Method for intelligent monitoring and early warning of hydrogen storage pressure vessel based on WCS algorithm optimization

Abstract

This invention relates to the field of hydrogen storage safety monitoring technology. It provides a method for intelligent monitoring and early warning of hydrogen storage pressure vessels based on the WCS algorithm optimization. The method includes: collecting and storing full lifecycle data of hydrogen storage pressure vessels of the monitored category to form an original historical database; collecting and storing 6-dimensional dynamic data of in-service hydrogen storage pressure vessels to be monitored in real time; calculating the weighted cosine similarity between the real-time collected 6-dimensional dynamic data and historical benchmarks; selecting data groups of similar states from the historical database; and using a sliding window algorithm to calculate the rate of deviation change for comprehensive trend analysis; performing graded early warning and response based on the weighted cosine similarity comparison results; and simultaneously recording the current early warning record, 6-dimensional real-time feature vector, weighted cosine similarity matching results, and deviation analysis data into the original historical database. This invention solves the core problems of existing similarity matching algorithms, such as insensitivity to key parameters, high false alarm rate, and low early warning accuracy.

Description

Technical Field

[0001] This invention relates to the field of hydrogen storage safety monitoring technology, and in particular to a method for intelligent monitoring and early warning of hydrogen storage pressure vessels based on WCS algorithm optimization. Background Technology

[0002] Hydrogen storage pressure vessels are core equipment in hydrogen energy storage and transportation systems, with design pressures up to 140 MPa and operating temperatures ranging from -40°C to 65°C. The operational status of these vessels directly determines the safety and reliability of the entire hydrogen storage system. Therefore, accurate monitoring and timely early warning of their operational status have become crucial for ensuring the safe and stable operation of hydrogen energy infrastructure.

[0003] In practical monitoring, metrics such as Euclidean distance and Manhattan distance are easily affected by differences in the dimensions of different parameters, which may lead to false alarms. While standard cosine similarity can measure the consistency of vector directions, it fails to distinguish the importance of different monitoring parameters in safety assessment and cannot reflect the differentiated contributions of parameters such as pressure, temperature, and strain to fault characterization. Pearson correlation coefficient mainly focuses on linear correlation trends and is insufficient in identifying nonlinear anomalies such as sudden pressure increases and decreases, easily leading to missed detections. These methods have weak adaptability to complex operating conditions and are difficult to meet the needs of multi-parameter, high-dynamic monitoring of hydrogen storage containers.

[0004] In actual operation, hydrogen storage containers experience natural fluctuations in environmental conditions such as temperature and humidity. If alarms are triggered using only fixed thresholds, it is easy to generate a large number of invalid alarms, which not only increases the maintenance burden but may also mask real potential hazards. At the same time, different monitoring parameters have varying degrees of impact on safety. The fixed threshold method cannot achieve differentiated judgment and graded early warning based on the importance of parameters, resulting in insufficient sensitivity to changes in critical safety parameters while being overly sensitive to normal fluctuations in non-critical parameters.

[0005] In terms of data support and analysis methods, the historical database is not sufficiently comprehensive and lacks in-depth analysis tools for real-time monitoring data. Currently, most systems only perform simple data recording and comparison, failing to establish in-depth analysis models based on time weighting and operating condition weighting, making it difficult to extract effective fault evolution patterns from long-term operating data.

[0006] Therefore, there is an urgent need for intelligent algorithms that can distinguish the importance of parameters, take into account the characteristics of dimensionlessness and direction recognition, and combine high-sensitivity identification of high-risk faults with anti-interference filtering of environmental interference. Summary of the Invention

[0007] To address the above issues, this invention utilizes the weighted cosine similarity algorithm as its core technology. By constructing a dynamic comparison mechanism between a large-scale historical database and a real-time database, and setting an adaptive dynamic early warning threshold, it achieves hierarchical intelligent early warning. This solves the core problems of existing similarity matching algorithms, such as insensitivity to key parameters, high false alarm rate, and low early warning accuracy, thereby improving the scientific and engineering reliability of hydrogen storage pressure vessel safety monitoring.

[0008] According to embodiments of the present invention, a method for intelligent monitoring and early warning of hydrogen storage pressure vessels based on WCS algorithm optimization is provided.

[0009] In a first aspect of the present invention, a method for intelligent monitoring and early warning of hydrogen storage pressure vessels based on WCS algorithm optimization is provided. The method includes: Step S01: Collect and store the full life cycle data of hydrogen storage pressure vessels of the category to be monitored to form an original historical database, and collect and store the 6-dimensional dynamic data of the in-service hydrogen storage pressure vessels to be monitored in real time. Step S02: Based on the real-time collected 6-dimensional dynamic data, calculate its weighted cosine similarity with the historical benchmark, select similar state data groups from the historical database, and combine the sliding window algorithm to calculate the deviation change rate for comprehensive trend analysis. Step S03: Implement graded early warning and response based on the weighted cosine similarity comparison results; Step S04: Synchronously enter the current warning record, 6-dimensional real-time feature vector, weighted cosine similarity matching results, and deviation analysis data into the original historical database.

[0010] Furthermore, the 6-dimensional dynamic data mentioned in step S01 includes: pressure, hydrogen leakage concentration, container wall temperature, ambient humidity, ambient oxygen concentration, and solar radiation intensity.

[0011] Furthermore, step S02 also includes preprocessing the real-time acquired 6-dimensional dynamic data: using Min-Max normalization to uniformly map different dimensional parameters to the [0,1] interval, and using the 3σ criterion to remove abnormal noise data.

[0012] Furthermore, the formula for calculating the weighted cosine similarity mentioned in step S02 is as follows: , in, This is a 6-dimensional normalized feature vector of historical benchmark data. To provide a 6-dimensional normalized feature vector for real-time monitoring data, For the first The weighting coefficients of each parameter.

[0013] Furthermore, the weights for pressure, hydrogen leakage concentration, container wall temperature, ambient humidity, ambient oxygen concentration, and solar radiation intensity are all set at 0.20 to 0.40.

[0014] Furthermore, the specific steps for comprehensive trend analysis in step S02, which involve combining the sliding window algorithm to calculate the rate of change of deviation, are as follows: using the absolute deviation... and relative deviation rate The difference is calculated using two indicators, and the rate of change of deviation is calculated using a sliding window algorithm. Combining the deviation quantification results, the rate of change, and the predicted remaining life of the container, a logistic regression model is used to output a quantitative value of the risk development trend to assess the trend. To provide a 6-dimensional normalized feature vector for real-time monitoring data, The 6-dimensional normalized feature vector of historical benchmark data.

[0015] Furthermore, the tiered early warning and response described in step S03 specifically includes: When the weighted cosine similarity is <0.85, a Level I warning is triggered. The system will initiate an emergency pressure relief procedure within 10 seconds until the target pressure is ≤10MPa, and simultaneously cut off the gas supply and lock the manual shut-off valve; when 0.85 ≤ weighted cosine similarity... When the weighted cosine similarity is less than 0.90, a Level II warning is triggered, and the system will instruct the directional sealing airbag to be in place and remotely shut off the electromagnetic shut-off valve; when the weighted cosine similarity is less than 0.94, a Level III warning is triggered, and the system will automatically increase the sampling rate of the core sensor by one level and generate a Level III priority maintenance work order; when the weighted cosine similarity is less than 0.97, a Level IV warning is triggered, and the system will push a three-dimensional health report of the equipment and adaptively adjust the sensor sampling frequency to reduce system power consumption.

[0016] In a second aspect of the invention, a device for intelligent monitoring and early warning of hydrogen storage pressure vessels based on WCS algorithm optimization is provided. The device includes: Data acquisition module: used to collect and store the full life cycle data of hydrogen storage pressure vessels of the category to be monitored to form the original historical database, and to collect and store the 6-dimensional dynamic data of the in-service hydrogen storage pressure vessels to be monitored in real time; Similarity calculation module: Based on real-time collected 6-dimensional dynamic data, it calculates the weighted cosine similarity between the data and the historical benchmark, selects data groups of the same state from the historical database, and combines the sliding window algorithm to calculate the rate of change of deviation for comprehensive trend analysis. Early warning and response module: used for hierarchical early warning and response based on weighted cosine similarity comparison results; Results storage module: used to synchronously input the current warning record, 6-dimensional real-time feature vector, weighted cosine similarity matching results and deviation analysis data into the original historical database.

[0017] In a third aspect of the invention, an electronic device is provided. The electronic device includes a memory and a processor, the memory storing a computer program, the processor executing the program to implement the method according to a first aspect of the invention.

[0018] In a fourth aspect of the invention, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the method according to a first aspect of the invention.

[0019] This invention achieves a tiered intelligent early warning system by using the weighted cosine similarity algorithm as its core technology breakthrough. It constructs a dynamic comparison mechanism between a large-scale historical database and a real-time database, sets an adaptive dynamic early warning threshold, and solves the core problems of similarity matching algorithms in existing technologies, such as insensitivity to key parameters, high false alarm rate, and low early warning accuracy. This improves the scientific and engineering reliability of safety monitoring of hydrogen storage pressure vessels.

[0020] It should be understood that the description in the Summary of the Invention is not intended to limit the key or essential features of the embodiments of the present invention, nor is it intended to restrict the scope of the invention. Other features of the invention will become readily apparent from the following description.

[0021] The beneficial effects of this invention are: 1. By adopting the weighted cosine similarity algorithm and implementing differentiated weight configuration based on the safety importance of parameters, a core breakthrough has been achieved: key parameters are targeted and sensitive, while non-key parameters have a tolerant response. This significantly improves the accuracy and reliability of early warning while effectively reducing the probability of false alarms and false alarms. 2. The output monitoring results can respond smoothly and stably to abnormal conditions of different severity, providing a reliable basis for establishing clear and reasonable multi-level early warning boundaries. At the same time, it can maintain stable low false alarm characteristics within a wide threshold setting range, demonstrating excellent system adaptability and decision robustness. 3. By continuously feeding back early warning records and corresponding deviation analysis results and integrating them into the historical database, dynamic optimization and self-improvement of benchmark data are achieved, thereby promoting the evolution of the entire monitoring system from static judgment to a data-driven predictive maintenance framework. Attached Figure Description

[0022] The above and other features, advantages, and aspects of the various embodiments of the present invention will become more apparent from the accompanying drawings and the following detailed description. Wherein: Figure 1 A flowchart of a method for intelligent monitoring and early warning of hydrogen storage pressure vessels based on WCS algorithm optimization according to an embodiment of the present invention is shown; Figure 2 A schematic diagram of the real-time pressure monitoring signal of a real-time hydrogen storage pressure vessel according to an embodiment of the present invention is shown. Figure 3 A schematic diagram of a real-time hydrogen leak concentration monitoring signal according to an embodiment of the present invention is shown; Figure 4 A schematic diagram of a real-time wall temperature monitoring signal according to an embodiment of the present invention is shown; Figure 5 A schematic diagram of a real-time ambient humidity monitoring signal according to an embodiment of the present invention is shown; Figure 6 A schematic diagram of a real-time ambient oxygen concentration monitoring signal according to an embodiment of the present invention is shown; Figure 7 A schematic diagram of a real-time radiation intensity monitoring signal according to an embodiment of the present invention is shown; Figure 8 The illustration shows a real-time score change diagram of each window data point after normalization and weighted cosine similarity matching with 500 historical benchmarks according to an embodiment of the present invention. Figure 9 A block diagram of a device for intelligent monitoring and early warning of hydrogen storage pressure vessels based on WCS algorithm optimization according to an embodiment of the present invention is shown. Figure 10 A schematic diagram of a device for intelligent monitoring and early warning of hydrogen storage pressure vessels based on the WCS algorithm optimization according to an embodiment of the present invention is shown. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only 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.

[0024] According to an embodiment of the present invention, a method for intelligent monitoring and early warning of hydrogen storage pressure vessels based on WCS algorithm optimization is proposed. The method takes the weighted cosine similarity algorithm as the core technology breakthrough, and achieves hierarchical intelligent early warning by constructing a dynamic comparison mechanism between a large-scale historical database and a real-time database and setting an adaptive dynamic early warning threshold. This solves the core problems of similarity matching algorithms in the prior art, such as insensitivity to key parameters, high false alarm rate and low early warning accuracy, thereby improving the scientific and engineering reliability of hydrogen storage pressure vessel safety monitoring.

[0025] The principles and spirit of the present invention will be explained in detail below with reference to several representative embodiments.

[0026] Figure 1 This is a schematic flowchart of a method for intelligent monitoring and early warning of hydrogen storage pressure vessels based on WCS algorithm optimization according to an embodiment of the present invention. The method includes: Step S01: Collect and store the full life cycle data of hydrogen storage pressure vessels of the category to be monitored to form an original historical database, and collect and store the 6-dimensional dynamic data of the in-service hydrogen storage pressure vessels to be monitored in real time. Step S02: Based on the real-time collected 6-dimensional dynamic data, calculate its weighted cosine similarity with the historical benchmark, select similar state data groups from the historical database, and combine the sliding window algorithm to calculate the deviation change rate for comprehensive trend analysis. Step S03: Implement graded early warning and response based on the weighted cosine similarity comparison results; Step S04: Synchronously enter the current warning record, 6-dimensional real-time feature vector, weighted cosine similarity matching results, and deviation analysis data into the original historical database.

[0027] It should be noted that although the operation of the method of the present invention has been described in a specific order in the above embodiments and figures, this does not require or imply that the operations must be performed in that specific order, or that all the operations shown must be performed to achieve the desired result. Additionally or alternatively, certain steps may be omitted, multiple steps may be combined into one step, and / or one step may be broken down into multiple steps.

[0028] To provide a clearer explanation of the above-mentioned intelligent monitoring and early warning method for hydrogen storage pressure vessels based on the WCS algorithm optimization, a specific embodiment will be used for illustration below. However, it is worth noting that this embodiment is only for better illustrating the present invention and does not constitute an improper limitation of the present invention.

[0029] The following example will further illustrate the intelligent monitoring and early warning method for hydrogen storage pressure vessels based on the WCS algorithm optimization. In this embodiment, the simulation parameters are set in accordance with GB / T 34542.1-2017, GB50516-2021 and related standards, and all algorithm matching calculations are obtained through actual execution of Python code.

[0030] This embodiment uses a 4130X steel hydrogen storage pressure vessel, which has been in service for 5 years and operates at a daily pressure of 35 MPa, as the monitoring object. It demonstrates the operational results at each stage of a 120-minute operating scenario, encompassing four phases: normal, abnormal development, severe, and recovery. The simulation employs standard parameters such as a rated operating pressure of 35 MPa, a temperature range of -40°C to 65°C, and hydrogen concentration alarm thresholds of 25 / 500 / 1000 ppm. The pressure change rate conforms to the TSG 21-2016 regulations for overpressure conditions in pressure vessels, and the online monitoring sampling frequency is 5 times / min. 500 historical baseline data points are included, covering different operating conditions and service life.

[0031] Step S01: Collect and store the full life cycle data of hydrogen storage pressure vessels of the category to be monitored to form an original historical database, and collect and store the 6-dimensional dynamic data of the in-service hydrogen storage pressure vessels to be monitored in real time.

[0032] Specifically, the database collects factory parameters for this type of high-pressure hydrogen storage container, including material (4130X, Cr-Mo steel), rated pressure (45 MPa), and volume (10 m³), ​​as well as operational data from the past 10 years (1-10 years) at different service lifespans, maintenance records of corrosion / fatigue damage cases, and operational characteristic data under different ambient temperatures (-40°C to 65°C) and working pressures (5-45 MPa). The database is updated monthly, incorporating data on newly added containers and updated data on existing containers.

[0033] The system collects and stores six-dimensional dynamic data of the in-service hydrogen storage pressure vessels under monitoring in real time, including pressure, hydrogen leakage concentration, vessel wall temperature, ambient humidity, ambient oxygen concentration, and solar radiation intensity. Specifically, the tank pressure is collected by a piezoelectric sensor with a range of 0–100 MPa and an accuracy of ±0.1%FS; the hydrogen leakage concentration is collected by an electrochemical sensor with a detection limit of 0.1 ppm; the vessel wall temperature is collected by a platinum resistance thermometer (PT100) with an accuracy of ±0.1°C; the ambient humidity is obtained by a dew point meter; the ambient oxygen concentration is measured by an ultrasonic sensor; and the solar radiation intensity is collected by a radiation sensor. These three data are incorporated into the feature vector as auxiliary monitoring parameters.

[0034] In this embodiment, the real-time pressure monitoring signal of the hydrogen storage pressure vessel in the simulation scenario is as follows: Figure 2 As shown, the real-time hydrogen leak concentration monitoring signal is as follows: Figure 3 As shown, the real-time wall temperature monitoring signal is as follows: Figure 4As shown, the real-time ambient humidity monitoring signal is as follows: Figure 5 As shown, the real-time ambient oxygen concentration monitoring signal is as follows: Figure 6 As shown, the real-time radiation intensity monitoring signal is as follows: Figure 7 As shown.

[0035] Step S02: Based on the real-time collected 6-dimensional dynamic data, calculate its weighted cosine similarity with the historical benchmark, select similar state data groups from the historical database, and combine the sliding window algorithm to calculate the deviation change rate for comprehensive trend analysis.

[0036] The 6-dimensional dynamic data collected in real time is preprocessed. Min-Max normalization is used to map the parameters of different dimensions to the [0,1] interval to eliminate the difference in dimensions. At the same time, abnormal noise data is removed by the 3σ criterion.

[0037] The six parameters are assigned differentiated weights according to their importance to hydrogen storage safety, forming a weight vector W. Among them, the weight of pressure ranges from 0.20 to 0.40, preferably from 0.25 to 0.35; the weight of hydrogen leakage concentration ranges from 0.20 to 0.40, preferably from 0.25 to 0.35; the weight of container wall temperature ranges from 0.10 to 0.30, preferably from 0.15 to 0.25; the weight of ambient humidity ranges from 0.01 to 0.15, preferably from 0.03 to 0.08; the weight of ambient oxygen concentration ranges from 0.01 to 0.15, preferably from 0.03 to 0.08; and the weight of solar radiation intensity ranges from 0.01 to 0.20, preferably from 0.05 to 0.15. Specifically, the sum of the weights of the key safety parameters (pressure, hydrogen leakage concentration, container wall temperature) is greater than the sum of the weights of the environmental auxiliary parameters (ambient humidity, ambient oxygen concentration, solar radiation intensity) in order to achieve a monitoring strategy that provides a high-sensitivity response to key safety parameters and a low-weight, tolerant treatment of environmental auxiliary parameters.

[0038] Similarity matching is performed using weighted cosine similarity, and the formula is: , in, This is a 6-dimensional normalized feature vector of historical benchmark data. To provide a 6-dimensional normalized feature vector for real-time monitoring data, For the first The weighting coefficients of each parameter are determined. The similarity threshold is set to S≥90%, and similar container data groups are selected from the historical database as the comparison benchmark.

[0039] By absolute deviation and relative deviation rate The difference between the two indicators is calculated, among which, The 6-dimensional normalized feature vector of the real-time monitoring data, i.e., the 6-dimensional dynamic data of the currently collected in-service hydrogen storage pressure vessels, is mapped to the [0,1] interval after Min-Max normalization processing. The 6-dimensional normalized feature vector is derived from historical baseline data. This vector consists of 6-dimensional parameter values ​​of similar containers selected from the original historical database under normal / baseline conditions, normalized using the same Min-Max method. The rate of change of deviation is calculated using a sliding window algorithm. Combining the deviation quantification results, the rate of change, and the container's remaining life prediction data, a logistic regression model is used to output a quantitative value of the risk development trend to assess the trend.

[0040] In this embodiment, the weights are: pressure (0.30), hydrogen concentration (0.30), wall temperature (0.20), humidity (0.05), oxygen concentration (0.05), and radiation (0.10). Weighted cosine similarity is calculated for each historical record. Filter out 442 historical records of similar containers with a value ≥0.90 were used as the comparison benchmark group, and the best matching record was selected. =0.9992.

[0041] Calculate absolute deviation—pressure =1.65MPa ( =4.2%), hydrogen concentration =0.26ppm ( =13.1%), wall temperature =0.9°C ( =2.6%), all within the normal range. A 6-hour sliding window was used to continuously calculate the rate of change of the deviation, and the trend was comprehensively analyzed by combining the deviation quantification results with the rate of change.

[0042] Figure 8 The data displays the real-time score changes when a sliding window of 40 sampling points (8 minutes) is used to match the weighted cosine similarity of each window's data after Min-Max normalization with 500 historical benchmarks. During the normal phase, the similarity remains stable above 0.98; after anomalies develop, it gradually decreases to below 0.90; during the severe phase, it bottoms out at approximately 0.897; and during the recovery phase, it rebounds as pressure returns. Three dashed lines mark the four warning levels: IV / III (S=0.97), III / II (S=0.94), and II / I (S=0.90).

[0043] Step S03: Perform graded early warning and response based on the weighted cosine similarity comparison results.

[0044] The risk level is directly determined based on the numerical range of the weighted cosine similarity comparison results, and the standardized response plan associated with that level is immediately implemented. Specifically, when... When the pressure is less than 0.85, a Level I (red) warning is triggered. The system will initiate an emergency pressure relief procedure within 10 seconds until the target pressure is ≤10MPa, and simultaneously cut off the gas supply and lock the manual shut-off valve. When 0.85 ≤ When the value is less than 0.90, a Level II (orange) warning is triggered. The system will instruct the directional sealing airbag to be in place and remotely shut off the electromagnetic shut-off valve. When 0.90 ≤ When the value is less than 0.94, a Level III (yellow) warning is triggered. The system will automatically increase the sampling rate of the core sensor by one level and generate a Level III priority maintenance work order. When 0.94 ≤ When the value is less than 0.97, a Level IV (blue) warning is triggered. The system will push a 3D health report of the device and adaptively adjust the sensor sampling frequency to reduce system power consumption.

[0045] Based on the above process operation examples, as shown in Table 1, the contribution percentages of each parameter to the decrease in similarity at the lowest similarity time (t≈86min) are as follows: Pressure (32%) and hydrogen concentration (26%), as high-weight key parameters, contributed 58% of the total deviation, strongly verifying the rationality of the weight vector W=[0.30,0.30,0.2,0.05,0.05,0.1]—high-weight parameters do indeed dominate the decrease in similarity in actual anomalies. Wall temperature (16%) is the second highest, while the three low-weight non-key parameters, environmental humidity (6%), oxygen concentration (9%), and radiation (11%), contributed only 26% in total. Their noise disturbances were effectively suppressed, achieving the design goal of directional sensitivity of key parameters and tolerant response of non-key parameters.

[0046] Table 1

[0047] Step S04: Synchronously enter the current warning record, 6-dimensional real-time feature vector, weighted cosine similarity matching results, and deviation analysis data into the original historical database.

[0048] The warning record, 6-dimensional real-time feature vector, weighted cosine similarity matching results, and deviation analysis data were simultaneously entered into the original historical database, expanding the historical baseline from 500 to 501 records, completing a closed-loop monitoring and warning process. Algorithm verification data: Historical baseline matching operations were performed using five different algorithms on 100 sets of identical real-time monitoring data. Due to interference from environmental parameters, Euclidean distance and Manhattan distance matched a large number of historical records with similar environmental values ​​but vastly different container properties (such as different materials and service lives), rendering the baseline data incomparable. This comparison result fully demonstrates that, within the technical framework of dynamic weighted comparison of hydrogen storage pressure vessels using dual databases, the weighted cosine similarity algorithm has irreplaceable technical value in ensuring the reliability of baseline matching and the accuracy of subsequent warning decisions.

[0049] In this embodiment, the performance advantages of the weighted cosine similarity algorithm in the monitoring scenario of hydrogen storage pressure vessels are also verified.

[0050] Four sets of comparative experiments were conducted under simulation conditions to evaluate five algorithms, including weighted cosine similarity, standard cosine similarity, Pearson correlation coefficient, Euclidean distance, and Manhattan distance. The simulation parameters all conformed to the pressure (≤140MPa) and temperature (-40°C to 65°C) ranges specified in GB / T 34542.1-2017, and the hydrogen concentration alarm thresholds (25 / 500 / 1000ppm) specified in GB 50516-2021. Historical baseline data consisted of 500 records from vessels under different operating conditions and with varying ages, while real-time monitoring data comprised 600 mixed samples including environmental noise and anomalies of varying degrees.

[0051] Experiment 1: Robustness Verification of Weight Range. To further verify that the weight vector can effectively perform the monitoring and early warning function within the reasonable value range defined in the claims, this embodiment selects five sets of weight vectors that meet the requirements. =1 and key safety parameters ( + + The sum of the weights is greater than the environmental auxiliary parameters. + + The weight configuration of the sum of weights constraint covers the upper limit, lower limit and intermediate value of the range described in the claims, as shown in Table 2.

[0052] Table 2

[0053] Under the same simulation conditions (500 historical baselines and 600 real-time monitoring data including environmental noise and anomalies of different degrees), weighted cosine similarity matching and graded early warning were performed on the five weights respectively. The evaluation indicators included ROC-AUC, PR-AUC, environmental noise false alarm rate (FAR@S=0.90) and multi-severity level response monotonicity. The results are shown in Table 3.

[0054] Among them, the rights reorganization A ( =[0.30,0.30,0.20,0.05,0.05,0.10]) is the preferred weight configuration adopted in the embodiments of the present invention; weighting group B takes the upper limit of pressure weight 0.40 and the lower limit of hydrogen concentration weight 0.20; weighting group C takes the upper limit of hydrogen concentration weight 0.40 and the lower limit of pressure weight 0.20; weighting group D takes the weight of key parameters close to the upper limit, totaling 0.85; weighting group E takes the weights to be balanced. All five configurations cover the upper boundary, lower boundary and middle region of the scope defined by the claims.

[0055] Table 3

[0056] As shown in Table 3, all five weight configurations achieved ROC-AUC values ​​above 0.98, and the false alarm rate for environmental noise was controlled below 5%. This verifies that within the weight range defined in the claims, any weight combination that meets the constraints can effectively execute the monitoring and early warning method, achieving engineering-acceptable monitoring accuracy. Among them, weight configuration A achieved optimal results in ROC-AUC (0.9949), PR-AUC (0.9935), and zero false alarm rate, indicating that this weight configuration achieved the best balance between the sensitivity of key parameters and the suppression of environmental noise, making it the optimal weight configuration of this invention. Weighted recombination B (pressure weight cap 0.40) performed exceptionally well in pressure-dominated anomaly scenarios, with an ROC-AUC of 0.9917; weighted recombination D (total key parameters 0.85) maintained a high overall performance (ROC-AUC = 0.9902); weighted recombination E (weights tending towards equilibrium) had acceptable overall performance (ROC-AUC = 0.9806), but the false alarm rate for environmental noise rose to 4.2%, indicating that the ability to suppress noise in non-key dimensions decreased after reducing the weights of key parameters.

[0057] The experimental results above demonstrate that the weight range set in this invention possesses sufficient engineering adaptability and robustness. Under different application scenarios (such as different container types, service life, installation environments, etc.), the weight values ​​can be flexibly adjusted within the range defined in the claims according to actual needs, and the adjusted method still maintains effective monitoring and early warning performance. Furthermore, the global optimality of weight recombination A provides sufficient experimental support for the preferred weight configuration described in the embodiments.

[0058] Experiment 2: The false alarm suppression capability for environmental noise is shown in Table 4. The false alarm rate for weighted cosine similarity is 0.0%, for standard cosine similarity it is 5.3%, for Pearson correlation it is 1.0%, for Euclidean distance it is 89.7%, and for Manhattan distance it is 21.7%. When the threshold is increased to S=0.96, the false alarm rate for weighted cosine similarity rises to 2.3%, and for standard cosine similarity it rises to 15.0%, demonstrating the robustness advantage of the weighted cosine threshold. The above false alarm rate results are statistical values ​​from a single experiment based on 600 mixed samples under the simulation conditions. In actual engineering deployments, the false alarm rate may fluctuate slightly due to factors such as sensor drift and extreme environmental transients, but the significant advantage of weighted cosine similarity over other algorithms remains stable.

[0059] Table 4

[0060] Experiment 3: Overall detection accuracy. The results are shown in Table 5. The weighted cosine similarity (ROC-AUC) is 0.9949 and PR-AUC is 0.9935, both significantly better than the other four algorithms. The weighted cosine similarity algorithm achieves the highest AUC while keeping the false alarm rate at an extremely low level.

[0061] Table 5

[0062] Experiment 3: Monotonicity of Responses at Multiple Severity Levels. As shown in Tables 6 and 7, across the seven severity levels from normal to critical, the weighted cosine score smoothly decreased from 1.0000 to 0.9977, a decrease of 0.23%; the standard cosine score decreased from 1.0000 to 0.9906, a decrease of 0.94%. Both response curves were smooth and stable, maintaining a good monotonically decreasing relationship with the degree of deviation. Furthermore, the weighted cosine score showed a smaller decrease, indicating a stronger ability to suppress noise in non-critical dimensions, with score changes more concentrated on the response to real security threats. In contrast, the Pearson correlation coefficient decreased by 3.7%, the Manhattan distance by 5.3%, and the Euclidean distance by 16.8%, indicating that they overreacted to changes in deviation and lacked a proportional relationship with the degree of deviation.

[0063] Table 6

[0064] Table 7

[0065] Experiment 4: Threshold reliability. As shown in Tables 8-9, at S=0.90, the proportion of abnormal sample scores exceeding the normal median using the weighted cosine method is 0.0%, and the standard cosine method is 0.0%, with almost no overlap between normal and abnormal distributions. When the threshold is increased to S=0.96, the weighted cosine FAR is 2.3%, indicating a significantly optimal safety margin.

[0066] Table 8

[0067] Table 9

[0068] Based on the combined results of the five experiments, the weighted cosine similarity performed best in four dimensions: false alarm suppression, overall detection accuracy, adaptability to graded early warning, and threshold reliability. It has irreplaceable technical advantages in the monitoring of hydrogen storage pressure vessels.

[0069] Based on the same inventive concept, this invention also proposes a device for intelligent monitoring and early warning of hydrogen storage pressure vessels based on WCS algorithm optimization. The implementation of this device can be found in the implementation of the method described above; repeated details will not be elaborated further. Figure 9 As shown, the device 100 includes: Data acquisition module 101: used to collect and store the full life cycle data of hydrogen storage pressure vessels of the category to be monitored to form an original historical database, and to collect and store 6-dimensional dynamic data of in-service hydrogen storage pressure vessels to be monitored in real time; Similarity calculation module 102: It is used to calculate the weighted cosine similarity between the 6-dimensional dynamic data collected in real time and the historical benchmark, select the data group of the same state from the historical database, and combine the sliding window algorithm to calculate the deviation change rate for comprehensive trend analysis. Early warning and response module 103: used for hierarchical early warning and response based on weighted cosine similarity comparison results; Result storage module 104: used to synchronously input the current warning record, 6-dimensional real-time feature vector, weighted cosine similarity matching results and deviation analysis data into the original historical database.

[0070] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the described module can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0071] like Figure 10 As shown, the device includes a central processing unit (CPU), which can perform various appropriate actions and processes based on computer program instructions stored in read-only memory (ROM) or loaded from storage units into random access memory (RAM). The RAM can also store various programs and data required for device operation. The CPU, ROM, and RAM are interconnected via a bus. Input / output (I / O) interfaces are also connected to the bus.

[0072] Multiple components in the device are connected to the I / O interface, including: input units such as keyboards and mice; output units such as various types of displays and speakers; storage units such as disks and optical discs; and communication units such as network interface cards (NICs), modems, and wireless transceivers. The communication unit allows the device to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0073] The processing unit executes the various methods and processes described above, such as method steps S01 to S04. For example, in some embodiments, method steps S01 to S04 may be implemented as a computer software program tangibly contained in a machine-readable medium, such as a storage unit. In some embodiments, part or all of the computer program may be loaded and / or installed on the device via ROM and / or a communication unit. When the computer program is loaded into RAM and executed by the CPU, one or more steps of method steps S01 to S04 described above may be performed. Alternatively, in other embodiments, the CPU may be configured to execute method steps S01 to S04 by any other suitable means (e.g., by means of firmware).

[0074] The functions described above in this document can be performed at least in part by one or more hardware logic components. For example, exemplary types of hardware logic components that can be used, without limitation, include: field programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload programmable logic devices (CPLDs), and so on.

[0075] The program code used to implement the methods of the present invention can be written in any combination of one or more programming languages. This program code can be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing device, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code can be executed entirely on the machine, partially on the machine, as a standalone software package partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0076] In the context of this invention, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. Machine-readable media can include, but are not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0077] Furthermore, although the operations are described in a specific order, this should be understood as requiring that such operations be performed in the specific order shown or in sequential order, or requiring that all illustrated operations be performed to achieve the desired result. In certain environments, multitasking and parallel processing may be advantageous. Similarly, although several specific implementation details are included in the above discussion, these should not be construed as limiting the scope of the invention. Certain features described in the context of individual embodiments may also be implemented in combination in a single implementation. Conversely, various features described in the context of a single implementation may also be implemented individually or in any suitable sub-combination in multiple implementations.

[0078] Although the subject matter has been described using language specific to structural features and / or methodological logic, it should be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or actions described above. Rather, the specific features and actions described above are merely illustrative examples of implementing the claims.

Claims

1. A method for intelligent monitoring and early warning of hydrogen storage pressure vessels based on WCS algorithm optimization, characterized in that, The method includes: Step S01: Collect and store the full life cycle data of hydrogen storage pressure vessels of the category to be monitored to form an original historical database, and collect and store the 6-dimensional dynamic data of the in-service hydrogen storage pressure vessels to be monitored in real time. Step S02: Based on the real-time collected 6-dimensional dynamic data, calculate its weighted cosine similarity with the historical benchmark, select similar state data groups from the historical database, and combine the sliding window algorithm to calculate the deviation change rate for comprehensive trend analysis. Step S03: Implement graded early warning and response based on the weighted cosine similarity comparison results; Step S04: Synchronously enter the current warning record, 6-dimensional real-time feature vector, weighted cosine similarity matching results, and deviation analysis data into the original historical database.

2. The method for intelligent monitoring and early warning of hydrogen storage pressure vessels based on WCS algorithm optimization according to claim 1, characterized in that, The 6-dimensional dynamic data mentioned in step S01 includes: pressure, hydrogen leakage concentration, container wall temperature, ambient humidity, ambient oxygen concentration, and solar radiation intensity.

3. The method for intelligent monitoring and early warning of hydrogen storage pressure vessels based on WCS algorithm optimization according to claim 1, characterized in that, Step S02 further includes preprocessing the real-time acquired 6-dimensional dynamic data: using Min-Max normalization to uniformly map different dimensional parameters to the [0,1] interval, and using the 3σ criterion to remove abnormal noise data.

4. The method for intelligent monitoring and early warning of hydrogen storage pressure vessels based on WCS algorithm optimization according to claim 1, characterized in that, The formula for calculating the weighted cosine similarity mentioned in step S02 is as follows: ， in, This is a 6-dimensional normalized feature vector of historical benchmark data. To provide a 6-dimensional normalized feature vector for real-time monitoring data, For the first The weighting coefficients of each parameter.

5. The method for intelligent monitoring and early warning of hydrogen storage pressure vessels based on WCS algorithm optimization according to claim 4, characterized in that, The weights for pressure, hydrogen leakage concentration, container wall temperature, ambient humidity, and solar radiation intensity are 0.20-0.40, 0.20-0.40, 0.10-0.30, 0.01-0.15, 0.01-0.15, and 0.01-0.20, respectively.

6. The method for intelligent monitoring and early warning of hydrogen storage pressure vessels based on WCS algorithm optimization according to claim 1, characterized in that, The specific steps for comprehensive trend analysis by combining the sliding window algorithm to calculate the rate of change of deviation in step S02 are as follows: using the absolute deviation... and relative deviation rate The difference is calculated using two indicators, and the rate of change of deviation is calculated using a sliding window algorithm. Combining the deviation quantification results, the rate of change, and the predicted remaining life of the container, a logistic regression model is used to output a quantitative value of the risk development trend to assess the trend. To provide a 6-dimensional normalized feature vector for real-time monitoring data, The 6-dimensional normalized feature vector of historical benchmark data.

7. The method for intelligent monitoring and early warning of hydrogen storage pressure vessels based on WCS algorithm optimization according to claim 1, characterized in that, The tiered early warning and response described in step S03 specifically refers to: When the weighted cosine similarity is <0.85, a Level I warning is triggered. The system will initiate an emergency pressure relief procedure within 10 seconds until the target pressure is ≤10MPa, and simultaneously cut off the gas supply and lock the manual shut-off valve. When 0.85 ≤ weighted cosine similarity When the weighted cosine similarity is less than 0.90, a Level II warning is triggered, and the system will instruct the directional sealing airbag to be in place and remotely shut off the electromagnetic shut-off valve; when the weighted cosine similarity is less than 0.94, a Level III warning is triggered, and the system will automatically increase the sampling rate of the core sensor by one level and generate a Level III priority maintenance work order; when the weighted cosine similarity is less than 0.97, a Level IV warning is triggered, and the system will push a three-dimensional health report of the equipment and adaptively adjust the sensor sampling frequency to reduce system power consumption.

8. A device for intelligent monitoring and early warning of hydrogen storage pressure vessels based on WCS algorithm optimization, characterized in that, The device implements the method as described in any one of claims 1 to 7, comprising: Data acquisition module: used to collect and store the full life cycle data of hydrogen storage pressure vessels of the category to be monitored to form the original historical database, and to collect and store the 6-dimensional dynamic data of the in-service hydrogen storage pressure vessels to be monitored in real time; Similarity calculation module: Based on real-time collected 6-dimensional dynamic data, it calculates the weighted cosine similarity between the data and the historical benchmark, selects data groups of the same state from the historical database, and combines the sliding window algorithm to calculate the rate of change of deviation for comprehensive trend analysis. Early warning and response module: used for hierarchical early warning and response based on weighted cosine similarity comparison results; Results storage module: used to synchronously input the current warning record, 6-dimensional real-time feature vector, weighted cosine similarity matching results and deviation analysis data into the original historical database.

9. An electronic device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the program, it implements the method as described in any one of claims 1 to 7.

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

Images