Hidden danger association rule mining method for ammonia hydrogen production and refueling station

Abstract

The invention relates to an ammonia hydrogen production hydrogen refueling station hidden danger association rule mining method, which comprises the steps of obtaining accident hidden danger investigation records and inputting the accident hidden danger investigation records into a system to form a historical hidden danger database; performing data processing on the data in the historical hidden danger database to enable the data to adapt to subsequent analysis requirements; optimizing an Apriori algorithm through a bald eagle search algorithm, constructing a BES-Apriori model, and taking the BES-Apriori model as an association algorithm model; mining association rules in the hidden danger data by using an association algorithm model, and identifying potential hidden danger association rules; and the occurrence of the frequent preceding item is mapped to the corresponding association rule, and the occurrence of the frequent succeeding item is predicted, so that prediction information is obtained. According to the method, potential high-risk factors can be quickly identified, and the hidden danger checking precision and efficiency are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of hidden danger analysis, and in particular to a method for mining hidden danger association rules of an ammonia-to-hydrogen refueling station. Background Art

[0002] As the infrastructure for hydrogen supply to hydrogen fuel cell vehicles, hydrogen refueling stations are an indispensable and important link in the hydrogen energy industry chain. Due to the characteristics of low density, high pressure, flammability, explosion and leakage of hydrogen, its storage technology is costly. "Green ammonia" is a good hydrogen-carrying medium with high hydrogen density and low combustion and explosion risk. It is a highly potential hydrogen energy storage and transportation medium. The ammonia decomposition on-site hydrogen production and hydrogenation integrated station effectively solves the key problems of hydrogen storage and transportation technology through distributed cracking and hydrogen production, provides a solution for the utilization of zero-carbon ammonia hydrogen energy, and strongly supports the construction of a clean energy system with ammonia hydrogen as the core. The safe operation of ammonia hydrogen production and hydrogenation stations is crucial to ensuring public safety. Any potential safety hazards may pose a serious threat to the safety of drivers and passengers, hydrogenation station employees and surrounding residents. Traditional hazard analysis methods mostly rely on experience judgment and traditional data analysis technology, while hazard analysis technology based on association rule mining can realize intelligent troubleshooting of potential hazards, improve work efficiency, and provide new ideas and methods for improving the safety management of gas stations. The Apriori algorithm is a classic association rule mining algorithm. Wang Heping et al. improved the traditional Apriori algorithm based on the comprehensive weight obtained by the combined weighting method, and proposed a weighted Apriori algorithm to mine association rules for the risks of each workspace in prefabricated buildings. Kang Jian et al. mined association rules for gas station risk factors based on the Apriori algorithm, systematically sorted out 71 typical accidents, established a metadata framework for people, objects, and management, quantitatively analyzed the association rules between accident causes and attributes from a multidimensional analysis perspective, mined 32 strong association rules, and intuitively showed the internal connection between various risk factors. Wang Ning et al. used the improved Apriori algorithm to conduct data mining and analysis on railway accidents and risk sources, considering the severity of casualties in railway accidents, proposed a new method for calculating support and confidence indicators, and weighted and quantified railway accident factors. At the same time, by adding time constraints, they explored the association rules of railway accident risk sources at different times.

[0003] Although the Apriori algorithm has been widely used in many fields, its application in hydrogen station hidden danger detection has not been reported. In addition, the existing improvements and applications of the Apriori algorithm mostly focus on improving computational efficiency, while paying less attention to the rationality of the mined association rules. Summary of the invention

[0004] The purpose of the present invention is to provide a method for mining association rules of hidden dangers in an ammonia-to-hydrogen refueling station, which can quickly identify potential high-risk factors and improve the accuracy and efficiency of hidden danger detection.

[0005] In order to achieve the above-mentioned purpose, the technical solution adopted by the present invention is: a method for mining association rules of hidden dangers of ammonia-to-hydrogenation stations, comprising:

[0006] Obtain accident hazard investigation records and enter them into the system to form a historical hazard database;

[0007] Process the data in the historical hidden danger database to adapt it to subsequent analysis needs;

[0008] The Apriori algorithm is optimized by the vulture search algorithm, and the BES-Apriori model is constructed and used as the association algorithm model;

[0009] Use the association algorithm model to mine the association rules in the hidden danger data and identify the potential hidden danger association rules;

[0010] By mapping the occurrence of frequent antecedents to corresponding association rules, the occurrence of frequent consequents is predicted, thereby obtaining prediction information.

[0011] Furthermore, the BES-Apriori model first searches for the optimal vector after determining the search range of support and confidence, determines the optimal minimum support min_sup and minimum confidence min_con, and calls the Apriori algorithm after weighing the quantity and quality of the rules, thereby reducing human intervention and experience dependence, and making the Apriori algorithm adaptable to different data sets.

[0012] Furthermore, the implementation steps of the BES-Apriori model are as follows:

[0013] (1) Randomly generate a certain number of vulture individuals, each of which represents a possible combination of support and confidence parameters;

[0014] (2) For each individual, the Apriori algorithm is used to mine association rules under given support and confidence, and its fitness value is calculated according to the predefined objective function;

[0015] (3) updating candidate solutions in the population according to the operation steps of the vulture search algorithm, including soaring, searching, and pouncing; these operations simulate the hunting behavior of vultures in different scenarios, thereby updating the parameter configuration of candidate solutions;

[0016] (4) Determine whether the stopping condition is met. If the stopping condition is met, that is, the maximum number of iterations is reached or the set fitness threshold is reached, the final converged main vulture individual is output, which is the best parameter combination found, that is, the best support and confidence; if the stopping condition is not met, return to step (2) to continue optimization;

[0017] (5) On this basis, the Apriori algorithm is re-executed with optimized parameter settings to mine more effective association rules.

[0018] Furthermore, the average support and the average confidence As the evaluation index of Apriori algorithm association rule mining performance:

[0019]

[0020] In the formula, ∑ i∈n sup(I i ) represents the sum of the support of association rules, N represents the number of rules, ∑ i∈n confd(I i ) represents the sum of the confidences of the association rules; the average support and the average confidence The larger it is, the higher the accuracy.

[0021] Compared with the prior art, the present invention has the following beneficial effects:

[0022] (1) The performance of the Apriori algorithm depends on correctly setting parameters, such as minimum support and minimum confidence. BES can automatically adjust these parameters to adapt to different data sets and problem requirements, making the Apriori algorithm more general and flexible. Automated parameter optimization can reduce human intervention and trial and error time, and improve the practicality of the algorithm.

[0023] (2) A major challenge of the Apriori algorithm is that as the size of the dataset increases, the number of candidate item sets increases rapidly, resulting in an exponential increase in computational complexity. By using BES to optimize Apriori, the search space can be explored more efficiently, with priority given to areas that are more likely to generate high-quality association rules, reducing unnecessary candidate item set generation and thus reducing computational costs.

[0024] (3) Combined with the optimization capability of BES, frequent item sets with high confidence and high support can be identified more accurately, which will help discover more meaningful and practical association rules. BES can help the Apriori algorithm avoid mining too many low-value rules and improve the accuracy and reliability of the rules. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 is a schematic diagram of a method implementation principle of an embodiment of the present invention;

[0026] Figure 2 is a schematic diagram of the implementation principle of the BES-Apriori algorithm in an embodiment of the present invention;

[0027] Figure 3 It is a comparison diagram of the average support and average confidence of the traditional Apriori algorithm and the BES-Apriori algorithm in the embodiment of the present invention. DETAILED DESCRIPTION

[0028] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0029] It should be noted that the following detailed descriptions are exemplary and are intended to provide further explanation of the present application. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present application belongs.

[0030] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present application. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprise" and / or "include" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or combinations thereof.

[0031] like Figure 1 As shown, this embodiment provides a method for mining association rules of hidden dangers of ammonia-to-hydrogen refueling stations, including:

[0032] Obtain accident hazard investigation records and enter them into the system to form a historical hazard database;

[0033] Process the data in the historical hidden danger database to adapt it to subsequent analysis needs;

[0034] The Apriori algorithm is optimized by the condor search algorithm (BES), and the BES-Apriori model is constructed and used as the association algorithm model;

[0035] Use the association algorithm model to mine the association rules in the hidden danger data and deeply identify the potential hidden danger association rules;

[0036] By mapping the occurrence of frequent antecedents to corresponding association rules, the occurrence of frequent consequents is predicted, thereby obtaining valuable prediction information.

[0037] This method optimizes the support and confidence parameter combination through the vulture search algorithm, realizes the adaptive selection of parameters, and obtains more accurate and reliable hidden danger association rule mining results of ammonia-to-hydrogen refueling stations, providing strong support for hidden danger detection and safety management of gas stations.

[0038] Although the Apriori algorithm is very effective in the field of association rule mining, the range of support and confidence values ​​may be very wide during implementation, so there is a risk of overfitting when looking for optimal parameters. A high support threshold may result in a reduction in the number of frequent item sets and association rules generated, thereby missing potentially useful association rules; setting too low a minimum support may result in too many frequent item sets and association rules being generated, which may contain a large number of meaningless or unreliable rules, increasing the complexity of the analysis. Therefore, when choosing the minimum support, it is necessary to weigh the quantity and quality of the rules and the generalization ability of the model. Choosing an appropriate support threshold is very important for discovering meaningful association rules.

[0039] This method constructs an Apriori model optimized by the vulture search algorithm, namely the BES-Apriori model. Compared with the traditional Apriori algorithm, the BES-Apriori model first searches for the optimal vector after determining the search range of support and confidence, determines the optimal minimum support min_sup and minimum confidence min_con, and then calls the Apriori algorithm after weighing the quantity and quality of rules, thereby reducing human intervention and experience dependence, and making the Apriori algorithm adaptable to different data sets.

[0040] like Figure 2 As shown, the implementation steps of the BES-Apriori model are as follows:

[0041] (1) Randomly generate a certain number of vulture individuals, each of which represents a possible combination of support and confidence parameters.

[0042] (2) For each individual, the Apriori algorithm is used to mine association rules under given support and confidence, and its fitness value is calculated according to the predefined objective function.

[0043] (3) Update the candidate solutions in the population according to the operation steps of the vulture search algorithm, including soaring, searching, and pouncing. These operations simulate the hunting behavior of vultures in different scenarios, thereby updating the parameter configuration of the candidate solutions.

[0044] (4) Determine whether the stopping condition is met. If the stopping condition is met, that is, the maximum number of iterations is reached or the set fitness threshold is reached, the final converged main vulture individual is output, which is the best parameter combination found, that is, the best support and confidence; if the stopping condition is not met, return to step (2) to continue optimization.

[0045] (5) On this basis, the Apriori algorithm is re-executed with optimized parameter settings to mine more effective association rules.

[0046] In this embodiment, the average support is used and the average confidence As the evaluation index of Apriori algorithm association rule mining performance:

[0047]

[0048] In the formula, ∑ i∈n sup(I i ) represents the sum of the support of association rules, N represents the number of rules, ∑ i∈n confd(I i ) represents the sum of the confidences of the association rules; the average support and the average confidence The larger it is, the higher the accuracy.

[0049] The following is a further introduction to the Apriori algorithm and vulture search algorithm involved in this method.

[0050] 1. Apriori algorithm

[0051] The Apriori algorithm is a classic association rule mining algorithm. Its basic principle is based on the concept of "frequent item sets" and aims to discover interesting patterns and association rules between item sets from large-scale transaction databases. The following is the basic principle of the Apriori algorithm and its working process.

[0052] The support of an association rule is the ratio of the number of transactions of X and Y contained in the transaction set to the number of all transactions, expressed as:

[0053] support(X→Y)=support(X∪Y)=P(XY)

[0054] The confidence of an association rule is the ratio of the number of transactions containing X and Y to the number of all transactions containing X in the transaction set, expressed as:

[0055]

[0056] Lift refers to the relative probability between an item (or item set) and another item (or item set) under given conditions. It can help determine whether the association rule has real relevance, rather than just due to the high frequency of individual items. It is expressed as:

[0057]

[0058] The implementation steps of the Apriori algorithm are as follows:

[0059] Step 1: Scan the entire data warehouse, calculate all frequent item sets, and find the frequent item sets that are consistent with the previous set support;

[0060] Step 2: Based on the frequent item set generation rules in the previous step, by scanning the support of candidate frequent item sets, discard the data sets whose support of candidate frequent item sets is lower than the threshold;

[0061] Step 3: Get the frequent k-item set that meets the actual needs of the project. The result obtained through K iterations may be an empty set, and the frequent set of K-1 items is returned as the final result.

[0062] 2. Bald Eagle Search Algorithm (BES)

[0063] Bald eagles are powerful predators with excellent eyesight, and they can detect the movement of schools of fish far away while flying. They are opportunistic foragers. Since it is very difficult to catch fish while flying, they will evaluate the cost of hunting based on the target location and the number of fish before hunting. Therefore, the choice of foraging space is an important consideration. After selecting the target area, they will approach the target area and fly in circles to further search the target area. Once the target prey is located, they will quickly dive towards the prey.

[0064] The vulture search algorithm is a type of swarm intelligence optimization algorithm, which is inspired by the behavior of vultures hunting fish in groups. By imitating this simple and efficient hunting method, the vulture algorithm has the characteristics of strong convergence, few parameters, and easy implementation. According to the hunting strategy and social behavior of vultures, the algorithm is mainly divided into three stages: selection stage, search stage, and dive stage.

[0065] (1) Selection stage

[0066] In the selection phase, vulture individuals are first randomly distributed in the search space, and then determine the approximate location of the hunt based on the density of the prey, and give priority to areas with high concentrations of prey. This strategy ensures the randomness of the search range, improves the search efficiency, and speeds up the search for the ideal capture location. This behavior can be expressed by the following formula:

[0067] P new,i =P best +α×r×(P mean -P i ) (1)

[0068] Where α is a parameter used to control the position change, which is between 1.5 and 2, and r is a random number between 0 and 1.best represents the search area chosen by the vulture based on the best location found in the last search. mean It is the average distribution position calculated based on the positions of all vultures after the last search. i is the current position of the i-th vulture.

[0069] (2) Search phase

[0070] In the search phase, the vulture searches for the target area by spiral flight. In this process, it uses the concepts of spiral and group center, so this behavior can be seen as a combination of the whale optimization algorithm and the particle swarm optimization algorithm. The individual vultures move in spirals around the center point. This method can not only sweep the overall search space, increase algorithm diversity, get rid of local optimality, but also avoid repeated calculations, greatly improving the efficiency of the search. The mathematical model of spiral flight is expressed in polar coordinates. In this process, the mathematical model of the vulture position update is shown as follows:

[0071] P new,i =P i +y(i)×(P i -P i+1 )+x(i)×(P i -P mean ) (2)

[0072] θ(i)=a×π×rand,r(i)=θ(i)+R×rand (3)

[0073]

[0074] xr(i)=r(i)×sin(θ(i)), yr(i)=r(i)×cos(θ(i)) (5)

[0075] Where θ(i) and r(i) are the polar angle and polar diameter of the spiral equation in polar coordinates; a and R vary in the range of (5,10) and (0.5,2), respectively, and are mainly used to control the spiral trajectory and determine the search period; rand is a random number between (0,1); x(i) and y(i) represent the current position of the vulture in polar coordinates, and their value ranges are (-1,1).

[0076] (3) Subduction stage

[0077] During the dive phase, the vulture quickly dives from the optimal position in the search space toward the target prey, and other individuals in the population move in the direction of the optimal position under the influence of the environment and attack the prey at the same time. The following equation illustrates this behavior:

[0078] P new,i =rand×P best+x1(i)×(P i -c1×P mean )+y1(i)×(P i -c2×P best ) (6)

[0079]

[0080] xr(i)=r(i)×sinh(θ(i)), yr(i)=r(i)×cosh(θ(i)) (8)

[0081] θ(i)=a×π×rand,r(i)=θ(i)c1,c2∈[1,2] (9)

[0082] Where c1 and c2 increase the intensity of the vulture's movement toward the optimal point and the center point, and the value range is [1,2].

[0083] From the above three stages, we can see that concentration and diversity are the two most important features of the vulture search algorithm. The concept of group center ensures the efficiency of the search and reflects the concentration characteristics of the vulture search algorithm. The spiral flight method provides sufficient diversity for the vulture algorithm, which helps to get rid of the local optimum. This enables the vulture search algorithm to efficiently and accurately obtain the optimal solution.

[0084] The method is further described below with reference to a specific embodiment.

[0085] 1. Data Collection

[0086] The data comes from the daily inspection record data collected by the production supervision and early warning system of an ammonia hydrogenation station. According to the attributes of hidden dangers, hidden dangers are divided into four categories: management defects, illegal operations, environmental defects and equipment defects. The specific classification of hidden danger data is shown in Table 1.

[0087] Table 1 Classification of hidden dangers in a certain ammonia hydrogenation station

[0088]

[0089]

[0090] Since the original data stored in the risk database cannot be used as direct input data for the experiment, it is necessary to convert some text data of the original data into a Boolean type suitable for the association algorithm (aggregate the hidden danger records of each inspection in units of times, and mark it as 1 if the hidden danger occurs at that time, and mark it as 0 if it does not occur. Process the data into a 0-1 matrix mode), and delete data including invalid values ​​and missing values. Table 2 shows the preprocessed fragments of multiple hidden danger inspection data of a certain ammonia hydrogenation station.

[0091] Table 2 Preprocessed implicit screening data fragments

[0092]

[0093] 2. Improved Apriori algorithm performance analysis

[0094] In order to verify the improvement of accuracy and effectiveness of this method over the classic algorithm, this embodiment constructs multiple data sets with different sample sizes, and compares the performance of the BES-Apriori algorithm with the traditional Apriori algorithm under different sample sizes. The results are as follows: Figure 3 shown.

[0095] It can be seen that compared with the traditional Apriori method, the average support and average confidence of the BES-Apriori algorithm are significantly improved.

[0096] 3. Analysis of the correlation between accident hazards based on traditional Apriori

[0097] The traditional Apriori algorithm is implemented based on the data in Table 2, with the minimum support set to 0.45 and the minimum confidence set to 0.40. Table 3 shows the top ten confidence-based association rules for hidden dangers of ammonia-to-hydrogen refueling stations.

[0098] Table 3 Accident hazard association rules

[0099]

[0100]

[0101] It can be seen from Table 3 that in the process of hidden danger investigation of the gas station, the support degree of Rule 1 is 56%, the confidence degree is 85%, and the improvement degree is as high as 2.29. Therefore, it is necessary to establish a feedback mechanism to collect new problems and suggestions found in the relocation process and continuously improve safety management measures; the support degree of Rule 2 is 53%, the confidence degree is 80%, and the improvement degree is 1.86. Therefore, it is necessary to establish a complete safety production responsibility system, covering job responsibilities, assessment mechanisms, reward and punishment measures, etc.; the support degree of Rule 3 is 50%, the confidence degree is 75%, and the improvement degree is 1.77, reminding enterprises to choose the appropriate type of centrifugal pump; the support degree of Rule 4 is 0.50, the confidence degree is 0.75, and the improvement degree is 1.50, reminding enterprises to conduct surveys before relocation operations; the support degree of Rule 5 is 0.5, the confidence degree is 0.83, and the improvement degree is 1.93. Therefore, it is necessary to establish a complete safety production responsibility system, covering job responsibilities, assessment mechanisms, reward and punishment measures, etc. In this investigation, we can find that the logic of the strong association rules of Rule 8 and Rule 9 is not compact enough, and the association responsibility system of the previous and next rules is not directly corresponding. It may be due to the accidental factors in the hidden danger investigation that the two associations are caused, and the strong association rules corresponding to the hidden danger cannot be mined. In view of the certain defects of the classic Apriori algorithm, we use the Apriori algorithm optimized based on BES for further analysis. 4. Analysis of the association of accident hidden dangers based on BES-Apriori

[0102] As an optimization algorithm, BES will explore the data space more effectively and find more representative frequent item sets and association rules. In addition, the automatic optimization mentioned above selects the optimal minimum support and minimum confidence for our system, and no parameter comparison is required. The gas station accident association rules of BES-Apriori are shown in the table.

[0103] Table 4 Accident hazard association rules

[0104]

[0105]

[0106] Table 4 shows that the Apriori algorithm optimized by BES still retains high support rules, and on this basis, searches for frequent item sets and association rules more comprehensively. Rule 3 has a support of 0.55 after being optimized by the BES algorithm, and the logical connection between the association rules can be intuitively felt. We found that while new strong association rules appeared, the original strong association rules such as Rule 1, Rule 2, Rule 4, Rule 5, Rule 6, and Rule 7 were still retained.

[0107] The present invention provides a method for mining association rules of hidden dangers in ammonia-to-hydrogen refueling stations to solve the problems of low efficiency and poor accuracy in the traditional manual hidden danger investigation process. To this end, the method introduces the Apriori algorithm to achieve in-depth analysis of hidden dangers by mining potential association rules in accident data. Through the optimization processing of the algorithm, potential high-risk factors can be quickly identified, the accuracy and efficiency of hidden danger investigation can be improved, the interference of human factors can be reduced, and more scientific decision-making support can be provided for safety management.

[0108] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0109] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0110] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0111] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0112] The above is only a preferred embodiment of the present invention, and does not limit the present invention in other forms. Any technician familiar with the profession may use the above disclosed technical content to change or modify it into an equivalent embodiment with equivalent changes. However, any simple modification, equivalent change and modification made to the above embodiment according to the technical essence of the present invention without departing from the technical solution of the present invention still belongs to the protection scope of the technical solution of the present invention.

Claims

1. A method for mining association rules of hidden dangers in ammonia-to-hydrogen refueling stations, characterized in that: include: Obtain accident hazard investigation records and enter them into the system to form a historical hazard database; Process the data in the historical hidden danger database to adapt it to subsequent analysis needs; The Apriori algorithm is optimized by the vulture search algorithm, and the BES-Apriori model is constructed and used as the association algorithm model; Use the association algorithm model to mine the association rules in the hidden danger data and identify the potential hidden danger association rules; By mapping the occurrence of frequent antecedents to corresponding association rules, the occurrence of frequent consequents is predicted, thereby obtaining prediction information.

2. The method for mining hidden danger association rules of an ammonia-to-hydrogenation station according to claim 1 is characterized in that: The BES-Apriori model first searches for the optimal vector after determining the search range of support and confidence, determines the optimal minimum support min_sup and minimum confidence min_con, and then calls the Apriori algorithm after weighing the quantity and quality of the rules, thereby reducing human intervention and experience dependence, and making the Apriori algorithm adaptable to different data sets.

3. The method for mining hidden danger association rules of an ammonia-to-hydrogenation station according to claim 1 is characterized in that: The implementation steps of the BES-Apriori model are as follows: (1) Randomly generate a certain number of vulture individuals, each of which represents a possible combination of support and confidence parameters; (2) For each individual, the Apriori algorithm is used to mine association rules under given support and confidence, and its fitness value is calculated according to the predefined objective function; (3) updating candidate solutions in the population according to the operation steps of the vulture search algorithm, including soaring, searching, and pouncing; these operations simulate the hunting behavior of vultures in different scenarios, thereby updating the parameter configuration of candidate solutions; (4) Determine whether the stopping condition is met. If the stopping condition is met, that is, the maximum number of iterations is reached or the set fitness threshold is reached, the final converged main vulture individual is output, which is the best parameter combination found, that is, the best support and confidence; if the stopping condition is not met, return to step (2) to continue optimization; (5) On this basis, the Apriori algorithm is re-executed with optimized parameter settings to mine more effective association rules.

4. The method for mining hidden danger association rules of an ammonia-to-hydrogenation station according to claim 1 is characterized in that: Average support and the average confidence As the evaluation index of Apriori algorithm association rule mining performance: In the formula, ∑ i∈n sup(I i ) represents the sum of the support of association rules, N represents the number of rules, ∑ i∈n confd(I i ) represents the sum of the confidences of the association rules; the average support and the average confidence The larger it is, the higher the accuracy.

Images