Accident data and physical knowledge co-fused hydrogen refueling station risk assessment method

Abstract

The invention discloses a hydrogen refueling station risk assessment method based on co-fusion of accident data and physical knowledge. The method comprises the following steps: acquiring M marked samples; carrying out sample expansion on the M marked samples; performing statistical analysis on the expanded samples to obtain a hydrogen refueling station risk cause statistical result and a hydrogen refueling station accident type statistical result; based on the hydrogen refueling station risk cause statistical result, the hydrogen refueling station accident type statistical result and the hydrogen refueling station failure related physical knowledge, constructing a hydrogen refueling station failure cause Bayesian network; different structure learning algorithms are adopted to learn the network structure of the Bayesian network caused by the failure of the hydrogen refueling station, an EM algorithm is utilized to solve network parameters of the Bayesian network caused by the failure of the hydrogen refueling station, and different risk assessment models of the hydrogen refueling station are obtained; screening out an optimal hydrogen refueling station risk assessment model by using a K-fold cross validation method; and carrying out sensitivity analysis and risk priority ranking on the screened optimal hydrogen refueling station risk assessment model. The risk assessment accuracy of the hydrogen refueling station can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of risk assessment of hydrogen refueling stations, and more specifically to a risk assessment method for hydrogen refueling stations that integrates accident data and physical knowledge. Background Art

[0002] The existing risk assessment modeling for urban hydrogen refueling stations mainly relies on expert experience and domain knowledge, and the uncertainty of the model is high. The risk assessment of hydrogen refueling stations is mainly based on the experience of the assessor, rather than sufficient quantitative data support, which restricts the accuracy of risk assessment and further risk management.

[0003] Therefore, how to provide a hydrogen refueling station risk assessment method with higher risk assessment accuracy is an urgent problem that technicians in this field need to solve. Summary of the invention

[0004] In view of this, an object of the present invention is to provide a hydrogen station risk assessment method that integrates accident data and physical knowledge.

[0005] In order to achieve the above object, the present invention adopts the following technical solution:

[0006] A hydrogen refueling station risk assessment method integrating accident data and physical knowledge includes the following steps:

[0007] S1: Obtain M sets of accident report data related to hydrogen refueling stations and physical knowledge related to hydrogen refueling station failures;

[0008] Mark the risk causes of M sets of accident reports related to hydrogen refueling stations respectively, and obtain M marked samples;

[0009] S2: expanding the M marked samples; and using the hydrogen refueling station risk assessment literature to perform statistical analysis on the expanded samples to obtain statistical results of risk causes and types of accidents at hydrogen refueling stations;

[0010] S3: constructing a Bayesian network of hydrogen station failure causes based on the statistical results of the risk causes of the hydrogen station, the statistical results of the accident types of the hydrogen station, and the physical knowledge related to the failure of the hydrogen station;

[0011] S4: using different structural learning algorithms to learn the network structure of the hydrogen station failure causation Bayesian network, and using the EM algorithm to solve the network parameters of the hydrogen station failure causation Bayesian network to obtain different hydrogen station risk assessment models;

[0012] S5: Use K-fold cross validation method to select the optimal hydrogen station risk assessment model;

[0013] S6: Conduct sensitivity analysis and risk priority ranking on the selected optimal hydrogen refueling station risk assessment model.

[0014] Preferably, the statistical results of the risk causes of the hydrogen refueling station include various risk causes and the probability of each risk cause.

[0015] Preferably, the statistical results of accident types of hydrogen refueling stations include leakage, fire and explosion.

[0016] Preferably, the Bootstrap method is used for sample expansion: the M labeled samples are sampled with replacement to achieve sample expansion.

[0017] Preferably, S4 specifically includes:

[0018] Using the BS algorithm to learn the network structure of the hydrogen station failure causation Bayesian network; using the EM algorithm to solve the network parameters of the hydrogen station failure causation Bayesian network to obtain the first hydrogen station risk assessment model;

[0019] The PC algorithm is used to learn the network structure of the hydrogen station failure causation Bayesian network; the EM algorithm is used to solve the network parameters of the hydrogen station failure causation Bayesian network to obtain the second hydrogen station risk assessment model;

[0020] The GTT algorithm is used to learn the network structure of the Bayesian network of the hydrogen refueling station failure causes; the EM algorithm is used to solve the network parameters of the Bayesian network of the hydrogen refueling station failure causes, and the risk assessment model of the third hydrogen refueling station is obtained.

[0021] Preferably, S5 further includes using the overall accuracy, precision, recall and F1 value as verification criteria to screen out the optimal hydrogen station risk assessment model from the first hydrogen station risk assessment model, the second hydrogen station risk assessment model and the third hydrogen station risk assessment model.

[0022] Preferably, the sensitivity analysis specifically includes:

[0023] The probability of a certain risk factor is randomly adjusted to obtain its impact on the accident type of hydrogen refueling stations.

[0024] Preferably, the risk priority ranking specifically includes:

[0025] The accident type of hydrogen refueling station is fixed as one of the three types, and the probability of each risk cause is obtained by reverse calculation. The reversed probabilities are sorted to obtain several risk causes with the highest risk cause probability.

[0026] Preferably, the physical knowledge related to the failure of the hydrogen filling station includes that the failure of the pipeline joint seal and the rupture of the hose joint will directly lead to hydrogen leakage. If the environmental hydrogen detection device fails and the leakage cannot be detected in time, as the hydrogen accumulates in the surrounding environment, it will cause an explosion when encountering suitable conditions; wherein, the suitable conditions include abnormal discharge caused by electromagnetic threshold failure;

[0027] Failure of the pressure relief device causes abnormal increase in the pressure inside the equipment. When the pressure exceeds the equipment's tolerance limit, it will cause the equipment to rupture and leak, leading to fire and explosion accidents.

[0028] Earthquakes can damage storage containers and pipeline systems at hydrogen refueling stations, causing hydrogen leaks. If there is a fire source in the leakage area or the explosion limit concentration is reached due to poor ventilation, serious explosion accidents will occur.

[0029] An electronic device comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the above-mentioned hydrogen refueling station risk assessment method integrating accident data with physical knowledge when executing the computer program.

[0030] It can be seen from the above technical solutions that, compared with the prior art, the present invention discloses a hydrogen refueling station risk assessment method that integrates accident data and physical knowledge, which can improve the accuracy of hydrogen refueling station risk assessment. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0032] Figure 1 A flow chart of a hydrogen refueling station risk assessment method integrating accident data and physical knowledge provided by the present invention;

[0033] Figure 2 The present invention is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0034] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0035] like Figure 1 As shown, the embodiment of the present invention discloses a hydrogen station risk assessment method integrating accident data and physical knowledge, comprising the following steps:

[0036] S1: Obtain M sets of accident report data related to hydrogen refueling stations and physical knowledge related to hydrogen refueling station failures;

[0037] Mark the risk causes of M hydrogen refueling station-related accident report data to obtain M marked samples;

[0038] It is understandable that: conducting statistical analysis on accident data of urban hydrogen refueling stations, collating information from public hydrogen energy accident databases such as the EU Hydrogen Incident and Accident Database and Japan's High Pressure Gas Safety Act, and obtaining the M copies of hydrogen refueling station-related accident report data.

[0039] The risk cause marking is specifically as follows: the risk cause causing the accident is set to 1 (for the calculation of the "number of occurrences" in the subsequent Table 1, and the number of risk causes set to 1 can be counted to obtain the "number of occurrences" of the risk cause); the risk cause that did not cause the accident is set to 0;

[0040] Furthermore, the physical knowledge related to the failure of the hydrogen refueling station (physical knowledge is mainly used to construct the mandatory relationship of the Bayesian network) includes that the failure of the pipeline joint seal and the rupture of the hose joint will directly lead to hydrogen leakage. If the environmental hydrogen detection device fails and the leakage cannot be detected in time, as hydrogen accumulates in the surrounding environment, it will cause an explosion when encountering suitable conditions; wherein, the suitable conditions include abnormal discharge caused by electromagnetic threshold failure;

[0041] Failure of the pressure relief device causes the internal pressure of the equipment to increase abnormally. When the pressure exceeds the equipment's tolerance limit, it will cause the equipment to rupture and cause leakage, which in turn may lead to fire and explosion accidents.

[0042] Earthquakes can damage storage containers and pipeline systems at hydrogen refueling stations, causing hydrogen leaks. If there is a fire source in the leakage area or the explosion limit concentration is reached due to poor ventilation, serious explosion accidents will occur.

[0043] S2: expanding the M marked samples; and using the hydrogen refueling station risk assessment literature to perform statistical analysis on the expanded samples to obtain statistical results of risk causes and types of accidents at hydrogen refueling stations;

[0044] In one embodiment, the statistical results of the risk causes of the hydrogen refueling station include various risk causes and the probability of each risk cause.

[0045] In one embodiment, the statistical results of the types of accidents at the hydrogen refueling station include leakage, fire and explosion.

[0046] In a certain embodiment, the Bootstrap method is used for sample expansion: the M labeled samples are sampled with replacement to achieve sample expansion.

[0047] In one embodiment, the statistical results of the risk causes of the hydrogen refueling station are shown in Table 1;

[0048] Table 1

[0049]

[0050]

[0051] S3: constructing a Bayesian network of hydrogen station failure causes based on the statistical results of the risk causes of the hydrogen station, the statistical results of the accident types of the hydrogen station, and the physical knowledge related to the failure of the hydrogen station;

[0052] S4: using different structural learning algorithms to learn the network structure of the hydrogen station failure causation Bayesian network, and using the EM algorithm to solve the network parameters of the hydrogen station failure causation Bayesian network to obtain different hydrogen station risk assessment models;

[0053] In one embodiment, S4 specifically includes:

[0054] The BS algorithm (Binary Search Algorithm) is used to learn the network structure of the hydrogen station failure causation Bayesian network; the EM algorithm (Expectation Maximization Algorithm) is used to solve the network parameters of the hydrogen station failure causation Bayesian network to obtain the first hydrogen station risk assessment model;

[0055] The PC algorithm (Peter-Ciark Algorithm) is used to learn the network structure of the hydrogen station failure causation Bayesian network; the EM algorithm is used to solve the network parameters of the hydrogen station failure causation Bayesian network to obtain the second hydrogen station risk assessment model;

[0056] The GTT (Greedy Thick Thinning) algorithm is used to learn the network structure of the hydrogen station failure causation Bayesian network; the EM algorithm is used to solve the network parameters of the hydrogen station failure causation Bayesian network to obtain the third hydrogen station risk assessment model.

[0057] In one embodiment, the network parameters of the hydrogen station failure causal Bayesian network are optimized based on the following formula:

[0058]

[0059] Wherein, X represents observed data (i.e., the statistical results of the risk causes of the hydrogen refueling station and the statistical results of the accident types of the hydrogen refueling station); Z represents latent variables; θ represents the network parameters of the failure cause Bayesian network of the hydrogen refueling station; θ(t) represents the network parameters of the failure cause Bayesian network of the hydrogen refueling station obtained by the t-th round of iterative optimization; θ(t+1) represents the network parameters of the failure cause Bayesian network of the hydrogen refueling station obtained by the t+1-th round of iterative optimization; L(θ|X, Z) represents the joint probability distribution of the observed data X and the latent variable Z under the given parameter θ; represents the expectation of the latent variable z given the observed data x and the current parameter estimate θ(t); Q(θ|θ (t) ) represents the expected log-likelihood function, which represents the expected value of the log-likelihood of the joint distribution of the observed data X and the latent variable Z under the current parameter estimate θ(t).

[0060] In one embodiment, the EM algorithm is divided into an E step and an M step;

[0061] Step E: The algorithm estimates the distribution of latent variables using the current model parameters.

[0062] M-step: Maximize the expected log-likelihood obtained in the E-step to update the network parameters.

[0063] S5: Use K-fold cross validation method to select the optimal hydrogen station risk assessment model;

[0064] In one embodiment, S5 further includes using the overall accuracy, precision (P), recall (R) and F1 value as verification criteria to screen out the optimal hydrogen station risk assessment model from the first hydrogen station risk assessment model, the second hydrogen station risk assessment model and the third hydrogen station risk assessment model.

[0065] It is understandable that:

[0066] Overall accuracy = (number of correctly classified samples / total number of samples) × 100% (3)

[0067] P=TP / (TP+FP)(4)

[0068] R=TP / (TP+FN)(5)

[0069] F1=2×(P×R) / (P+R)(6)

[0070] Among them, TP represents true positive examples, FP represents false positive examples, and FN represents false negative examples.

[0071] For example, taking the fire accident type of a hydrogen refueling station as an example, a true positive example means that a fire accident occurred at the hydrogen refueling station, and the model also predicts that a fire occurred. A false positive example means that no fire accident occurred at the hydrogen refueling station, but the model predicts that a fire occurred. A false negative example means that a fire accident occurred at the hydrogen refueling station, but the model predicts that no fire occurred.

[0072] The larger the overall accuracy, precision (P), recall (R) and F1 value, the better the evaluation model is.

[0073] In one embodiment, the comparison of the effects of the three evaluation models of the present invention is shown in Table 2.

[0074] Table 2

[0075]

[0076] S6: Conduct sensitivity analysis and risk priority ranking on the selected optimal hydrogen refueling station risk assessment model.

[0077] In one embodiment, the sensitivity analysis specifically includes:

[0078] The probability of a certain risk factor is randomly adjusted to obtain its impact on the accident type of hydrogen refueling stations.

[0079] For example, randomly increase the probability of a certain risk cause and obtain its impact on the accident type of hydrogen refueling station.

[0080] In one embodiment, the risk priority ranking specifically includes:

[0081] The accident type of hydrogen refueling station is fixed as one of the three types, and the probability of each risk cause is obtained by reverse calculation. The reversed probabilities are sorted to obtain several risk causes with the highest risk cause probability.

[0082] The embodiment of the present invention also discloses a physical structure diagram of an electronic device, such as Figure 2 As shown, the electronic device may include: a processor 201, a communications interface 202, a memory 203 and a communication bus 204, wherein the processor 201, the communications interface 202 and the memory 203 communicate with each other through the communication bus 204. The processor 201 may call the logic instructions in the memory 203 to execute the above-mentioned hydrogen station risk assessment method integrating accident data and physical knowledge.

[0083] In addition, the logic instructions in the above-mentioned memory 203 can be implemented in the form of software functional units and can be stored in a computer-readable storage medium when sold or used as an independent product. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in the embodiment of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0084] On the other hand, an embodiment of the present invention further provides a computer program product, which includes a computer program. The computer program can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the above-mentioned hydrogen station risk assessment method that integrates accident data and physical knowledge.

[0085] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.

[0086] The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention is not limited to the embodiments shown herein.

[0087] Rather, it is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A hydrogen refueling station risk assessment method integrating accident data and physical knowledge, characterized in that: The following steps are involved: S1: Obtain M sets of accident report data related to hydrogen refueling stations and physical knowledge related to hydrogen refueling station failures; Mark the risk causes of M sets of accident reports related to hydrogen refueling stations respectively, and obtain M marked samples; S2: performing sample expansion on the M labeled samples; The expanded sample was statistically analyzed using the hydrogen refueling station risk assessment literature to obtain the statistical results of the risk causes and accident types of hydrogen refueling stations. S3: constructing a Bayesian network of hydrogen station failure causes based on the statistical results of the risk causes of the hydrogen station, the statistical results of the accident types of the hydrogen station, and the physical knowledge related to the failure of the hydrogen station; S4: using different structural learning algorithms to learn the network structure of the hydrogen station failure causation Bayesian network, and using the EM algorithm to solve the network parameters of the hydrogen station failure causation Bayesian network to obtain different hydrogen station risk assessment models; S5: Use K-fold cross validation method to select the optimal hydrogen station risk assessment model; S6: Conduct sensitivity analysis and risk priority ranking on the selected optimal hydrogen refueling station risk assessment model.

2. A hydrogen refueling station risk assessment method integrating accident data and physical knowledge according to claim 1, characterized in that: The statistical results of the risk causes of the hydrogen refueling station include various risk causes and the probability of each risk cause.

3. A hydrogen refueling station risk assessment method integrating accident data and physical knowledge according to claim 2, characterized in that: The statistical results of hydrogen refueling station accident types include leakage, fire and explosion.

4. A hydrogen refueling station risk assessment method integrating accident data and physical knowledge according to claim 1, characterized in that: The Bootstrap method is used for sample expansion: the M labeled samples are sampled with replacement to achieve sample expansion.

5. The hydrogen refueling station risk assessment method integrating accident data and physical knowledge according to claim 1 is characterized in that: S4 specifically includes: The BS algorithm is used to learn the network structure of the hydrogen station failure causation Bayesian network; the EM algorithm is used to solve the network parameters of the hydrogen station failure causation Bayesian network to obtain the first hydrogen station risk assessment model; The PC algorithm is used to learn the network structure of the hydrogen station failure causation Bayesian network; the EM algorithm is used to solve the network parameters of the hydrogen station failure causation Bayesian network to obtain the second hydrogen station risk assessment model; The GTT algorithm is used to learn the network structure of the Bayesian network of the hydrogen refueling station failure causes; the EM algorithm is used to solve the network parameters of the Bayesian network of the hydrogen refueling station failure causes, and the risk assessment model of the third hydrogen refueling station is obtained.

6. A hydrogen refueling station risk assessment method integrating accident data and physical knowledge according to claim 5, characterized in that: S5 further includes screening out the optimal hydrogen refueling station risk assessment model from the first hydrogen refueling station risk assessment model, the second hydrogen refueling station risk assessment model and the third hydrogen refueling station risk assessment model by using the overall accuracy, precision, recall and F1 value as verification criteria.

7. A hydrogen refueling station risk assessment method integrating accident data and physical knowledge according to claim 3, characterized in that: The sensitivity analysis specifically includes: The probability of a certain risk factor is randomly adjusted to obtain its impact on the accident type of hydrogen refueling stations.

8. A hydrogen refueling station risk assessment method integrating accident data and physical knowledge according to claim 3, characterized in that: The risk priority ranking specifically includes: The accident type of hydrogen refueling station is fixed as one of the three types, and the probability of each risk cause is obtained by reverse calculation. The reversed probabilities are sorted to obtain several risk causes with the highest risk cause probability.

9. A hydrogen refueling station risk assessment method integrating accident data and physical knowledge according to claim 5, characterized in that: The physical knowledge related to the failure of the hydrogen refueling station includes that the failure of the pipeline joint seal and the rupture of the hose joint will directly lead to hydrogen leakage. If the environmental hydrogen detection device fails and the leakage cannot be detected in time, as the hydrogen accumulates in the surrounding environment, it will cause an explosion when encountering suitable conditions; wherein, the suitable conditions include abnormal discharge caused by electromagnetic threshold failure; Failure of the pressure relief device causes abnormal increase in the pressure inside the equipment. When the pressure exceeds the equipment's tolerance limit, it will cause the equipment to rupture and leak, leading to fire and explosion accidents. Earthquakes can damage storage containers and pipeline systems at hydrogen refueling stations, causing hydrogen leaks. If there is a fire source in the leakage area or the explosion limit concentration is reached due to poor ventilation, serious explosion accidents will occur.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the hydrogen refueling station risk assessment method integrating accident data and physical knowledge as described in any one of claims 1 to 9 is implemented.

Images