FRAM-based dangerous chemical storage tank area fire explosion accident cause analysis method

Abstract

The invention discloses a hazardous chemical substance storage tank area fire explosion accident cause analysis method based on an FRAM. The method comprises the steps that accident cases are processed through a rooting theory, accident cause factors are extracted, function recognition in FRAM analysis is improved in combination with an abstract hierarchy theory, and a system function network diagram of a storage tank area is constructed; obtaining the grey relational degree of the accident cause factors to the occurrence of the fire and explosion accident through a grey relational degree method, substituting the grey relational degree as a weight into an Apriori algorithm, and mining a dangerous chemical storage tank area cause factor association rule; and establishing a key function connection network of fire and explosion accidents in the hazardous chemical substance storage tank area, and proposing targeted prevention measures based on a coupling relationship among cause factors. According to the method, the causes of the fire explosion accident in the hazardous chemical substance storage tank area can be comprehensively and effectively analyzed, frequently occurring safety risk factors or bad behavior modes can be mined, and the importance of the factors of the causes of the fire explosion accident in the hazardous chemical substance storage tank area and the coupling action relationship between the factors can be analyzed.

Description

Technical Field

[0001] The invention belongs to the technical field of accident analysis, and in particular relates to a method for analyzing the causes of fire and explosion accidents in a hazardous chemical storage tank area based on FRAM. Background Art

[0002] With the continuous increase in the production and reserves of hazardous chemicals, storage tanks, as important storage facilities in the industry, the capacity and volume of a single storage tank and the scale of the storage tank area are showing a situation of continuous expansion. Moreover, as one of the key units in the chemical industry, the safety and stability of the storage tank area directly affects the safety and production efficiency of the entire chemical enterprise. According to statistics, the current rate of accidents in the petroleum and chemical industry due to hazardous chemical storage tanks is second only to accidents caused by process equipment failure. In accidents in hazardous chemical storage tank areas, the types of accidents include fire, explosion, poisoning, etc., among which fire and explosion accidents are the main ones. At the same time, given that hazardous chemicals are often stored in storage tank areas, accidents in storage tank areas often have the characteristics of a large impact range, serious consequences of accidents, and arduous rescue difficulties. Therefore, it is necessary to conduct an in-depth analysis of the causes of accidents in hazardous chemical storage tank areas in order to explore the laws of accident occurrence and prevention points.

[0003] The existing domestic accident analysis of hazardous chemical tank areas is mainly based on the frequency analysis of single descriptive accident causes and the calculation and analysis of explosion and fire models, and lacks the discussion of the interaction between causal factors. At the same time, when an accident occurs in a hazardous chemical storage tank area, there will be many factors such as personnel behavior and organizational safety management, and there are also complex characteristics such as nonlinearity and emergence between factors. It can be seen that the occurrence of accidents in the tank area is not caused by a single functional failure or malfunction, but is the result of the mutual coupling of various factors. It is difficult to fully reveal the potential laws of accident occurrence by relying solely on linear accident analysis models. Therefore, it is necessary to seek appropriate analysis methods to better understand and analyze the complexity and uncertainty of accidents in hazardous chemical storage tank areas.

[0004] Functional resonance method (FRAM) starts from the perspective of system function, and provides ideas for analyzing the coupling of factors in accidents by analyzing the variability of various factors in the system and the interactive mode of changes, so as to obtain a more thorough situation of the coupling of factors in the system. In-depth research on the coupling relationship of causal factors in the hazardous chemical storage tank area system is of great significance for understanding the law of accident occurrence. At the same time, the existing hazardous chemical storage tank area accident cases record the process of the accident, forming a large amount of text data for analysis, which provides a large number of accident case support for the FRAM analysis of hazardous chemical storage tank areas. Summary of the invention

[0005] In order to address the deficiencies of the prior art, the present invention aims to provide a method for analyzing the causes of fire and explosion accidents in hazardous chemical storage tank areas based on FRAM. The method can effectively solve the problems in the prior art that it is impossible to understand and describe increasingly complex systems by relying solely on simple causal relationships, and that it is difficult to fully reveal the potential laws of accident occurrence by relying solely on linear accident analysis models. The method can more thoroughly analyze the complexity and uncertainty of accidents in hazardous chemical storage tank areas, explore the coupling relationship between the causal factors of fire and explosion accidents in hazardous chemical storage tank areas, and provide ideas for the prevention and safety management of fire and explosion accidents in hazardous chemical storage tank areas.

[0006] In order to achieve the above object, the technical solution adopted by the present invention is:

[0007] A method for analyzing the causes of fire and explosion accidents in hazardous chemical storage tank areas based on FRAM includes the following steps:

[0008] S1. Process accident cases through grounded theory and extract the causes of accidents;

[0009] S2. Combine the abstract hierarchy theory to improve the function identification in FRAM analysis and construct the system function network diagram of the tank area;

[0010] S3. Calculate the grey correlation of the fire and explosion accidents for the accident causal factors obtained by the grey correlation method;

[0011] S4. Substitute the grey correlation degree as the weight into the Apriori algorithm to mine the association rules of the causal factors of the hazardous chemicals storage tank area;

[0012] S5. Establish a key functional connection network for fire and explosion accidents in hazardous chemical storage tank areas, and propose targeted preventive measures based on the coupling relationship between causal factors.

[0013] Furthermore, the accident-causing factors extracted in step S1 include:

[0014] S1-1. Collect information;

[0015] S1-2. Construct a case library: By collecting accident information data of hazardous chemical storage tank areas in a certain year, since the collected data should provide more detailed data support for subsequent analysis, only cases with accident descriptions are not selected, and accident investigation reports with detailed records are selected to extract text information to construct a case library;

[0016] S1-3. Accident data processing: First, the cause of the accident is extracted from the accident report and processed to form data entries; then the jieba toolkit is used to perform Chinese word segmentation processing to obtain the accident Chinese word segmentation entries;

[0017] S1-4. Summarization of causal factors: Using word segmentation technology, the text data is divided into independent vocabulary units, and the entries with the same semantics are summarized and organized into causal factors.

[0018] Furthermore, the information collected in step S1-1 includes accident cases and case libraries involved in previously published documents, accident investigation reports on hazardous chemical storage tank areas in published books, the website of the Ministry of Emergency Management of the People's Republic of China, local government emergency management websites, official websites such as the National Bureau of Statistics, and news reports.

[0019] Furthermore, step S1-4 specifically includes:

[0020] By processing the accident cases through text processing and Chinese word segmentation, we obtain the set of accident causative factors, which is the first-level coding of grounded theory analysis. On the basis of the first-level coding, we classify the factors with similarity in the first-level coding in combination with relevant laws and regulations as well as the accident causal model to obtain the second-level coding. Similarly, we classify the second-level coding to obtain the third-level coding, and finally obtain the set of accident causative factors that can be used for subsequent FRAM analysis.

[0021] Furthermore, step S2 specifically includes:

[0022] S2-1. Use the abstract hierarchy theory to divide the tank area system into functional levels;

[0023] S2-2. Use the three coding levels obtained from grounded theory analysis to identify the different functional levels of the system;

[0024] S2-3. Finally, the relationship between functions is determined by describing the identified functions in detail to construct a system function network diagram of the tank area;

[0025] By analyzing the correlation between functions and the integrity of the relationship between functions, a functional network diagram is finally established based on the functional relationship questionnaire.

[0026] Furthermore, step S2-1 specifically includes:

[0027] Combined with the abstract hierarchy theory, the hazardous chemicals storage tank area system is abstracted into two functional layers: the generalized functional layer and the physical functional layer; combined with the characteristics of the hazardous chemicals storage tank area, the scope of each functional layer is further clarified: for the generalized functional layer, this layer further refines the goals of the functional purpose layer into specific functional requirements; for the physical functional layer, the functional requirements of the generalized functional layer are converted into actual operating procedures and steps. At this layer, it is necessary to plan the daily operation process of the hazardous chemicals storage tank area in detail, including the daily operations of the storage tank area, as well as emergency measures for dealing with emergencies.

[0028] Furthermore, step S2-2 specifically includes: analyzing and describing the hexagonal features of each identified function based on the FRAM model, thereby analyzing the connection between the functions to construct the relationship rules between the functions; if the upstream function affects the downstream function, the relevant function attribute column is marked as 1 in the function relationship table; if there is no connection between the functions, it is marked as 0 in the function relationship table, and there may be a situation where a single function is connected to multiple functions.

[0029] Furthermore, the grey correlation method is used to calculate the grey correlation degree in step S3, which includes the following steps:

[0030] S3-1. Establishing the reference sequence and comparison sequence:

[0031] Under given conditions, if the reference sequence is divided into j, there are k comparison sequences as evaluation factors, and there are i samples in total, then the reference sequence x i (k) represents the actual measured value of i in the kth evaluation factor in the reference sequence; where i ranges from 1 to n, and k ranges from 1 to t; the corresponding comparison sequence is y j (k) represents the specific value of the k evaluation factor in the comparison sequence, where j ranges from 1 to m;

[0032] S3-2. Standardization of raw data:

[0033] Before conducting grey correlation analysis, in order to ensure the comparability and accuracy of the data, the data needs to be dimensionally normalized to eliminate the dimensional differences between the data, so that the correlation analysis between different indicators or variables can be conducted fairly. To achieve this goal, the mean method is used to standardize the data and obtain the dimensionless new values ​​of each factor:

[0034]

[0035] In the formula, and Respectively represent x i (k) and y j The dimensionless new value of (k);

[0036] S3-3. Calculate the absolute difference:

[0037] Determine the maximum and minimum differences between the reference and comparison sequences, expressed as:

[0038]

[0039] In the formula, Δ min Indicates the minimum difference between the reference sequence and the comparison sequence, Δ max Indicates the maximum difference between the reference sequence and the comparison sequence, Δij (k) represents the difference between the reference sequence and the comparison sequence; and Represent the minimum difference between levels i and j, respectively. and Represent the maximum difference between levels i and j respectively;

[0040] S3-4. Calculation of correlation coefficients of various factors:

[0041] Find the relative difference between the ith comparison sequence and the reference sequence in terms of k indicators:

[0042]

[0043] In the formula, ε ij (k) is x i Corresponding to y i The correlation coefficient; θ is the resolution coefficient, and its value range is (0,1], but in the calculation, its value is generally 0.5;

[0044] S3-5. Calculate the grey relational degree:

[0045] According to all the data obtained above, calculate the contribution of each causal factor to the fire and explosion accident ρ ij :

[0046]

[0047] Furthermore, in step S4, the Apriori algorithm mines association rules of causal factors in the hazardous chemicals storage tank area, including the following steps:

[0048] S4-1. Data preprocessing: Convert the collected non-digital structured text into data labels suitable for Apriori analysis; specifically, construct a table of accident values. If an accident is caused by a certain consistent factor, then in the corresponding accident row and a certain consistent factor column, mark it as "1" in the corresponding table cell; conversely, if the accident is not caused by the factor, mark it as "0" in the table cell. Through labeling processing, a data table suitable for Apriori association analysis is obtained;

[0049] S4-2. Combine the preprocessed data with the grey correlation coefficient: use the calculated grey correlation coefficient of each factor as the weight of the factor and substitute it into the Apriori algorithm operation process;

[0050] S4-3. Calculate support and confidence: Based on the constructed weight library, perform a detailed search of the database to determine the set of items that frequently appear in the items of length 1; use combinatorial logic to generate candidate 2-item sets; screen the support of the candidate 2-item sets to determine the real frequent item sets; based on the identified frequent item sets, construct candidate k+1 item sets, and scan the database again to calculate the support of the candidate k+1 item sets to determine the k+1 frequent item sets that meet the set threshold; this process will be repeated until all frequent item sets are fully discovered and identified; on the known frequent item sets, calculate key indicators such as support and confidence;

[0051] S4-4. Determine strong association rules: Set the support and conditional support to 0.1, and the support confidence to 0.7. By analyzing the association, we can obtain association rules that are greater than a given threshold, thereby revealing rules with strong association.

[0052] Furthermore, step S5 specifically includes:

[0053] S5-1. Establishing the key functional connection network of fire and explosion accidents in hazardous chemical storage tank areas: The functional failure links are judged by mining the accident factor association links through the improved Apriori association rule. The confidence between factors is used to determine the coupling of functions. The connections with functional confidence less than 0.7 are eliminated. Finally, the functional failure links are obtained, and the key functional connection network of fire and explosion accidents in hazardous chemical storage tank areas is established;

[0054] S5-2. Targeted preventive measures: By analyzing the coupling between causal factors, the occurrence patterns of fire and explosion accidents in hazardous chemical storage tank areas are obtained, and targeted accident prevention measures are proposed.

[0055] The beneficial effects of the present invention are:

[0056] 1. The occurrence of accidents in the tank area is not caused by a single functional failure or malfunction, but is the result of the mutual coupling of various factors. However, the existing technology simply relies on linear accident analysis models, which makes it difficult to fully reveal the potential laws of accident occurrence. The present invention adopts the functional resonance method (FRAM) method, starting from the perspective of system function, to analyze the variability of each function in the system and the interaction mode of the changes, to provide ideas for analyzing the coupling of factors in accidents, so as to obtain a more thorough understanding of the coupling of causal factors, and in-depth research on the coupling relationship of causal factors in the hazardous chemical storage tank area system, which can better understand and analyze the complexity and uncertainty of accidents in hazardous chemical storage tank areas, and provide ideas for the prevention and safety management of fire and explosion accidents in hazardous chemical storage tank areas.

[0057] 2. Traditional accident cause analysis methods often rely on subjective judgment and experience summary, and there is a situation where subjective influence is too large. However, the present invention avoids subjective influence to a certain extent by mining and analyzing a large number of accident case text information and using grey correlation analysis method. It also performs calculations based on objective data and mathematical models, which can more accurately reveal the correlation and regularity between data, thereby more objectively and scientifically analyzing the correlation between different causal factors and the occurrence of fire and explosion accidents, and improving the reliability and scientificity of the research.

[0058] 3. The present invention can present the functional network of the hazardous chemicals storage tank area in a graphical manner, which facilitates more intuitive observation and analysis and improves the visualization and applicability of the research. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 This is a flow chart of the method for analyzing the causes of fire and explosion accidents in hazardous chemical storage tank areas of the present invention;

[0060] Figure 2 It is a generalized functional layer functional network diagram of the hazardous chemicals storage tank area of ​​the present invention;

[0061] Figure 3 It is a functional network diagram of the physical functional layer of the hazardous chemicals storage tank area of ​​the present invention;

[0062] Figure 4 It is a key function link network diagram of the generalized function layer of the fire and explosion accident in the hazardous chemicals storage tank area of ​​the present invention;

[0063] Figure 5 This is a key function connection diagram of the physical function layer of a fire and explosion accident in a hazardous chemicals storage tank area of ​​the present invention. DETAILED DESCRIPTION

[0064] The principles and features of the present invention are described below in conjunction with the accompanying drawings. The examples given are only used to explain the present invention and are not intended to limit the scope of use of the present invention.

[0065] like Figure 1 As shown, the present invention proposes a method for analyzing the causes of fire and explosion accidents in hazardous chemical storage tank areas based on FRAM, with the aim of solving the problems that as social and technical systems become increasingly complex, it is no longer possible to understand and describe increasingly complex systems by relying solely on simple causal relationships, and it is difficult to fully reveal the potential laws of accidents by relying solely on linear accident analysis models. The method specifically includes the following steps:

[0066] S1. Use grounded theory to process accident cases and extract the causes of accidents. Specifically:

[0067] S1-1. Collect information: including accident cases and case libraries involved in previously published documents, accident investigation reports on hazardous chemical storage tank areas in published books, the website of the Ministry of Emergency Management of the People's Republic of China, local government emergency management websites, official websites such as the National Bureau of Statistics, and news reports.

[0068] S1-2. Construct a case library: By collecting accident information data of hazardous chemical storage tank areas in a certain year, considering that the collected data must provide more detailed data support for subsequent analysis, cases with only accident descriptions will not be selected. Instead, accident investigation reports with detailed records will be selected and text information will be extracted to construct a case library.

[0069] S1-3. Accident data processing: First, extract the cause of the accident from the accident report and process it to form data entries; then use the jieba toolkit to perform Chinese word segmentation processing to obtain the accident Chinese word segmentation entries.

[0070] S1-4. Summarization of causal factors: Through the above process, word segmentation technology is used to divide the text data into independent vocabulary units, and the terms with the same semantics are summarized and organized into causal factors.

[0071] By processing the accident cases through text processing and Chinese word segmentation, we obtain the set of accident causative factors, which is the first-level coding of grounded theory analysis. On the basis of the first-level coding, we combine relevant laws and regulations and the accident causal model to classify the factors with similarity in the first-level coding to obtain the second-level coding. Similarly, we classify the second-level coding to obtain the third-level coding, and finally obtain the set of accident causative factors that can be used for subsequent FRAM analysis.

[0072] S2. Combine the abstract hierarchy theory to improve the function identification in FRAM analysis and construct the system function network diagram of the tank area. Specifically:

[0073] S2-1. Use the abstract hierarchy theory to divide the functional levels of the storage tank area system; combine the abstract hierarchy theory to abstract the hazardous chemicals storage tank area system into two functional layers: the generalized functional layer and the physical functional layer and physical. By combining the characteristics of the hazardous chemicals storage tank area, the scope of each functional layer can be further clarified: 1) The generalized functional layer, this layer further refines the goals of the functional purpose layer into specific functional requirements. For example, in order to achieve the goal of safe storage of hazardous chemicals, the system needs to have good equipment, facilities, rules and regulations and other functions; 2) The physical functional layer converts the functional requirements of the generalized functional layer into actual operating procedures and steps. At this layer, it is necessary to plan the daily operation process of the hazardous chemicals storage tank area in detail, including the daily operations of the storage tank area, as well as emergency measures for dealing with emergencies.

[0074] S2-2. Use the three coding levels obtained from grounded theory analysis to identify the different functional levels of the system;

[0075] Based on the FRAM model, the hexagonal features of the identified functions are analyzed and described, so as to analyze the connection between the functions and construct the relationship rules between the functions. If the upstream function affects the downstream function, the relevant function attribute column is marked as 1 in the function relationship table; if there is no connection between the functions, it is marked as 0 in the function relationship table. At the same time, there may be a situation where a single function is connected to multiple functions.

[0076] S2-3. Finally, the relationship between functions is determined by describing the identified functions in detail to construct a system function network diagram for the tank area.

[0077] By analyzing the correlation between functions and the integrity of the relationship between functions, a functional network diagram is finally established based on the functional relationship questionnaire. The composition process can be modeled using FMV software (FRAM Model Visualiser).

[0078] S3. Calculate the grey correlation of the fire and explosion accidents for the obtained accident causal factors by grey correlation method. This includes the following steps:

[0079] S3-1. Establishing the reference sequence and comparison sequence:

[0080] Under given conditions, if the reference sequence is divided into j, there are k comparison sequences as evaluation factors, and there are i samples in total, then the reference sequence x i (k) represents the actual measured value of i in the kth evaluation factor in the reference sequence; where i ranges from 1 to n, and k ranges from 1 to t; the corresponding comparison sequence is y j (k) represents the specific value of the k evaluation factor in the comparison sequence, where j ranges from 1 to m;

[0081] S3-2. Standardization of raw data:

[0082] Before conducting grey correlation analysis, in order to ensure the comparability and accuracy of the data, the data needs to be dimensionally normalized to eliminate the dimensional differences between the data, so that the correlation analysis between different indicators or variables can be conducted fairly. To achieve this goal, the mean method is used to standardize the data and obtain the dimensionless new values ​​of each factor:

[0083]

[0084] In the formula, and Respectively represent xi (k) and y j The dimensionless new value of (k);

[0085] S3-3. Calculate the absolute difference:

[0086] Determine the maximum and minimum differences between the reference and comparison sequences, expressed as:

[0087]

[0088] In the formula, Δ min Indicates the minimum difference between the reference sequence and the comparison sequence, Δ max Indicates the maximum difference between the reference sequence and the comparison sequence, Δ ij (k) represents the difference between the reference sequence and the comparison sequence;. and They represent the minimum difference between levels i and j respectively. The calculation steps are to first find the minimum value inside each row (or each data sequence), and then find the minimum value again from these minimum values. and They represent the maximum difference between levels i and j respectively, and the calculation steps are the same as the minimum difference steps.

[0089] S3-4. Calculation of correlation coefficients of various factors:

[0090] Find the relative difference between the ith comparison sequence and the reference sequence in terms of k indicators:

[0091]

[0092] In the formula, ε ij (k) is x i Corresponding to y i The correlation coefficient; θ is the resolution coefficient, and its value range is (0,1], but in the calculation, its value is generally 0.5;

[0093] S3-5. Calculate the grey relational degree:

[0094] According to all the data obtained above, calculate the contribution of each causal factor to the fire and explosion accident ρ ij :

[0095]

[0096] S4. Substitute the grey correlation degree as the weight into the Apriori algorithm to mine the association rules of the causal factors of the hazardous chemicals storage tank area. The following steps are included:

[0097] S4-1. Data preprocessing, converting the collected non-digital structured text into data labels suitable for Apriori analysis. Specifically, a table of accident values ​​is constructed. If an accident is caused by a certain consistent factor, then in the corresponding accident row and a certain causal factor column, the corresponding table cell is marked as "1". On the contrary, if the accident is not caused by this factor, it is marked as "0" in the table cell. Through such labeling processing, a data table suitable for Apriori association analysis is obtained.

[0098] S4-2. Combine the preprocessed data with the grey correlation coefficient, and use the grey correlation degree of each factor calculated above as the weight of the factor and substitute it into the Apriori algorithm calculation process.

[0099] S4-3. Calculate the support and confidence, and conduct an exhaustive search of the database based on the constructed weight library to determine the set of items that frequently appear in the items of length 1; use combinatorial logic to generate candidate 2-item sets; screen the support of the candidate 2-item sets to determine the true frequent item sets; based on the identified frequent item sets, construct candidate k+1 item sets, and scan the database again to calculate the support of the candidate k+1 item sets to determine the k+1 frequent item sets that meet the set threshold; this process will be repeated until all frequent item sets are fully discovered and identified; on the known frequent item sets, calculate key indicators such as support and confidence.

[0100] S4-4. Determine the strong association rules, set the support and conditional support to 0.1, and the support confidence to 0.7. Through the analysis of the association, obtain the association rules that are greater than the given threshold, thereby revealing the rules with strong association.

[0101] S5. Establish a key functional connection network for fire and explosion accidents in hazardous chemical storage tank areas, and propose targeted preventive measures based on the coupling relationship between causal factors.

[0102] S5-1. Establishing the key functional connection network of fire and explosion accidents in hazardous chemical storage tank areas: The functional failure links are judged by mining the accident factor association links through the improved Apriori association rule. The confidence between factors is used to determine the coupling of functions. The connections with functional confidence less than 0.7 are eliminated. Finally, the functional failure links are obtained, and the key functional connection network of fire and explosion accidents in hazardous chemical storage tank areas is established;

[0103] S5-2. Targeted preventive measures: By analyzing the coupling between causal factors, the occurrence patterns of fire and explosion accidents in hazardous chemical storage tank areas are obtained, and targeted accident prevention measures are proposed.

[0104] Simulation experiment:

[0105] The present invention collects 60 hazardous chemical storage tank area accidents from 2000 to 2023, mainly from journal literature, book materials, official websites, news reports, etc., aiming to construct an accident case library for analysis by the present invention.

[0106] The 60 collected accident cases were processed for data, Chinese word segmentation and causal factors were summarized, and 73 first-level coding factors leading to fire and explosion accidents in hazardous chemical storage tank areas were obtained. Based on the grounded theory, 17 second-level codes for fire and explosion accidents in hazardous chemical storage tank areas and 6 third-level codes for fire and explosion accidents in hazardous chemical storage tank areas were obtained. The specific factors are shown in Table 2.

[0107] Table 2

[0108]

[0109]

[0110] Based on the division of functions by abstract hierarchy theory and the analysis results of grounded theory, functions are identified. Through the analysis of the entire life cycle of the hazardous chemicals storage tank area system, six third-level coding factors can cover the life process of the hazardous chemicals storage tank area. Therefore, these six categories are abstracted into the functions that the generalized functional layer should have: GFF1 project construction, GFF2 system construction and implementation, GFF3 process management, GFF4 on-site operations, GFF5 environmental management, GFF6 emergency response and emergency decision-making. The factors of the secondary coding layer of the grounded theory analysis were abstracted to obtain the functions included in the physical functions: PF1 project design and construction, PFF2 safety training, PFF3 safety system construction and implementation, PFF4 human resource allocation, PFF5 internal supervision, PFF6 contractor management, PFF7 process technology design and management, PFF8 operating procedure design and management, PFF9 change management, PFF10 equipment maintenance and equipment management, PFF11 on-site personnel operation, PFF12 on-site operation procedure management, PFF13 on-site communication, PFF14 equipment and operation status control, PFF15 production environment control, PFF16 personnel emergency, PFF17 emergency equipment.

[0111] Through the description of each function and based on the characteristics of the hazardous chemicals storage tank area and various standards and regulations, the six characteristics of the six functions in the generalized functional layer are shown in Table 3, and the six characteristics of the 17 functions in the physical functional layer are shown in Table 4.

[0112] Table 3

[0113]

[0114] Table 4

[0115]

[0116]

[0117]

[0118] After completing the attribute analysis of all functions in the two-layer function, the association between each function is constructed according to the function details. Take the GFF1 project construction and GFF2 system construction execution in the generalized function as an example for analysis. The upstream function GFF1 has an interactive relationship with the downstream functions GFF3, GFF4, GFF5, and GFF6. At the same time, the output of GFF1 is the input of GFF3. Therefore, it is marked as 1 in the corresponding table in the functional relationship table, and the other features are marked as 0. GFF2 is the control of GFF3, GFF4, GFF5, and GFF6 functions. The subscript of the function control attribute corresponding to GFF2 is 1, and the subscripts of the other attributes are marked as 0, as shown in Table 5.

[0119] Table 5

[0120]

[0121] By analyzing the correlation between functions and the integrity of the relationship between functions, a functional network diagram is finally established based on the functional relationship questionnaire. The generalized functional layer functional network of the hazardous chemicals storage tank area is as follows: Figure 1 As shown in the figure, the physical function layer functional network of the hazardous chemicals storage tank area is as follows: Figure 2 shown.

[0122] The original sequence matrix of each factor in the three-level coding layer is constructed as shown in Table 6. 1 (k)(k=1,2,…,24), where the k value represents the number of accidents occurring in different years, which is used as the reference benchmark sequence.

[0123] Table 6

[0124]

[0125] The mean method was used to standardize the factors and obtain the dimensionless new values ​​of each factor. The results are shown in Table 7.

[0126] Table 7

[0127]

[0128] The absolute difference was calculated and the maximum and minimum differences between the reference sequence and the comparison sequence were determined. The results are shown in Table 8, where Δ min =0.0172,Δ max =0.6969.

[0129] Table 8

[0130]

[0131] The correlation coefficients of each factor were calculated and the results are shown in Table 9.

[0132] Table 9

[0133]

[0134] The grey correlation between each factor in the three-level coding layer and the occurrence of accidents is calculated, and the results are shown in Table 10. Since the minimum threshold of the correlation is set to 0.7, it can be concluded that the main causal factors in hazardous chemical fire and explosion accidents are on-site operation defects, system defects, and process management defects.

[0135] Table 10

[0136]

[0137] The grey correlation degree of the secondary coding layer can be obtained through the above process, as shown in Table 11. Under the constraint of the minimum correlation threshold of 0.7, there are 11 factors that are strongly correlated with the fire and explosion accidents in the hazardous chemical storage tank area.

[0138] Table 11

[0139]

[0140] The 60 collected non-digital structured texts were converted into data labels suitable for Apriori analysis. The grey correlation coefficients of the factors were then taken as weights into the Apriori algorithm. The support and conditional support were set to 0.1, and the support confidence was set to 0.7. Through the analysis of the correlation, more than 30 association rules of the third-level coding layer factors were obtained. The top 10 strong association rules were selected as shown in Table 12. Since the second-level coding layer involves more factors and generates more association rules, the top 20 strong association rules are selected as shown in Table 13.

[0141] Table 12

[0142]

[0143] Table 13

[0144]

[0145] The confidence between factors is used to determine the coupling of functions, and the connections with function confidence less than 0.7 are eliminated. Finally, four function failure links are obtained in the generalized function layer, and the fire and explosion failure function diagrams (representing failure function links) of this function layer are drawn, as shown in Table 14 and Table 14. Figure 4Similarly, 21 functional failure links were finally obtained in the physical functional layer, and the fire and explosion failure function diagrams of this functional layer (representing the failure function links) were drawn, as shown in Table 15 and Table 15. Figure 5 Shown

[0146] Table 14

[0147]

[0148] Table 15

[0149]

[0150]

[0151] Through the above analysis, we can get the occurrence law of fire and explosion accidents in hazardous chemical storage tank areas: (1) There is a strong causal relationship between enterprise system, process management and on-site operation functions in fire and explosion accidents in hazardous chemical storage tank areas. At the same time, defects in enterprise system are likely to affect process management and on-site operations, leading to their failure; (2) In fire and explosion accidents in hazardous chemical storage tank areas, 11 causal factors including operating procedure defects, safety training defects, unsafe behaviors, safety system and implementation defects are the key factors leading to fire and explosion accidents in hazardous chemical storage tank areas; (3) Before the accident, at the micro level, in addition to safety training defects, operating procedure defects, and equipment design defects, inspection and maintenance defects are also the cause of fire and explosion accidents. Key causes of accidents, and the failure of safety management system construction and implementation will have a great impact on the occurrence of the above-mentioned causes; (4) On-site operation management and operation status control are important functions that directly cause accidents, and there are also many upstream functions that cause them to change. Therefore, if unsafe behaviors and unsafe states are to be controlled, it is necessary to control their corresponding upstream functions, including suppressing the variability of functions such as safety management system construction and implementation, safety training, and contractor management; (5) During the long-term operation of equipment, the accumulation of hazardous substances, equipment design defects, and insufficient maintenance may cause hazardous substances to leak and accumulate through the joints, valves, and other parts of the tank, thereby causing the tank area system to fall into a dangerous state. If unsafe behaviors or missing operating procedures occur at this time, fire or explosion accidents will be triggered; (6) When grassroots employees are performing special operations, if the construction and implementation of the safety management system, safety training, and the design and implementation of operating procedures fail, it will lead to the person in charge of production safety not being rigorous in approving special operations or failing to obtain special operation permits. In addition, if employees who have not received professional training have deficiencies in operating procedures and work procedures, they will lack the ability to detect risk factors such as dangerous gas accumulation before work; (7) Among the various factors that lead to fire and explosion accidents in hazardous chemical storage tank areas, contractor management deficiencies also play an important role. Compared with internal employees of the company, contractors are more susceptible to multiple factors such as insufficient safety training and lax qualification review during the operation process, which increases the possibility of accidents.

[0152] Accident prevention measures for fire and explosion accidents in hazardous chemical storage tank areas: (1) For insufficient safety training, enterprises should strengthen safety training for employees. Specifically, they should: increase safety training for special operations; regularly train employees in safety knowledge and skills; strengthen assessments after safety training, and include them in salary assessments and promotion assessments; share safety accident cases to make employees aware of the dangers of their work; combine network experimental simulations to strengthen employees' awareness of accidents; strengthen safety culture construction, such as conducting safety activities and establishing safety bonuses; (2) For deficiencies in system construction and implementation, the following measures should be taken: fully understand the actual situation and needs of production and environment before formulating the system; have leaders take the lead in formulating and verifying the rationality of the system; make timely changes to the system according to changes in national standards, laws and regulations; organize expert seminars to demonstrate the system Rationality; establish a sound system implementation supervision mechanism and reward mechanism to improve employees' enthusiasm for implementing the system; (3) For deficiencies in operating procedures, the following measures should be taken: ensure the popularization and training of operating procedures so that employees can implement them correctly; enterprises should focus on updating and revising operating procedures to ensure the rigor and effectiveness of the procedures; enterprises should regularly organize experts in related fields to hold discussions, deeply analyze the deficiencies and defects of existing operating procedures, and propose improvement measures; when formulating operating procedures, enterprises should comprehensively consider various dangerous operation scenarios to ensure that the procedures cover all possible risk points; at the same time, enterprises should also strengthen communication and exchanges with peers to learn from each other's experience and information; (4) To address the shortcomings of lax approval of operations, the following measures should be taken: improve the approval system for special operations, clarify the responsibilities of the person in charge of safety and the on-site supervisor; establish a supervision mechanism for the approval process of special operations, track and record the entire approval process of special operations, and ensure traceability and effectiveness; the approval system for special operations should be combined with the actual situation of the enterprise, ensuring both safety and production efficiency; the division of responsibilities should be clear, and a safety responsibility system should be implemented; (5) implement good preventive management measures or punishment measures for employees' unsafe behaviors: establish a production safety management department, strengthen internal supervision during the production process; ensure the supply and update of safety equipment and tools; carry out effective prevention and punishment for different positions. Targeted safety training; establish a suitable reward and punishment system and link the occurrence rate of unsafe behaviors with salary promotion; clarify the safety production responsibility system and implement safety responsibilities on each employee; employ qualified and experienced employees; (6) In terms of equipment management deficiencies, the following measures should be taken: clarify the person responsible for each equipment to ensure that equipment problems can be handled in a timely manner; continuously improve the equipment maintenance and inspection and repair system, and formulate a reasonable maintenance plan; regularly conduct equipment risk assessments; establish equipment usage records to dynamically manage equipment and ensure that usage is traceable; provide sufficient investment in equipment management to update testing equipment and introduce professional talents to improve the level of equipment management;(7) Strengthen the management of contractors: conduct a comprehensive review of the qualifications of contractors in accordance with regulations; provide necessary safety education and training to contractors to ensure that they understand and comply with relevant safety regulations; conduct regular safety inspections on construction sites to promptly identify and eliminate potential safety hazards to ensure the safety of the project; and at the same time, an effective incentive mechanism should be established for the management of contractors to encourage contractors to ensure work safety through a reasonable reward and punishment system.

[0153] Obviously, the embodiments described above are only some embodiments of the present application, rather than all embodiments. The preferred embodiments of the present application are given in the accompanying drawings, but they do not limit the patent scope of the present application. The present application can be implemented in many different forms. On the contrary, the purpose of providing these embodiments is to make the understanding of the disclosure of the present application more thorough and comprehensive. Although the present application is described in detail with reference to the aforementioned embodiments, for those skilled in the art, it is still possible to modify the technical solutions recorded in the aforementioned specific implementation methods, or to perform equivalent replacement of some of the technical features therein. Any equivalent structure made using the contents of the specification and drawings of this application, directly or indirectly used in other related technical fields, is similarly within the scope of patent protection of this application.

Claims

1. A method for analyzing the causes of fire and explosion accidents in hazardous chemical storage tank areas based on FRAM, characterized by: The following steps are involved: S1. Process accident cases through grounded theory and extract the causes of accidents; S2. Combine the abstract hierarchy theory to improve the function identification in FRAM analysis and construct the system function network diagram of the tank area; S3. Calculate the grey correlation of the fire and explosion accidents for the accident causal factors obtained by the grey correlation method; S4. Substitute the grey correlation degree as the weight into the Apriori algorithm to mine the association rules of the causal factors of the hazardous chemicals storage tank area; S5. Establish a key functional connection network for fire and explosion accidents in hazardous chemical storage tank areas, and propose targeted preventive measures based on the coupling relationship between causal factors.

2. The method for analyzing the cause of fire and explosion accidents in hazardous chemical storage tank areas based on FRAM according to claim 1 is characterized in that: The accident-causing factors extracted in step S1 include: S1-1. Collect information; S1-2. Construct a case library: By collecting accident information data of hazardous chemical storage tank areas in a certain year, since the collected data should provide more detailed data support for subsequent analysis, only cases with accident descriptions are not selected, and accident investigation reports with detailed records are selected to extract text information to construct a case library; S1-3. Accident data processing: First, the cause of the accident is extracted from the accident report and processed to form data entries; then the jieba toolkit is used to perform Chinese word segmentation processing to obtain the accident Chinese word segmentation entries; S1-4. Summarization of causal factors: Using word segmentation technology, the text data is divided into independent vocabulary units, and the entries with the same semantics are summarized and organized into causal factors.

3. The method for analyzing the cause of fire and explosion accidents in hazardous chemical storage tank areas based on FRAM according to claim 2 is characterized in that: The information collected in step S1-1 includes accident cases and case libraries involved in previously published documents, accident investigation reports on hazardous chemical storage tank areas in published books, the website of the Ministry of Emergency Management of the People's Republic of China, local government emergency management websites, official websites such as the National Bureau of Statistics, and news reports.

4. The method for analyzing the cause of fire and explosion accidents in hazardous chemical storage tank areas based on FRAM according to claim 2 is characterized in that: Step S1-4 specifically includes: By processing the accident cases through text processing and Chinese word segmentation, we obtain the set of accident causative factors, which is the first-level coding of grounded theory analysis. On the basis of the first-level coding, we classify the factors with similarity in the first-level coding in combination with relevant laws and regulations as well as the accident causal model to obtain the second-level coding. Similarly, we classify the second-level coding to obtain the third-level coding, and finally obtain the set of accident causative factors that can be used for subsequent FRAM analysis.

5. The method for analyzing the cause of fire and explosion accidents in hazardous chemical storage tank areas based on FRAM according to claim 1 is characterized in that: Step S2 specifically includes: S2-1. Use the abstract hierarchy theory to divide the tank area system into functional levels; S2-2. Use the three coding levels obtained from grounded theory analysis to identify the different functional levels of the system; S2-3. Finally, the relationship between functions is determined by describing the identified functions in detail to construct a system function network diagram of the tank area; By analyzing the correlation between functions and the integrity of the relationship between functions, a functional network diagram is finally established based on the functional relationship questionnaire.

6. The method for analyzing the causes of fire and explosion accidents in hazardous chemical storage tank areas based on FRAM according to claim 5 is characterized in that: Step S2-1 specifically includes: Combined with the abstract hierarchy theory, the hazardous chemicals storage tank area system is abstracted into two functional layers: the generalized functional layer and the physical functional layer; combined with the characteristics of the hazardous chemicals storage tank area, the scope of each functional layer is further clarified: for the generalized functional layer, this layer further refines the goals of the functional purpose layer into specific functional requirements; for the physical functional layer, the functional requirements of the generalized functional layer are converted into actual operating procedures and steps. At this layer, it is necessary to plan the daily operation process of the hazardous chemicals storage tank area in detail, including the daily operations of the storage tank area, as well as emergency measures for dealing with emergencies.

7. The method for analyzing the causes of fire and explosion accidents in hazardous chemical storage tank areas based on FRAM according to claim 5 is characterized by: Step S2-2 specifically includes: analyzing and describing the hexagonal features of each identified function based on the FRAM model, thereby analyzing the connection between the functions to construct the relationship rules between the functions; if the upstream function affects the downstream function, the relevant function attribute column is marked as 1 in the function relationship table; if there is no connection between the functions, it is marked as 0 in the function relationship table, and there may be a situation where a single function is connected to multiple functions.

8. The method for analyzing the causes of fire and explosion accidents in hazardous chemical storage tank areas based on FRAM according to claim 1 is characterized by: In step S3, the grey correlation method is used to calculate the grey correlation degree, which includes the following steps: S3-1. Establishing the reference sequence and comparison sequence: Under given conditions, if the reference sequence is divided into j, there are k comparison sequences as evaluation factors, and there are i samples in total, then the reference sequence x i (k) represents the actual measured value of i in the kth evaluation factor in the reference sequence; where i ranges from 1 to n, and k ranges from 1 to t; the corresponding comparison sequence is y j (k) represents the specific value of the k evaluation factor in the comparison sequence, where j ranges from 1 to m; S3-2. Standardization of raw data: Before conducting grey correlation analysis, in order to ensure the comparability and accuracy of the data, the data needs to be dimensionally normalized to eliminate the dimensional differences between the data, so that the correlation analysis between different indicators or variables can be conducted fairly. To achieve this goal, the mean method is used to standardize the data and obtain the dimensionless new values ​​of each factor: In the formula, and Respectively represent x i (k) and y j The dimensionless new value of (k); S3-3. Calculate the absolute difference: Determine the maximum and minimum differences between the reference and comparison sequences, expressed as: In the formula, Δ min Indicates the minimum difference between the reference sequence and the comparison sequence, Δ max Indicates the maximum difference between the reference sequence and the comparison sequence, Δ ij (k) represents the difference between the reference sequence and the comparison sequence; and Represent the minimum difference between levels i and j, respectively. and Represent the maximum difference between levels i and j respectively; S3-4. Calculation of correlation coefficients of various factors: Find the relative difference between the ith comparison sequence and the reference sequence in terms of k indicators: In the formula, ε ij (k) is x i Corresponding to y i The correlation coefficient; θ is the resolution coefficient, and its value range is (0,1], but in the calculation, its value is generally 0.5; S3-5. Calculate the grey relational degree: According to all the data obtained above, calculate the contribution of each causal factor to the fire and explosion accident ρ ij :

9. The method for analyzing the causes of fire and explosion accidents in hazardous chemical storage tank areas based on FRAM according to claim 4 is characterized in that: In step S4, the Apriori algorithm mines association rules of causal factors in the hazardous chemicals storage tank area, including the following steps: S4-1. Data preprocessing: Convert the collected non-digital structured text into data labels suitable for Apriori analysis; specifically, construct a table of accident values. If an accident is caused by a certain consistent factor, then in the corresponding accident row and a certain causal factor column, mark it as "1" in the corresponding table cell; conversely, if the accident is not caused by the factor, then mark it as "0" in the table cell. Through labeling processing, a data table suitable for Apriori association analysis is obtained; S4-2. Combine the preprocessed data with the grey correlation coefficient: use the calculated grey correlation coefficient of each factor as the weight of the factor and substitute it into the Apriori algorithm operation process; S4-3. Calculate support and confidence: Based on the constructed weight library, perform a detailed search of the database to determine the set of items that frequently appear in the items of length 1; use combinatorial logic to generate candidate 2-item sets; screen the support of the candidate 2-item sets to determine the real frequent item sets; based on the identified frequent item sets, construct candidate k+1 item sets, and scan the database again to calculate the support of the candidate k+1 item sets to determine the k+1 frequent item sets that meet the set threshold; this process will be repeated until all frequent item sets are fully discovered and identified; on the known frequent item sets, calculate key indicators such as support and confidence; S4-4. Determine strong association rules: Set the support and conditional support to 0.1, and the support confidence to 0.

7. By analyzing the association, we can obtain association rules that are greater than a given threshold, thereby revealing rules with strong association.

10. The method for analyzing the causes of fire and explosion accidents in hazardous chemical storage tank areas based on FRAM according to claim 1 is characterized by: Step S5 specifically includes: S5-1. Establishing the key functional connection network of fire and explosion accidents in hazardous chemical storage tank areas: The functional failure links are judged by mining the accident factor association links through the improved Apriori association rule. The confidence between factors is used to determine the coupling of functions. The connections with functional confidence less than 0.7 are eliminated. Finally, the functional failure links are obtained, and the key functional connection network of fire and explosion accidents in hazardous chemical storage tank areas is established; S5-2. Targeted preventive measures: By analyzing the coupling between causal factors, the occurrence patterns of fire and explosion accidents in hazardous chemical storage tank areas are obtained, and targeted accident prevention measures are proposed.

Images