (4) The assessment strategy lacks adaptability. The time preference coefficient of the existing methods is usually set in a fixed manner, which cannot dynamically adjust the assessment focus period according to the real-time characteristics of the monitoring signal. It is difficult to adapt to the changes in sensor consistency and short-term fluctuation characteristics under different leakage scenarios, and the adaptability and stability of the assessment results are limited.

[0007] The maximum entropy principle can determine the most objective probability distribution under finite constraints. Introducing it into the time-weighted solution can satisfy time preference constraints while avoiding the subjective arbitrariness of manually setting weights. The entropy weight method automatically determines the weight of each monitoring point by utilizing the statistical dispersion of monitoring data, effectively eliminating local extreme value interference. By integrating the subjective time weight determined by the maximum entropy principle with the objective information weight determined by the entropy weight method, a comprehensive evaluation mechanism that takes into account both time evolution and data difference characteristics is established. Combined with graded thresholds and duration criteria, graded early warning information is output, providing a new approach to solving the technical bottleneck of dynamic assessment of hydrogen leakage in hydrogen refueling areas. Summary of the Invention

[0008] The purpose of this invention is to address the problems of existing assessment methods that rely primarily on fixed thresholds, making it difficult to reflect the dynamic evolution of the leakage process, and making it difficult to comprehensively quantify the concentration information from multiple monitoring points and the characteristics of time changes. The invention provides a comprehensive dynamic assessment method and device for hydrogen leakage concentration based on subjective and objective weights.

[0009] To achieve the above objectives, the technical solution adopted by the present invention is as follows: Firstly, a comprehensive dynamic assessment method for hydrogen leakage concentration based on subjective and objective weights is provided, including the following steps: Several monitoring points were set up in the hydrogen refueling area to obtain hydrogen concentration data from multiple monitoring points, and the data were time-aligned to construct a time series set of hydrogen concentration data from multiple monitoring points. The hydrogen concentration time series set is preprocessed by introducing a duration constraint rule; Based on the preprocessed hydrogen concentration time series set, evaluation indicators for characterizing the hydrogen leakage concentration change characteristics were selected, an evaluation indicator matrix was constructed and dimensionless processing was performed to obtain a standardized evaluation matrix. Define subjective time weights (vectors) and establish time preference coefficients; Based on the principle of maximum entropy, subjective time weights are solved with the goal of maximizing time information entropy, while satisfying the constraints of time preference coefficient and normalization. Based on the standardized evaluation matrix, the objective weights of each evaluation index are calculated using the entropy method. A product combination model is used to fuse subjective time weights and objective weights to obtain a comprehensive weight; Based on the comprehensive weight, the evaluation indicators at each time point are weighted and calculated to obtain a comprehensive weighted concentration sequence that changes over time. Based on the comprehensive weighted concentration sequence, combined with the grading threshold and duration criteria, graded early warning information is output.

[0010] Furthermore, the monitoring points are deployed by setting up several hydrogen concentration sensors in the hydrogen refueling area. Under a unified sampling period, the hydrogen concentration (hydrogen gas integral) data of each hydrogen concentration sensor is collected during the monitoring period. The moment when the hydrogen leakage signal is first identified is used as the reference starting point, and the hydrogen concentration data collected by all hydrogen concentration sensors are time-aligned.

[0011] Furthermore, the time preference coefficient is established according to the following model: , In the formula, T To assess the total number of time points within a time window, t This refers to the time node number. oh t Subjective time weights for the corresponding time points; l This is the time preference coefficient, reflecting the decision-maker's preference for information at different times. l The smaller the value, the more the decision-maker values ​​information at the current moment.

[0012] Furthermore, the model for solving the subjective time weight with the objective of maximizing the temporal information entropy is as follows: , , In the formula, T To assess the total number of time points within a time window, t This refers to the time node number. oh t Subjective time weights for the corresponding time points; l This represents the time preference coefficient.

[0013] Furthermore, the calculation process of the entropy method includes: Calculate the weight of the j-th evaluation indicator in the i-th evaluation object. : , , Calculate the information entropy of the j-th evaluation index. : , , Calculate the coefficient of variation for the j-th evaluation indicator. : , , Determine the objective weight of the j-th evaluation indicator : , , In the formula, Indicates the first i The evaluation object is in the first j Standardized values ​​for each evaluation indicator k Let m and n be constants used for entropy normalization.

[0014] Furthermore, the calculation formula for the product combination model is as follows: , In the formula, w c For the first c The overall weight of each evaluation indicator w sc For the first c The subjective weight of each evaluation indicator, w oc For the first c The objective weight of each evaluation indicator, It represents the sum of the products of the subjective and objective weights of all evaluation indicators.

[0015] Furthermore, the grading thresholds include a first warning threshold and a second warning threshold. The first warning threshold is 50% of the standard concentration threshold, and the second warning threshold is the standard concentration threshold. When the comprehensive weighted concentration sequence exceeds the first warning threshold for a continuous period of not less than a preset duration, a first-level warning is output. When the comprehensive weighted concentration sequence exceeds the second warning threshold for a continuous period of not less than the preset duration, a second-level warning is output.

[0016] Furthermore, this method also includes an online adaptive adjustment step for the time preference coefficient: Based on the spatial structure and ventilation conditions of the hydrogen refueling area, a base value for the time preference coefficient is pre-set. l base and upper and lower limits of values l min , l max ; Set the runtime layer update cycle and the width of the sliding time window; Based on data from multiple monitoring points within a sliding time window, the consistency score SCS(t) is calculated. The short-term volatility index GI(t) is calculated based on data from multiple monitoring points within a sliding time window. The time preference coefficient is adjusted online based on the consistency score and the short-term fluctuation index. The adjusted time preference coefficients are subjected to exponential smoothing and minimum residence time constraints to obtain the actual effective time preference coefficients.

[0017] Furthermore, the consistency score is calculated based on the Pearson correlation coefficient, which characterizes the degree of consistency in the hydrogen concentration change trends at different monitoring points; the short-term fluctuation index is calculated based on the standard deviation of the high-frequency components after detrending, which characterizes the fluctuation characteristics of the monitoring signal within a short time scale.

[0018] In a second aspect, an electronic device is provided, the electronic device including at least one processor and at least one memory connected to the processor; wherein the processor is used to call program instructions in the memory to execute the comprehensive dynamic assessment method for hydrogen leakage concentration based on subjective and objective weights as described above.

[0019] A computer-readable storage medium includes a stored program that, when executed by a processor, implements the comprehensive dynamic assessment method for hydrogen leakage concentration based on subjective and objective weights as described above.

[0020] Compared with existing technologies, the beneficial effects of this invention are: 1. Dynamic evolution characteristics can be quantified. Subjective time weights are solved based on the maximum entropy principle, avoiding the subjective arbitrariness of manually setting weight distributions, and enabling the comprehensive weighted concentration sequence to dynamically reflect the cumulative trend of leakage risk over time; 2. Multi-monitoring point information is comprehensively quantified. The objective weights of each monitoring point are automatically determined using the entropy method, eliminating interference from local extreme values. Subjective and objective weights are fused through a product combination model, avoiding the dominance of single monitoring point extreme values ​​in overall risk judgment; 3. Evaluation robustness is enhanced. Subjective weights are determined based on the maximum entropy principle under limited constraints to achieve the most objective distribution form, while objective weights are automatically calculated based on the data dispersion. The fusion of the two significantly improves the robustness of the early warning triggering strategy; 4. Early warning reliability is improved. The comprehensive weighted concentration is combined with graded thresholds and duration criteria to output early warnings, which are triggered only when the judgment conditions are continuously met, effectively avoiding false triggering of short-term spikes; 5. The evaluation strategy is adaptive. The time preference coefficient is adjusted online based on sensor consistency scores and short-term fluctuation indices, and the adjustment stability is ensured through exponential smoothing and minimum residence time constraints, making the evaluation method adaptable to different leakage scenarios. Attached Figure Description

[0021] Figure 1 This is a flowchart illustrating a comprehensive dynamic assessment method for hydrogen leakage concentration based on subjective and objective weights according to the present invention. Detailed Implementation

[0022] The technical solution of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] Example 1: A comprehensive dynamic assessment method for hydrogen leakage concentration based on subjective and objective weights is provided, combined with... Figure 1 As shown, it includes the following steps: S101: Several monitoring points are set up in the hydrogen refueling area to acquire hydrogen concentration data from multiple monitoring points and perform time alignment to construct a time series set of hydrogen concentration data from multiple monitoring points; this step mainly performs hydrogen concentration data acquisition and time alignment from multiple monitoring points. S102: Introduce duration constraint rules to preprocess the hydrogen concentration time series set; this step mainly involves preprocessing the concentration signal and imposing duration constraints. S103: Based on the preprocessed hydrogen concentration time series set, evaluation indicators for characterizing the hydrogen leakage concentration change characteristics are selected, an evaluation indicator matrix is ​​constructed and dimensionless processing is performed to obtain a standardized evaluation matrix. S104: Define subjective time weights (subjective time weight vector) and establish time preference coefficients; S105: Based on the principle of maximum entropy, subjective time weights are solved with the goal of maximizing time information entropy, under the conditions of satisfying the constraints of time preference coefficient and normalization. S106: Based on the standardized evaluation matrix, calculate the objective weight of each evaluation index using the entropy method; S107: The subjective time weight and the objective weight are integrated using a product combination model to obtain a comprehensive weight; S108: Based on the comprehensive weight, the evaluation indicators at each time point are weighted and calculated to obtain a comprehensive weighted concentration sequence that changes over time; S109: Based on the comprehensive weighted concentration sequence, combined with the grading threshold and duration criterion, output graded early warning information.

[0024] The purpose of this comprehensive dynamic assessment method for hydrogen leakage concentration is to construct a subjective time-weighted model and an objective information-weighted model, and then integrate the two to form a comprehensive weighted concentration assessment result that changes over time. This result is then combined with a grading threshold and duration criterion to output graded early warning information with engineering significance, thereby meeting the needs of early identification and risk grading response for hydrogen leakage in hydrogen refueling areas.

[0025] The method described in this invention addresses the characteristics of hydrogen refueling areas being semi-enclosed spaces, where hydrogen leakage and diffusion processes are significantly affected by ventilation conditions and environmental disturbances, and where concentration distribution exhibits obvious spatiotemporal nonuniformity. It utilizes time series data of hydrogen concentration from multiple monitoring points to dynamically characterize and comprehensively assess leakage risks.

[0026] The subjective time weight model is solved based on the maximum entropy principle. Under the conditions of satisfying the time preference coefficient constraint and normalization constraint, it determines the time weight distribution with the goal of maximizing the time information entropy, avoiding the subjective arbitrariness of manually setting the weight distribution form and characterizing the attention characteristics of leakage risk evolving over time. The objective information weight model is calculated based on the entropy weight method. It automatically determines the weight of each monitoring point by utilizing the statistical dispersion of the monitoring data, eliminating local extreme value interference, enhancing the robustness of the early warning triggering strategy, and reflecting the differences in objective information contained in different evaluation indicators.

[0027] By employing a product-combination model to fuse the subjective and objective weights, a comprehensive weight is obtained, establishing a comprehensive evaluation mechanism that simultaneously considers temporal evolution characteristics and data difference characteristics. Based on this comprehensive weight, the evaluation indicators at each time point are weighted and calculated to obtain a comprehensive weighted concentration sequence that changes over time. This comprehensive weighted concentration sequence can dynamically capture the evolution trend of risk over time, avoiding the dominance of extreme values ​​at a single monitoring point in the overall risk assessment, while also reflecting the dynamic characteristics of hydrogen diffusion under different leakage intensities, leakage directions, and environmental conditions.

[0028] Based on this, an early warning output mechanism is established by combining graded thresholds and duration criteria. When the comprehensive weighted concentration exceeds the corresponding threshold for a continuous period of no less than the preset duration, graded early warning information is output, which effectively avoids false triggering caused by short-term spikes and improves the reliability and engineering applicability of the early warning.

[0029] Furthermore, in step S101, the monitoring points are deployed by setting up several hydrogen concentration sensors in the hydrogen refueling area, collecting hydrogen concentration data (hydrogen gas integral data) from each hydrogen concentration sensor during the monitoring period under a unified sampling period, and using the moment when the hydrogen leak signal is first identified as a reference starting point to time-align the hydrogen concentration data collected by all hydrogen concentration sensors.

[0030] Specifically, let the number of hydrogen concentration sensors be N. Under a uniform sampling period, the integral data of hydrogen gas concentration from each hydrogen concentration sensor during the monitoring period are collected. Let the hydrogen concentration monitored by the i-th hydrogen concentration sensor at time t be... x i ( t ), in, i = 1, 2, ..., N ,t = 1, 2, …, T , T To evaluate the number of discrete time points within a time window; Using the moment when the hydrogen leak signal was first detected as a reference starting point, the data collected by all hydrogen concentration sensors were time-aligned to construct a time series dataset of hydrogen concentrations at multiple monitoring points. {xi(t)} .

[0031] Further, in step S102, the hydrogen concentration time series obtained in step S101 is preprocessed to reduce the impact of noise and instantaneous fluctuations on the evaluation results. Let the smoothed concentration series be... , Based on this, a duration constraint rule is introduced; the duration constraint rule is that a judgment condition is applied only if the duration is continuous for a period of not less than a preset duration. t The condition is considered valid only if it continues to be true under certain conditions, thus avoiding false triggering caused by short-term spikes.

[0032] Furthermore, in step S103, the constructed evaluation index matrix As shown below: , in, m For the number of evaluation objects, n To evaluate the number of indicators, x ij Indicates the first i The evaluation object is in the first j The evaluation index matrix is ​​then processed to obtain a standardized evaluation matrix for subsequent weight calculations.

[0033] Standardized evaluation matrix It is expressed as follows: , y ij Indicates the first i The evaluation object is in the first j The standardized values ​​of each evaluation indicator.

[0034] Furthermore, in step S104, assuming the evaluation time window contains T time nodes, the subjective time weight vector is defined as follows: , in, oh t Indicates the first t The time weight of each moment in the comprehensive evaluation.

[0035] ,and .

[0036] To characterize the differences in the degree of attention paid to data in different time periods during the evaluation process, a time preference coefficient is introduced, which is established according to the following model: , In the formula, T To assess the total number of time points within a time window, t This refers to the time node number. oh t Subjective time weights for the corresponding time points; l This is the time preference coefficient, reflecting the decision-maker's preference for information at different times. l The smaller the value, the more the decision-maker values ​​information at the current moment.

[0037] Furthermore, in step S105, to avoid artificially setting the weight distribution while satisfying time preference constraints, an information entropy function for time weights is introduced. The model for solving subjective time weights with the goal of maximizing time information entropy is as follows: , , In the formula, T To assess the total number of time points within a time window, t This refers to the time node number. oh t Subjective time weights for the corresponding time points; l This represents the time preference coefficient.

[0038] Under the conditions of satisfying the normalization constraint of subjective time weight and the constraint of time preference coefficient, a subjective time weight solution model is constructed with the goal of maximizing time information entropy, and the subjective time weight vector is obtained by solving the model. oh .

[0039] Further, in step S106, the calculation process of the entropy method includes: Based on the standardized evaluation matrix Y obtained in step S103, calculate the weight of the j-th evaluation index in the ith evaluation object. : , , Calculate the information entropy of the j-th evaluation index. : , , In the formula, kLet m and n be constants used for entropy normalization, where m and n are integers. According to information entropy Calculate the difference coefficient of the j-th evaluation index. : , , Determine the objective weight of the j-th evaluation indicator : , .

[0040] Further, in step S107, the calculation formula for the product combination model is: , In the formula, w c For the first c The overall weight of each evaluation indicator w sc For the first c The subjective weight of each evaluation indicator, w oc For the first c The objective weight of each evaluation indicator, This represents the sum of the products of the subjective and objective weights of all evaluation indicators, used for normalization.

[0041] Based on the comprehensive weights obtained in step S107, the evaluation indicators at each time point are weighted and calculated to obtain the comprehensive weighted concentration: , In the formula, W c (t) This represents the weighted concentration value at time t. y j (t) Indicates time t Next j Standardized values ​​for each evaluation indicator.

[0042] This forms a comprehensive dynamic evaluation sequence that changes over time. {W c (t)} .

[0043] Further, in step S109, the grading threshold includes a first warning threshold and a second warning threshold, wherein the first warning threshold is 50% of the standard concentration threshold, and the second warning threshold is the standard concentration threshold; when the comprehensive weighted concentration sequence exceeds the first warning threshold for a continuous period of not less than a preset duration, a first-level warning is output; when the comprehensive weighted concentration sequence exceeds the second warning threshold for a continuous period of not less than the preset duration, a second-level warning is output.

[0044] Specifically, the standard concentration threshold is set as follows: C std And based on this, set graded early warning thresholds. When the comprehensive weighted concentration W c (t) If the first warning threshold is exceeded for a continuous duration of at least τ, a Level 1 warning is output; when the comprehensive weighted concentration... W c (t) If the second warning threshold is exceeded for a duration of not less than τ, a level 2 warning is output.

[0045] Furthermore, this method also includes an online adaptive adjustment step for the time preference coefficient: Based on the spatial structure and ventilation conditions of the hydrogen refueling area, a base value for the time preference coefficient is pre-set. l base and upper and lower limits of values l min , l max ; Set the runtime layer update cycle and the width of the sliding time window; Based on data from multiple monitoring points within a sliding time window, the consistency score SCS(t) of the monitoring points (hydrogen concentration sensor) is calculated. Based on data from multiple monitoring points within a sliding time window, the short-term fluctuation index GI(t) of the monitoring point (hydrogen concentration sensor) is calculated. The time preference coefficient is adjusted online based on the consistency score and the short-term fluctuation index. The adjusted time preference coefficients are subjected to exponential smoothing and minimum residence time constraints to obtain the actual effective time preference coefficients.

[0046] The consistency score is calculated based on the Pearson correlation coefficient, which characterizes the degree of consistency in the hydrogen concentration change trend at different monitoring points; the short-term fluctuation index is calculated based on the standard deviation of the high-frequency component after detrending, which characterizes the fluctuation characteristics of the monitoring signal within a short time scale.

[0047] Example 2: This example selects a typical hydrogen leakage scenario in the hydrogen refueling area of ​​a hydrogen refueling station to verify and explain the comprehensive dynamic assessment method of hydrogen leakage concentration based on subjective and objective weights in Example 1.

[0048] Step 1: Construct the original data matrix X ij In this embodiment, a scaled-down experimental model of the hydrogen refueling area of ​​a hydrogen refueling station was constructed to simulate the diffusion and concentration evolution process of hydrogen leakage in an actual hydrogen refueling station. The experimental model was scaled to a 1:4 ratio, simplifying and replicating key structures and shielding structures in the hydrogen refueling area while maintaining the consistency of the main structural layout and spatial constraints, to simulate the hydrogen diffusion characteristics under semi-enclosed space conditions in a real hydrogen refueling station. High-purity hydrogen was used as the leakage medium, with a purity of not less than 99.99%. The leakage occurred through a circular nozzle located within the hydrogen refueling area, with a nozzle diameter of 8 mm. The leakage flow rate was precisely controlled by the pressurized gas source and gas distribution system. To balance physical representativeness and experimental safety, four nominal leakage flow rate conditions were set up in the experiment, each with a flow rate of 5000... mL / min 10000 mL / min 15000 mL / min and 20000 mL / min .

[0049] Under 1:4 scaling and environmental conditions, the above leakage flow rates cover the typical operating conditions transitioning from buoyancy-dominated diffusion to diffusion significantly influenced by momentum. Specifically, the lowest leakage flow rate corresponds to a weak leakage situation dominated by buoyancy, representing a low-intensity leakage caused by minor failures in the sealing or connection parts of the hydrogen refueling equipment, with far-field hydrogen concentrations typically below the alarm threshold. The medium leakage flow rate gradually increases the initial momentum of the leakage, exhibiting a diffusion state influenced by both buoyancy and momentum, and forming more pronounced backflow and accumulation characteristics below the upper structure of the hydrogen refueling area. The highest leakage flow rate is close to the maximum allowable discharge capacity of the experimental device at this scale under safe conditions, used to simulate boundary leakage situations where momentum effects dominate. These four leakage flow rate conditions, while ensuring the safe operation of the experimental device, continuously cover the variation range of common leakage severity in hydrogen refueling stations, enabling the acquired concentration time series data to reflect the dynamic characteristics of hydrogen diffusion under different leakage intensities, thus providing a data foundation for the applicability and stability of the comprehensive dynamic assessment method.

[0050] During the experiment, four calibrated hydrogen concentration sensors were deployed in the hydrogen refueling area to synchronously collect data on changes in hydrogen concentration during the leakage process, constructing a raw data matrix. X ijThe experiment was conducted under repeatable and controlled conditions, and environmental parameters, including ambient temperature, atmospheric pressure, wind speed, and atmospheric stability, were recorded during the experiment. To facilitate engineering application analysis, the surface roughness parameters of the experimental site were estimated to characterize the influence of surrounding structures on airflow organization and diffusion behavior. Based on the above experimental setup, a dataset of hydrogen concentration variations over time at multiple monitoring points under different leakage conditions was obtained.

[0051] Step 2: Subjective Weight Calculation W sc With objective weight calculation W oc Based on the original data matrix constructed in step one above. X ij Subjective weight calculation and objective weight calculation are performed in parallel.

[0052] Subjective weighting is calculated using a time-preference information entropy model, reflecting the information contribution of concentration changes over time. Multiple time preference coefficient values ​​are selected for comparative calculations based on the assessment time window length and leakage duration characteristics. l Values ​​of 0.1, 0.2, 0.3, and 0.4 were used to reflect the varying degrees of attention paid to monitoring data across different time periods during the evaluation process. Under these parameter settings, different... l Subjective time weights under given value conditions.

[0053] The objective weights are calculated using the information entropy method, reflecting the concentration differences between each sensor. The concentration time-series data collected by each sensor are standardized, and the information entropy, redundancy, and objective weights corresponding to each evaluation index are calculated.

[0054] Step 3: Calculate the combined weights W c The subjective weights obtained in step two W sc With objective weight W oc By merging, we obtain the combined weights. W c This combines subjective time information with objective difference information. By comparing different... l The distribution characteristics of the combined weights under different values ​​show that the time preference coefficient has a significant impact on the allocation of combined weights in the time series, thus providing different emphase strategies for the dynamic evaluation of the subsequent comprehensive weighted concentration.

[0055] Step 4: Weighted Concentration Calculation C j Based on the combined weights obtained in step threeW c The hydrogen concentration data from multiple monitoring points at different time points are weighted and calculated to obtain the comprehensive weighted concentration. C j A comprehensive dynamic assessment sequence is formed that changes over time, serving as the final comprehensive hydrogen concentration index. In each experimental scenario, the comprehensive weighted concentration shows a continuous increasing trend with leakage time, reaching a peak at the end of the leakage duration phase; after the leakage source is closed, the comprehensive weighted concentration decreases rapidly over time. By adjusting the time preference coefficient, the peak position and rate of change of the comprehensive weighted concentration in the time series can be changed accordingly, thereby reflecting the risk evolution characteristics under different assessment emphasis strategies.

[0056] To further analyze the impact of hydrogen leakage direction on the diffusion morphology, concentration distribution, and temporal evolution of hydrogen jets within the hydrogen refueling area, various typical leakage direction scenarios were set up under the aforementioned experimental platform and scale conditions, and comparative experiments were conducted under the same leakage intensity conditions. Specifically, using the hydrogen refueling station's hydrogen dispenser as the leakage source location, four different leakage direction scenarios were set up within the hydrogen refueling area. Scenario 1 was a free leakage scenario along the direction directly in front of the hydrogen dispenser, where the hydrogen jet was ejected along an open direction within the hydrogen refueling area, without being directly obstructed by surrounding structures. Scenario 2 was a leakage scenario where the hydrogen jet was directed towards one side of a building, used to simulate the diffusion and backflow characteristics of the leaking jet when it encounters a solid building structure at close range. Scenario 3 was a leakage scenario where the hydrogen jet was directed towards an adjacent hydrogen dispenser, used to reflect the diffusion behavior under conditions of mutual obstruction between equipment. Scenario 4 was a leakage scenario where the hydrogen jet was directed towards the top canopy of the hydrogen refueling area, used to simulate the situation where the jet directly impacts the upper structure and accumulates and backflows below it.

[0057] In the four leakage direction scenarios described above, the leakage flow rate was set to 15000 mL / min to ensure comparability between different experimental scenarios. This leakage flow rate corresponds to the typical operating condition under the combined effects of buoyancy and momentum in the aforementioned proportional model, and can adequately reflect the dynamic characteristics of hydrogen diffusion under common moderate-intensity leakage conditions at hydrogen refueling stations. In each experiment, the continuous hydrogen leakage time was set to 240 s, and hydrogen concentrations at multiple monitoring points were simultaneously collected throughout the leakage process. By keeping the leakage flow rate, leakage duration, and environmental conditions consistent, and only changing the leakage direction setting, the influence of different spatial constraints within the hydrogen refueling area on the spatiotemporal evolution characteristics of hydrogen concentration can be systematically investigated. It should be noted that although the leakage velocity and direction do not change the order in which hydrogen arrives at each sensor, they significantly affect the arrival time and concentration rise rate at each monitoring point, and the hazard of hydrogen is closely related to the concentration rise rate. Therefore, systematically studying the concentration evolution law under different leakage direction conditions is of great significance for establishing an assessment method that can dynamically capture the spatiotemporal characteristics of risk, and provides representative input data for subsequent comprehensive dynamic assessment methods based on multi-monitoring point concentration data.

[0058] Based on the time series data of hydrogen concentration obtained from multiple monitoring points under different leakage direction conditions, the subjective weight, objective weight, and combined weight are calculated according to steps one to four of this invention, and a dynamic evaluation result of the comprehensive weighted concentration is formed. In the specific implementation process, numerical calculation software is used to solve for the subjective time weight, objective information weight, and their fused combined weight.

[0059] Significant differences exist in the overall weighted concentration obtained at the same time point across different leakage direction scenarios, indicating that the overall weighted concentration effectively reflects the impact of environmental conditions, spatial constraints, and changes in leakage direction on the hydrogen diffusion process. In this embodiment, a leakage direction scenario where the hydrogen concentration sensor deployment path is relatively continuous and the jet diffusion path is less affected by obstruction is selected as a representative operating condition. The characteristics of the combined weight changing with the leakage duration are further analyzed to illustrate the applicability of the method of the present invention under typical operating conditions.

[0060] Based on the changing characteristics of the hydrogen jet path under different leakage directions, hydrogen concentration sensors in the hydrogenation zone were strategically deployed. In scenarios where the leaking jet was not significantly obstructed by structures, multiple sensors were arranged sequentially along the leak centerline to acquire information on hydrogen concentration changes along the main diffusion path. In scenarios where the leaking jet was affected by buildings or roof structures, the sensors were symmetrically deployed relative to the leak centerline to reflect the diffusion and accumulation characteristics of hydrogen under space-constrained conditions. Each sensor synchronously collected hydrogen concentration data throughout the entire leakage period, forming a multi-monitoring-point time-series dataset. Based on this sensor deployment, hydrogen concentration time-series data were acquired under different leakage directions at a leakage flow rate of 15000 mL / min and a leakage duration of 240 s. The data was preprocessed, time-aligned, and dimensionless, and then used as input data for a comprehensive dynamic evaluation model.

[0061] Regarding parameter settings, considering that the hydrogen refueling area is a semi-enclosed space and the risk of leakage accumulates and increases over time, a time preference coefficient was selected as an adjustment parameter to reflect the change in assessment focus over time. Its value range was set to be greater than 0.2, and comparative calculations under different value conditions were used to verify the response characteristics of the comprehensive assessment results to changes in time weight. In the comprehensive dynamic assessment process, firstly, the objective weights corresponding to each evaluation indicator were calculated according to step two. Then, the subjective time weights and objective weights were merged according to step three to obtain the combined weights. Subsequently, according to step four, the hydrogen concentration data from multiple monitoring points at each time node were weighted and calculated to obtain the assessment sequence of the comprehensive weighted concentration changing over time.

[0062] During the early warning output phase, based on engineering application requirements, a tiered threshold and duration criterion are introduced. A standard concentration threshold is set as the corresponding value of a predetermined safety standard, and based on this, the first early warning threshold is set to 50% of the standard concentration threshold. A continuous time determination is performed on the comprehensive weighted concentration sequence; when the comprehensive weighted concentration remains within a range of at least a preset duration... t When the overall weighted concentration exceeds the first warning threshold, a level one warning signal is output; when the overall weighted concentration remains above the threshold for a duration not less than the stated duration... t When the concentration exceeds the standard threshold, a secondary warning signal is output. Through the above implementation process, under different leakage directions and spatial constraints, the comprehensive weighted concentration sequence can reflect the dynamic changes in the risk level during hydrogen leakage and can output time-oriented graded warning information in the early stages of leakage. The comprehensive weighted concentration, as a unified assessment index, can be used to characterize the overall diffusion risk of hydrogen leakage in the hydrogenation area and provide a basis for initiating subsequent safety response measures.

[0063] As can be seen from the above implementation process, while maintaining consistency in leakage intensity, leakage duration, and environmental conditions, the method described in this invention can flexibly control the time period emphasized in hydrogen leakage risk assessment by adjusting the time preference coefficient, and generate a comprehensive dynamic assessment result with clear time orientation. This method avoids single monitoring point extreme values ​​dominating the overall risk judgment, dynamically captures the evolution trend of risk over time, enhances the objectivity of risk assessment and the timeliness of early warning response, and provides basic data for the subsequent application of graded early warning criteria.

[0064] Example 3: Based on the same inventive concept, this example further presents an online adaptive adjustment method for the time preference coefficient, building upon the experimental data from Example 2. l The implementation method. Except l Except for the change of the value retrieval method from a fixed setting to online updating, the remaining steps are the same as steps S101 to S109 described in Example 1.

[0065] I. Data Sources and Inputs In this embodiment, the time series data of hydrogen concentration at multiple monitoring points obtained in Example 1 is selected as the input data. The input data is generated by synchronously collecting data from multiple hydrogen concentration sensors deployed in the hydrogen refueling area, denoted as... x i ( t ), In the formula, i Indicates the sensor number, t Represents a discrete time point.

[0066] The concentration time series is time-aligned, smoothed, and dimensionless according to steps S101 to S103 to obtain a preprocessed data series for online evaluation. This is then used as input for subsequent adaptive parameter calculations and comprehensive dynamic evaluation.

[0067] II. A Two-Layer Framework for Obtaining the Time Preference Coefficient λ To balance the stability of project deployment with the flexibility of operation, this embodiment adopts a two-layer acquisition framework that combines site-level initialization and operation-level online adjustment to determine the time preference coefficient λ.

[0068] Step S201: Setting the base value of the time preference coefficient at the site level; Based on the spatial structure and ventilation conditions of the hydrogen refueling area, a base value for the time preference coefficient is pre-selected. l base In this embodiment, the hydrogen refueling area is a semi-enclosed space. l base Select a value range that is not lower than 0.2.

[0069] At the same time, upper and lower limits are set for the value of λ, respectively. min =0.20, λ max =0.30; used to constrain the range of λ change during online adjustment, ensuring it remains within a pre-defined safe range.

[0070] Step S202: Fine-tune the update cycle and window width settings in the runtime layer; Set the runtime layer update cycle to one second, and set the sliding time window width to [value missing]. W Seconds are used to calculate sensor consistency and short-term fluctuation metrics within a window. Window length. W The time can be set to either 10 seconds or 20 seconds to meet the requirements of real-time performance and stability.

[0071] III. Construction of Online Monitoring Indicators at the Operational Layer Step S203: Calculate the consistency score of the hydrogen concentration sensor SCS(t) ; At each time point t, the four sensor sequences within the interval [t−W+1, t] are taken, and the Pearson correlation coefficient ρ between any two hydrogen concentration sensors i and j is calculated. ij (t;W), its calculation formula is: , In the formula, and These represent the average values ​​of hydrogen concentration sensors i and j within the window, respectively.

[0072] The consistency score of the hydrogen concentration sensors is obtained by averaging the pairwise correlation coefficients of the four hydrogen concentration sensors and limiting them to the range of zero to one. , In the formula, clip ( x , a , b ) indicates that the variable x It is restricted to the interval [a, b]. SCS The value range is from zero to one. The consistency index is used to characterize the degree of consistency in the trend of hydrogen concentration changes at different monitoring points, and its value range is limited to zero to one.

[0073] Step S204: Calculate the short-term volatility index GI(t) ; To characterize the short-term fluctuation intensity, the detrended high-frequency components are calculated on each sensor sequence. The preprocessed sequence for hydrogen concentration sensor i is shown below. The detrending sequence is obtained according to the formula: , In the formula, This represents a 10-second moving average.

[0074] In the window [ tW +1, t Internal calculation r i ( t Standard deviation The fluctuation intensities of the four hydrogen concentration sensors were summarized and normalized from zero to one to obtain the short-term fluctuation index. Within the same sliding time window, the concentration sequences of each hydrogen concentration sensor were detrended to extract the short-term fluctuation components and calculate their fluctuation intensity.

[0075] , In the formula, ref As a reference scale for normalization, robust statistics can be provided from the background segment or historical window, making... GI The value range is consistently between zero and one. The short-time fluctuation index is used to characterize the fluctuation characteristics of the monitoring signal within a short time scale, and its value range is also limited to zero and one.

[0076] IV. Online Adjustment and Stability Constraints of Time Preference Coefficient Step S205: Original update of λ based on monitoring indicators; At each point in time, based on the consistency index SCS(t) With short-term volatility indicators GI(t) The time preference coefficient is adjusted online to obtain the original updated value of λ: , In the formula, λ(t-1) The time preference coefficient of the previous moment; α This represents the consistency adjustment gain with respect to λ; β This represents the adjustment gain of the fluctuation with respect to λ; l min and l max They are respectively l The lower and upper limits.

[0077] Specifically, when the monitoring signal consistency is high and short-term fluctuations are weak, the value of λ is increased to enhance the focus on data from the early stages of leakage; when consistency decreases or short-term fluctuations increase, the value of λ is decreased to enhance the robustness of the assessment results to instantaneous fluctuations. The updated λ is limited by the aforementioned upper and lower limits [ l min , l max ].

[0078] Step S206: Exponential smoothing and minimum dwell time constraint; To avoid frequent changes in λ within a short period of time, l raw (t) Smoothing is performed to obtain the actual effective time preference coefficient. : , In the formula, c As a smoothing coefficient, this embodiment takes... c =0.2; simultaneously set the minimum dwell time. D In this embodiment, D =10 seconds, used to limit the adjustment direction of λ to remain consistent during the dwell time, thereby reducing the impact of short-term jitter on λ update.

[0079] At the same time, a minimum dwell time constraint is set to ensure that the adjustment direction of λ remains consistent during the dwell time, thereby improving the stability during online operation.

[0080] V. Adaptive λ-substitution comprehensive dynamic evaluation and early warning output Step S207: Calculation of comprehensive weighted concentration; At each time point, the updated time preference coefficient will be... Substitute the subjective time weights from steps S104 to S106 into the calculation process, and merge them with the objective weights obtained in step S107 according to step S108 to obtain the comprehensive weight. Then, calculate the comprehensive weighted concentration according to step S109 to form a comprehensive dynamic evaluation sequence that changes over time.

[0081] Step S208: Output of graded early warning criteria; Based on the comprehensive weighted concentration sequence, the same dual threshold and duration criteria as in Example 1 are used for judgment; when the comprehensive weighted concentration exceeds the first threshold for a continuous period of not less than a preset duration, a first-level warning signal is output; when the comprehensive weighted concentration exceeds the second threshold for a continuous period of not less than the preset duration, a second-level warning signal is output.

[0082] VI. Description of the operational characteristics of this embodiment Through the above implementation process, the online adaptive acquisition of the time preference coefficient λ was achieved without changing the overall structure of the comprehensive dynamic assessment method. This adaptive acquisition method relies solely on existing multi-monitoring point concentration time series data, requires no additional hardware, and can directly substitute the updated λ into the comprehensive assessment process during online operation, thereby outputting graded early warning results with clear time orientation.

[0083] Example 4: An electronic device is provided, the electronic device including at least one processor and at least one memory connected to the processor; wherein, the processor is used to call program instructions in the memory to execute the comprehensive dynamic assessment method for hydrogen leakage concentration based on subjective and objective weights as described above.

[0084] A computer-readable storage medium includes a stored program that, when executed by a processor, implements the comprehensive dynamic assessment method for hydrogen leakage concentration based on subjective and objective weights as described above.

[0085] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A comprehensive dynamic assessment method for hydrogen leakage concentration based on subjective and objective weights, characterized in that, Includes the following steps: Several monitoring points were set up in the hydrogen refueling area to obtain hydrogen concentration data from multiple monitoring points, and the data were time-aligned to construct a time series set of hydrogen concentration data from multiple monitoring points. The hydrogen concentration time series set is preprocessed by introducing a duration constraint rule; Based on the preprocessed hydrogen concentration time series set, evaluation indicators for characterizing the hydrogen leakage concentration change characteristics were selected, an evaluation indicator matrix was constructed and dimensionless processing was performed to obtain a standardized evaluation matrix. Define subjective time weights and establish time preference coefficients; Based on the principle of maximum entropy, subjective time weights are solved with the goal of maximizing time information entropy, while satisfying the constraints of time preference coefficient and normalization. Based on the standardized evaluation matrix, the objective weights of each evaluation index are calculated using the entropy method. A product combination model is used to fuse subjective time weights and objective weights to obtain a comprehensive weight; Based on the comprehensive weight, the evaluation indicators at each time point are weighted and calculated to obtain a comprehensive weighted concentration sequence that changes over time. Based on the comprehensive weighted concentration sequence, combined with the grading threshold and duration criteria, graded early warning information is output.

2. The comprehensive dynamic assessment method for hydrogen leakage concentration based on subjective and objective weights according to claim 1, characterized in that, The monitoring points are deployed by setting up several hydrogen concentration sensors in the hydrogen refueling area. Under a unified sampling period, the hydrogen concentration (hydrogen gas integral) data of each hydrogen concentration sensor is collected during the monitoring period. The moment when the hydrogen leakage signal is first identified is used as the reference starting point, and the hydrogen concentration data collected by all hydrogen concentration sensors are time-aligned.

3. The comprehensive dynamic assessment method for hydrogen leakage concentration based on subjective and objective weights according to claim 1, characterized in that, The time preference coefficient is established according to the following model: ， In the formula, T To assess the total number of time points within a time window, t This refers to the time node number. ω t Subjective time weights for the corresponding time points; λ This represents the time preference coefficient.

4. The comprehensive dynamic assessment method for hydrogen leakage concentration based on subjective and objective weights according to claim 1, characterized in that, The model for solving subjective time weights with the objective of maximizing time information entropy is as follows: ， ， In the formula, T To assess the total number of time points within a time window, t This refers to the time node number. ω t Subjective time weights for the corresponding time points; λ This represents the time preference coefficient.

5. The comprehensive dynamic assessment method for hydrogen leakage concentration based on subjective and objective weights according to claim 1, characterized in that, The calculation process of the entropy method includes: Calculate the first j The evaluation index is in the first i The proportion of each evaluation object : ， ， Calculate the first j Information entropy of each evaluation indicator : ， ， Calculate the first j Coefficient of difference of each evaluation indicator : ， ， Determine the first j Objective weights of each evaluation indicator : ， ， In the formula, Indicates the first i The evaluation object is in the first j Standardized values ​​for each evaluation indicator k Let m and n be constants used for entropy normalization.

6. The comprehensive dynamic assessment method for hydrogen leakage concentration based on subjective and objective weights according to claim 1, characterized in that, The calculation formula for the product combination model is as follows: ， In the formula, w c For the first c The overall weight of each evaluation indicator w sc For the first c The subjective weight of each evaluation indicator, w oc For the first c The objective weight of each evaluation indicator, It represents the sum of the products of the subjective and objective weights of all evaluation indicators.

7. The comprehensive dynamic assessment method for hydrogen leakage concentration based on subjective and objective weights according to claim 1, characterized in that, The grading thresholds include a first warning threshold and a second warning threshold. The first warning threshold is 50% of the standard concentration threshold, and the second warning threshold is the standard concentration threshold. When the comprehensive weighted concentration sequence exceeds the first warning threshold for a continuous period of not less than a preset duration, a first-level warning is output. When the comprehensive weighted concentration sequence exceeds the second warning threshold for a continuous period of not less than the preset duration, a second-level warning is output.

8. The comprehensive dynamic assessment method for hydrogen leakage concentration based on subjective and objective weights according to claim 1, characterized in that, It also includes an online adaptive adjustment step for the time preference coefficient: Based on the spatial structure and ventilation conditions of the hydrogen refueling area, a base value for the time preference coefficient is pre-set. λ base and upper and lower limits of values λ min , λ max ; Set the runtime layer update cycle and the width of the sliding time window; A consistency score is calculated based on data from multiple monitoring points within a sliding time window. SCS(t) ; Short-term volatility index is calculated based on data from multiple monitoring points within a sliding time window. GI(t) ; The time preference coefficient is adjusted online based on the consistency score and the short-term fluctuation index. The adjusted time preference coefficients are subjected to exponential smoothing and minimum residence time constraints to obtain the actual effective time preference coefficients.

9. The comprehensive dynamic assessment method for hydrogen leakage concentration based on subjective and objective weights according to claim 8, characterized in that, The consistency score is calculated based on the Pearson correlation coefficient, which characterizes the degree of consistency in the hydrogen concentration change trend at different monitoring points; the short-term fluctuation index is calculated based on the standard deviation of the high-frequency component after detrending, which characterizes the fluctuation characteristics of the monitoring signal within a short time scale.

10. An electronic device, characterized in that, The electronic device includes at least one processor and at least one memory connected to the processor; wherein the processor is used to call program instructions in the memory to execute the comprehensive dynamic assessment method for hydrogen leakage concentration based on subjective and objective weights as described in any one of claims 1 to 9.

Images