Risk assessment method and system for dangerous goods at ports based on V-HAZOP model

Abstract

The present application provides a port dangerous goods risk assessment method and system based on the V-HAZOP model, which relates to the field of risk assessment technology. The method comprises: obtaining characteristic information and additional information of dangerous goods at the port, and establishing a basic data set for risk assessment; pre-processing the basic data set for risk assessment based on the HAZOP method, and establishing a risk data index library; analyzing the risk data index library based on the association analysis algorithm, and updating the risk data index library; obtaining final risk index data based on the fault tree analysis model to Bayesian network transformation method; performing visual data modeling based on the final risk index data, and constructing a V-HAZOP model; and using VIEW-Glass multi-dimensional visualization technology to display data obtained by the V-HAZOP model. The present application can solve the management difficulties of existing ports such as a wide variety of dangerous goods, different physical and chemical properties, and high operational requirements, comprehensively optimize the safe production operation of dangerous goods yards, and realize intelligent operation and management of specialized dangerous goods yards at ports.

Description

Technical Field

[0001] The present application relates to the technical field of risk assessment, and in particular to a method and system for risk assessment of dangerous goods at ports based on a V-HAZOP model. Background Art

[0002] With the continuous development of global trade, ports, as important hubs for the import and export of goods, are experiencing a continuous increase in the throughput of dangerous goods. Dangerous goods involve complex processes at ports, including storage, loading and unloading, and transportation. These dangerous goods may include explosives, flammable materials, and toxic substances. Accidents can not only cause serious damage to port facilities and personnel, but can also have catastrophic impacts on the surrounding environment, residents' health, and the smooth flow of international trade. Traditional port dangerous goods risk assessments often lack comprehensive data collection. Information on dangerous goods may focus only on certain characteristics, such as type and quantity, while ignoring other important additional information. Furthermore, this data comes from a wide variety of sources and lacks effective integration methods, making it difficult to ensure data accuracy. Furthermore, traditional risk analysis models often fail to deeply explore correlations between data and accurately identify the interactions between risk factors, resulting in inaccurate risk assessment results. Traditional methods lack effective visualization tools for presenting risk assessment results. This makes it difficult for decision makers to intuitively understand complex risk data and quickly identify key information from numerous risk indicators. Therefore, a method is needed to address these issues.

[0003] In summary, the traditional risk assessment of dangerous goods at ports has problems such as incomplete data collection, low accuracy, inaccurate risk assessment results, and unclear structure, which seriously affect the safe management of cargo operations at ports. Summary of the Invention

[0004] The present disclosure provides a port dangerous goods risk assessment method and system based on the V-HAZOP model to solve the technical problems in the existing technology such as vague port dangerous goods risk assessment, unclear definition of dangerous goods risk status, difficulty in achieving dangerous goods prevention and control, and even safety hazards.

[0005] According to a first aspect of the present disclosure, a method for risk assessment of dangerous goods at ports based on a V-HAZOP model is provided, comprising:

[0006] Acquire characteristic information and additional information of dangerous goods at ports, and establish a basic data set for risk assessment, wherein the characteristic information includes the type, quantity, storage conditions, and transportation route of dangerous goods, and the additional information includes monitoring data of storage yards, warehousing, meteorology, and environment; pre-process the basic data set for risk assessment based on the HAZOP method, divide the HAZOP nodes, integrate the risk indicator data obtained from the HAZOP analysis with the data in the MSDS file library, and establish a risk data indicator library, wherein the HAZOP nodes cover all key links in the operation of dangerous goods storage yards at ports; analyze the risk data indicator library based on the association analysis algorithm, mine the associations between item sets, filter out strong association rules based on the set minimum confidence, analyze the item sets in the strong association rules, filter out indicators with lower risk levels, and update the risk data indicator library, wherein the item sets include indicator data and dangerous results in the risk data indicator library, the minimum confidence is a preset minimum confidence threshold, and the strong association rules are association rules with confidence greater than or equal to the minimum confidence; transform the fault tree analysis model into the Bayesian network method Based on the method, the relationship between fault tree events and Bayesian network nodes is obtained, the logical relationship of the fault tree is mapped to the Bayesian network, the Bayesian network model is constructed, and the final risk indicator data is obtained. The Bayesian network model once again screens the risk indicator data through probability calculation; visual data modeling is performed according to the final risk indicator database, and on the basis of HAZOP analysis, a V-HAZOP model is constructed in combination with the visual data modeling results, and the processed and analyzed indicator data is data-connected with the V-HAZOP model. The construction of the V-HAZOP model is to determine the structure, nodes, guide words and parameter elements of the model, and to establish connections and relationships between nodes; VIEW-Glass multidimensional visualization technology is used to display the data obtained by the V-HAZOP model. The V-HAZOP model includes a decision-making VIEW-Glass model and a special management VIEW-Glass model. The decision-making VIEW-Glass model includes key indicators and dynamics, and the special management VIEW-Glass includes key functional indicators and thematic indicators.

[0007] According to a second aspect of the present disclosure, a port dangerous goods risk assessment system based on the V-HAZOP model is provided, comprising:

[0008] A risk assessment basic data set establishment module is used to obtain characteristic information and additional information of dangerous goods at ports and establish a risk assessment basic data set, wherein the characteristic information includes the type, quantity, storage conditions, and transportation routes of dangerous goods, and the additional information includes monitoring data of storage yards, warehousing, meteorology, and environment; a risk data indicator library establishment module is used to pre-process the risk assessment basic data set based on the HAZOP method, divide HAZOP nodes, integrate the risk indicator data obtained by HAZOP analysis with the data in the MSDS file library, and establish a risk data indicator library, wherein the HAZOP nodes cover all key links in the operation of the dangerous goods storage yard at ports; a risk data indicator library optimization module is used to optimize the risk data indicator library based on the correlation analysis algorithm. The method analyzes the risk data indicator library, mines the association between item sets, screens out strong association rules according to the set minimum confidence, analyzes the item sets in the strong association rules, filters out indicators with lower risk levels, and updates the risk data indicator library, wherein the item sets include indicator data and dangerous results in the risk data indicator library, the minimum confidence is a preset minimum confidence threshold, and the strong association rules are association rules with confidence greater than or equal to the minimum confidence; a Bayesian network risk assessment module, wherein the Bayesian network risk assessment module is used to obtain the relationship between fault tree events and Bayesian network nodes based on the fault tree analysis model to Bayesian network conversion method, map the fault tree logical relationship to the Bayesian network, complete the Bayesian network model construction, and obtain final risk indicator data, and the Bayesian network model once again screens the risk indicator data through probability calculation; V - HAZOP model construction module, the V-HAZOP model construction module is used to perform visual data modeling based on the final risk indicator database. Based on the HAZOP analysis, the V-HAZOP model is constructed in combination with the visual data modeling results. The processed and analyzed indicator data is data-connected with the V-HAZOP model. The construction of the V-HAZOP model is to determine the model structure, nodes, guide words and parameter elements, and establish connections and relationships between nodes; data display module, the data display module is used to use VIEW-Glass multi-dimensional visualization technology to display data obtained by the V-HAZOP model. The V-HAZOP model includes a decision-making VIEW-Glass model and a special management VIEW-Glass model. The decision-making VIEW-Glass model includes key indicators and dynamics, and the special management VIEW-Glass includes key functional indicators and thematic indicators.

[0009] One or more technical solutions provided in the present disclosure have at least the following technical effects or advantages: obtaining characteristic information and additional information of dangerous goods at ports, establishing a basic data set for risk assessment, wherein the characteristic information includes the type, quantity, storage conditions, and transportation routes of dangerous goods, and the additional information includes monitoring data of storage yards, warehousing, meteorology, and environment; preprocessing the basic data set for risk assessment based on the HAZOP method, dividing HAZOP nodes, integrating the risk indicator data obtained from the HAZOP analysis with the data in the MSDS file library, and establishing a risk data indicator library, wherein the HAZOP nodes cover all key links in the operation of dangerous goods storage yards at ports; analyzing the risk data indicator library based on an association analysis algorithm, mining the associations between item sets, screening out strong association rules based on a set minimum confidence, analyzing the item sets in the strong association rules, filtering out indicators with lower risk levels, and updating the risk data indicator library, wherein the item sets include indicator data and hazard results in the risk data indicator library, the minimum confidence is a preset minimum confidence threshold, and the strong association rules are association rules with a confidence greater than or equal to the minimum confidence; Based on the fault tree analysis model to Bayesian network conversion method, the relationship between fault tree events and Bayesian network nodes is obtained, the fault tree logical relationship is mapped to the Bayesian network, and the Bayesian network model is constructed to obtain final risk indicator data. The Bayesian network model further screens the risk indicator data through probability calculation. Visual data modeling is performed based on the final risk indicator database. Based on the HAZOP analysis, the V-HAZOP model is constructed in combination with the visual data modeling results. The processed and analyzed indicator data is data-connected with the V-HAZOP model. The V-HAZOP model construction involves determining the model structure, nodes, guide words and parameter elements, and establishing connections and relationships between nodes. VIEW-Glass multidimensional visualization technology is used to display the data obtained by the V-HAZOP model. The V-HAZOP model includes a decision-making VIEW-Glass model and a special management VIEW-Glass model. The decision-making VIEW-Glass model includes key indicators and dynamics, and the special management VIEW-Glass includes key functional indicators and thematic indicators. It solves the technical problems in the existing technology, such as vague risk assessment of dangerous goods at ports, unclear definition of dangerous goods risk status, difficulty in implementing dangerous goods prevention and control, and even safety hazards, and achieves the technical effect of improving the accuracy of risk indicator data, the level of dangerous goods safety management at ports, and the accuracy of on-site operations.

[0010] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] In order to more clearly illustrate the technical solutions in the present disclosure or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are merely exemplary, and a person of ordinary skill in the art can obtain other drawings based on the provided drawings without any creative work.

[0012] Figure 1 A flow chart of a method for risk assessment of dangerous goods at ports based on the V-HAZOP model provided in an embodiment of the present application;

[0013] Figure 2 A schematic diagram of the structure of a port dangerous goods risk assessment system based on the V-HAZOP model provided in an embodiment of the present application.

[0014] Figure 3 Port flow chart of the port dangerous goods risk assessment method and system based on the V-HAZOP model provided in the embodiment of the present application;

[0015] Explanation of the accompanying symbols: risk assessment basic data set establishment module 11, risk data indicator library establishment module 12, risk data indicator library optimization module 13, Bayesian network risk assessment module 14, V-HAZOP model construction module 15, data display module 16. DETAILED DESCRIPTION

[0016] The following description of exemplary embodiments of the present disclosure is provided in conjunction with the accompanying drawings, which include various details of the embodiments of the present disclosure to facilitate understanding. These details should be considered merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Similarly, for the sake of clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description.

[0017] Example 1: The risk assessment method for dangerous goods at ports based on the V-HAZOP model provided in the embodiment of the present disclosure is referred to Figure 1 For illustration, the methods include:

[0018] Obtain characteristic information and additional information of dangerous goods at ports and establish a basic data set for risk assessment. The characteristic information includes the type, quantity, storage conditions, and transportation routes of dangerous goods; the additional information includes monitoring data of the yard, warehousing, meteorology, and environment.

[0019] Specifically, systematic data collection includes characteristic information of the item itself and additional information about the external environment. Characteristic information primarily includes data on the physical and chemical properties of the item. This type of data should be collected from multiple reliable sources. On the one hand, it is necessary to consult the information provided by the manufacturer of the goods. This information usually contains detailed product specifications, such as basic physical properties such as the state of the substance, color, odor, density, melting point, boiling point, as well as chemical properties such as reactivity, oxidizing properties, reducing properties, and corrosiveness. On the other hand, reference should be made to relevant chemical databases, such as the chemical database of the American Chemical Society (ACS), which compiles a large amount of basic research data on chemical substances. Additional information primarily includes storage condition data. To collect storage condition data, it is necessary to review the warehouse facility documentation at the port, including the type of warehouse, storage temperature and humidity requirements, ventilation conditions, etc.

[0020] Specifically, to ensure data accuracy, collected data must be cross-validated. Physical and chemical property data must be verified by comparing data from multiple chemical databases or data from different manufacturers for the same product. Storage condition data must be verified against actual port facilities. Furthermore, a data inventory must be established, categorized by cargo type, properties, storage, and other factors, to ensure that no data is missed and data integrity is guaranteed.

[0021] Based on the HAZOP method, the risk assessment basic data set is preprocessed, HAZOP nodes are divided, and the risk indicator data obtained from the HAZOP analysis are integrated with the data in the MSDS file library to establish a risk data indicator library. The HAZOP nodes cover all key links of dangerous goods storage at ports.

[0022] Specifically, a detailed review of the established risk assessment basic dataset was conducted to understand the various types of information on dangerous goods at ports contained in the dataset and determine the appropriate HAZOP analysis direction. All key aspects of dangerous goods storage at ports were identified, including the arrival and acceptance of dangerous goods, the sorting and storage of different types of dangerous goods, and the operation of fire protection and safety facilities. Nodes were refined and boundaries were defined. Each key aspect was further refined into specific HAZOP nodes, and subnodes were set under each node. The boundaries of each node, including inputs, outputs, and operating conditions, were determined. Furthermore, corresponding HAZOP guide words were set for each HAZOP node, expressing deviations to facilitate statistical analysis of the indicator data involved in each node and the expression of abnormal status of the indicator data.

[0023] Specifically, the risk indicator data obtained from the HAZOP method analysis needs to be matched with the data in the MSDS file library. For example, if the HAZOP analysis shows that the risk of leakage of a certain dangerous goods may increase due to improper operation during storage, the physical and chemical properties of the goods, emergency treatment measures and other related data can be found in the MSDS file library through the name, UN number and other identification information of the goods. Based on the information obtained from both parties, the matched MSDS data and the HAZOP indicator data are integrated to jointly determine the comprehensive degree of harm to the surrounding environment and personnel in the event of a leak. A unified data structure is established to store the integrated data to form a risk data indicator library. This indicator library can be used for subsequent risk assessment, decision support and other work, such as formulating safety management strategies and emergency plans for dangerous goods storage yards at ports.

[0024] The risk data indicator library is analyzed based on an association analysis algorithm, associations between item sets are mined, strong association rules are screened out according to a set minimum confidence level, item sets in the strong association rules are analyzed, indicators with lower risk levels are filtered out, and the risk data indicator library is updated. The item sets include indicator data and dangerous results in the risk data indicator library, the minimum confidence level is a preset minimum confidence level threshold, and the strong association rules are association rules with a confidence level greater than or equal to the minimum confidence level.

[0025] Specifically, the risk data indicator library is analyzed and processed based on the association analysis algorithm. The main steps include: data preparation, formatting the data in the risk data indicator library to make it suitable for the input requirements of the association analysis algorithm, and determining the item sets in the data set; calculating the support and confidence of the item sets. The support represents the frequency of the item set in the entire data set, and the confidence represents the probability of the occurrence of another item set when one item set occurs.

[0026] Specifically, strong association rules are screened based on a set minimum confidence level. This involves: determining a minimum confidence threshold, which can be preset based on actual needs and experience. This threshold can be determined through historical data analysis, industry standards, or expert opinion; and screening strong association rules. All calculated association rules are traversed, and those with confidence levels greater than or equal to the minimum confidence threshold are selected. These strong association rules reflect highly credible relationships between risk indicators in the risk data indicator library.

[0027] Specifically, the item sets in the strong association rules are analyzed to exclude and filter out indicators with lower risk levels. This step includes: Risk assessment: For each item set in the strong association rules, its risk level is assessed according to predefined risk assessment criteria. For example, if certain indicator data has a low frequency of occurrence, it may be assessed as a low-risk indicator; while some inherent hazardous characteristics of dangerous goods may be assessed as high-risk indicators; indicator filtering: excluding those indicators with lower risk levels. This helps to simplify the risk data indicator library and focus on key indicators with a greater impact on risk. If an item set contains multiple risk indicators, and one of them is a low-risk protective measure, it can be removed from the item set.

[0028] Based on the method of transforming the fault tree analysis model into the Bayesian network, the relationship between the fault tree events and the Bayesian network nodes is obtained, the fault tree logical relationship is mapped to the Bayesian network, the Bayesian network model is constructed, and the final risk indicator data is obtained. The Bayesian network model once again screens the risk indicator data through probability calculation.

[0029] Specifically, the most basic event unit in the fault tree, which cannot be further subdivided, is directly mapped to the root node in the Bayesian network as a basic event. The root node in the Bayesian network is the starting node without a parent node, just as the basic event is the lowest-level element of the fault tree analysis. At the same time, the intermediate events in the fault tree, which are composed of basic events or obtained through certain logical relationship operations, are transformed into intermediate nodes in the Bayesian network. The intermediate node has a parent node in the Bayesian network, which may be the root node or other intermediate node, and its state is affected by the state of the parent node. The top event in the fault tree, which is the final result event of the entire fault tree analysis, needs to be converted into a leaf node in the Bayesian network.

[0030] Specifically, the logical relationship of the fault tree is mapped to the Bayesian network. The specific steps for converting the "logical AND" relationship in the fault tree into the conditional probability of the Bayesian network include:

[0031] S1: Analyze and extract the logical AND relationship in the fault tree. If events A and B jointly cause event C, then A, B and C are in a logical AND relationship. This logical relationship means that C will only occur when A and B occur at the same time.

[0032] S2: Find the nodes corresponding to events A, B, and C in the Bayesian network and determine the connections between the nodes. For the logical AND relationship, the conditional probability distribution of event C in the Bayesian network is expressed as: P ( C | A, B);

[0033] S3: When A=true and B=true, the value of P(C=true|A=true, B=true) is the value determined by the logical AND relationship in the fault tree. This value is usually 1, because if A and B occur at the same time, C will inevitably occur.

[0034] When A=false or B=false, the value of P(C=true|A, B) is 0, because the logical AND relationship requires A and B to occur at the same time.

[0035] Specifically, the logical relationship of the fault tree is mapped to the Bayesian network. The specific steps for converting the "logical OR" relationship in the fault tree into the conditional probability of the Bayesian network include:

[0036] S1: Analyze and extract the logic or relationship in the fault tree. In the fault tree, as long as event D or event E occurs, event F will occur. Then event D and event E lead to intermediate event F through the logic or relationship.

[0037] S2: Find the nodes corresponding to events D, E, and F in the Bayesian network and determine the connection between the nodes. For a logical OR relationship in the Bayesian network, the relationship between the corresponding nodes D, E, and F is expressed as: P(F|D, E);

[0038] S3: When D=true or E=true, P(F=true|D=true, E);

[0039] or

[0040] P(F=true|D,E=true);

[0041] The value of P is determined by the "logical OR" relationship in the fault tree. This value is usually 1, because if D or E occurs, F will inevitably occur.

[0042] When D=false and E=false,

[0043] The value of P ( F=true|D=false, E=false ) is 0.

[0044] Specifically, based on the aforementioned mapping and transformation logic, a directed acyclic graph and conditional probability table are formed, completing the Bayesian network modeling. Risk analysis is performed using the Bayesian network model, and data from the risk data indicator library is input into the model. Once the optimized risk data is obtained, the model calculates the probability value of each node based on the data and classifies the data based on the probability relationships between the nodes. The final risk assessment results are then derived based on the importance of the root node, resulting in the final risk indicator data.

[0045] Visual data modeling is performed based on the final risk indicator database. Based on the HAZOP analysis, a V-HAZOP model is constructed in combination with the visual data modeling results. The processed and analyzed indicator data is then data-connected with the V-HAZOP model. Constructing the V-HAZOP model involves determining the model structure, nodes, guide words, and parameter elements, and establishing connections and relationships between nodes.

[0046] Specifically, we first need to gain a deep understanding of the meaning and characteristics of risk indicator data categories, organize data of different categories, and ensure its completeness and accuracy. We then perform data preprocessing, including cleaning, removing outliers and duplicates, and standardizing data to transform data of varying magnitudes into comparable formats for subsequent modeling.

[0047] Specifically, select a visualization modeling method and build a visualization model. Choose an appropriate visualization modeling method based on the nature of the risk indicator data. For multi-dimensional risk indicator data, it is necessary to use complex visualization methods such as radar charts and Sankey diagrams, combined with modeling techniques such as cluster analysis and principal component analysis. Use radar charts to display the performance of different devices in multiple risk indicator dimensions. Through cluster analysis, classify devices with similar risk conditions into one category to build a visualization-based clustering model. When building a model using the selected visualization method and modeling technique, the relationship between different risk indicator data categories must be clearly expressed. For multi-dimensional visualization models, the coordinate axes or hierarchies must be accurately set to reflect the different dimensions of risk indicators and the relationships between them.

[0048] Specifically, the model structure is determined. Based on the HAZOP methodology and subsequent optimization, the basic structure of the V-HAZOP model is determined. The HAZOP analysis has already assessed the risks of various components within the system, and the structure of the V-HAZOP model should reflect the relationships between these components. Data connections are established. Data mapping and conversion are performed based on the determined V-HAZOP model, establishing the actual data connection. Database connection technologies or file transfer protocols are used to transfer and connect data from the analyzed indicator data storage location to the V-HAZOP model.

[0049] The data obtained by the V-HAZOP model is displayed using VIEW-Glass multi-dimensional visualization technology. The V-HAZOP model includes a decision-making VIEW-Glass model and a special management VIEW-Glass model. The decision-making VIEW-Glass model includes key indicators and dynamics, and the special management VIEW-Glass includes key functional indicators and thematic indicators.

[0050] Specifically, design the interactive interface and interactive elements within the VIEW - Glass visualization interface to display inter-model connections. Set up a navigation menu or sidebar with links to related models and indicators. Ensure that interactive operations are intuitive and convenient. For example, use icons or tooltips to guide users through interactive operations, and display available actions when the mouse hovers over relevant elements.

[0051] Specifically, plan the layout of the overall view, logically placing the visualizations of the decision-making and specialized management VIEW-Glass models within a single interface. Visualize key indicators for the decision-making model at the top of the interface to provide an overview of the overall situation. Visualize functional and thematic indicators for the specialized management model at the bottom, distinguishing them through columns or groupings. Consider the spatial proportions and visual balance between different visualization elements to ensure the overall aesthetics and readability of the view. Avoid overly large or small visualization elements, which can affect the overall visual effect.

[0052] Specifically, we integrated the visualization elements, established connections and interactive relationships between them, and used a unified visual style and color scheme to ensure consistency across the entire VIEW-Glass multi-dimensional visualization system.

[0053] The method provided in the embodiment of the present disclosure also includes:

[0054] Systematically collect data on dangerous goods at ports, including but not limited to the physical and chemical properties of the goods, storage conditions, and historical accident records;

[0055] Divide the handling processes of dangerous goods loading, unloading, storage and transportation at ports into HAZOP nodes to ensure comprehensive coverage of all possible hazardous scenarios during port transportation;

[0056] Based on the data of dangerous goods at the port, according to the detailed chemical and safety properties of dangerous goods in the MSDS file library, the danger of dangerous goods operation at each node port is analyzed, and a dangerous goods risk data indicator library is constructed.

[0057] Furthermore, HAZOP nodes are divided into multiple specific areas, including loading and unloading nodes and storage nodes. The loading and unloading node encompasses the transfer of cargo from the transport vehicle to the loading and unloading platform, the connection of loading and unloading equipment to the cargo, and the securing and supporting of cargo during loading and unloading. The storage node covers aspects such as warehouse entry operations, cargo stacking within the warehouse, storage monitoring, and warehouse safety and protection facilities. By categorizing these nodes in detail, we can comprehensively cover all possible hazardous scenarios during the transportation of dangerous goods at ports.

[0058] Furthermore, based on the nodes divided by HAZOP, a dangerous goods risk index library is constructed in combination with the MSDS file library. The construction of the risk index library should comprehensively consider the dangerousness of the goods, the complexity of the operation, and environmental factors. For the dangerousness of the goods, the corresponding indicators can be determined according to the dangerous classification in the MSDS file. The MSDS file library contains detailed chemical and safety property information of various dangerous goods. Aspects of operational complexity, such as the number of equipment involved in the loading and unloading process, and the operating steps can be used as indicators of operational complexity. Environmental factors, such as the population density around the storage warehouse, should also be included in the risk index library. Based on the information obtained above, the dangerousness of each node of the dangerous goods operation at the export port can be analyzed, and a preliminary risk index database can be compiled.

[0059] The method provided in the embodiment of the present disclosure also includes:

[0060] Read the data set in the risk data indicator library, set each risk indicator data as a separate element, calculate the number of occurrences of each element, and sort them from high to low according to the number of occurrences;

[0061] Starting from a single element, gradually generate item sets containing multiple elements;

[0062] Calculate the support of each item set in the data set, which is used to measure the frequency of the item set in the entire data set;

[0063] Filter out frequent item sets according to the preset minimum support, where the frequent item sets are item sets whose support is greater than or equal to the minimum support;

[0064] For each filtered frequent item set, generate an association rule and calculate the confidence, which is the probability that the result part of the right element of the association rule appears under the condition that the condition part of the left element appears;

[0065] Filtering out strong association rules according to a preset minimum confidence, wherein the strong association rules are association rules whose confidence is greater than or equal to the minimum confidence;

[0066] Obtain a strongly associated item set, filter low-risk indicator data, and update the risk data indicator library. The strongly associated item set is an item set with a strong association rule.

[0067] Specifically, we first read a dataset from the risk indicator database and perform data cleansing, including removing duplicate records and handling missing values, to ensure data accuracy and consistency. Each indicator data point is then set as a separate element, with each different risk type as a separate element. We then traverse the entire dataset, counting the number of occurrences of each element and using a data structure to store each element and its corresponding occurrence count. Finally, we sort these elements by occurrence count, from highest to lowest. This is achieved by sorting the database that stores the elements and their occurrence counts.

[0068] We then generate itemsets containing multiple elements, starting with a single element and gradually building up multiple itemsets. First, consider an itemset containing two elements. By looping through the set of single elements, we generate all possible two-element itemsets. Then, in a similar manner, we generate three-element itemsets from the two-element itemset, and so on, completing the construction of the itemset, for example, {a}, {a, b}, {a, b, c}.

[0069] Furthermore, for each item set, its support in the dataset is calculated. The support of an association rule refers to the proportion of transactions that contain both the antecedent and the consequent among all transactions. Support reflects the prevalence of the pattern appearing in the rule and the frequency with which the rule appears in the dataset. If the support of a rule exceeds a preset minimum support threshold, the rule is considered frequent and a meaningful candidate rule. Support is calculated by counting the number of times the item set appears in the entire dataset and dividing it by the total number of records in the dataset. Following this method, support is calculated for all generated item sets (from single-element to multi-element item sets). Frequent item sets are filtered out based on the set minimum support. For example, the minimum support is set to 0.03, and item sets with support greater than or equal to 0.03 are filtered as frequent item sets. This process can be performed by traversing all item sets for which support has been calculated, comparing their support with the minimum support value, and retaining the item sets that meet the criteria.

[0070] Furthermore, for each filtered frequent item set, all possible association rules are generated and the confidence of each association rule is calculated. Association rules are generally expressed in the form X→Y, where X and Y are item sets. Confidence is a key metric, indicating the proportion of transactions that contain the rule's antecedent also containing the rule's consequent. The confidence calculation formula is: Confidence (X→Y) = Support (X∪Y) / Support (X), where Support (X∪Y) represents the proportion of transactions containing both X and Y to the total number of transactions, and Support (X) represents the proportion of transactions containing X to the total number of transactions.

[0071] Furthermore, strong association rules are filtered out based on the set minimum confidence. For example, the minimum confidence is set to 0.3, and association rules with confidence greater than or equal to 0.3 are filtered out as strong association rules. Similarly, by traversing all association rules with calculated confidence, their confidence is compared with the minimum confidence value, and the association rules that meet the conditions are retained. Based on the final set of strongly correlated items, low-risk indicator data is identified. If the strong association rules indicate that certain indicator combinations are highly correlated with low-risk situations, then the indicators in these indicator combinations may be considered low-risk indicator data. The records corresponding to these low-risk indicator data are deleted from the data set, and the low-risk indicator data in the risk data indicator library is eliminated. Finally, the risk data indicator library is updated so that the data in the library reflects the results of the filtering and screening, so that subsequent risk analysis and decision-making operations can be based on the updated and more accurate risk indicator data.

[0072] The method provided in the embodiment of the present disclosure also includes:

[0073] Construct a fault tree model, determine the node corresponding to each event in the fault tree in the Bayesian network, convert the logical relationship of the fault tree into a conditional probability relationship in the Bayesian network, and achieve a complete mapping of the node relationship and conditional probability relationship of each event in the fault tree on the Bayesian network, forming a directed acyclic graph and a conditional probability table, and completing the Bayesian network modeling. The directed acyclic graph is a structural representation of the Bayesian network;

[0074] Collect data, calculate the probability of the bottom event of the fault tree model, obtain the prior probability of the Bayesian network, and infer the probabilities of other nodes based on the prior probability of the root node and the conditional probability relationship. At the same time, train the basic data to obtain the importance of the root node of the Bayesian network. Evaluate safety hazards based on the probability and importance of the node, identify key operational management risks, and integrate and summarize the final risk indicator data;

[0075] The bottom event in the fault tree model is mapped to the root node of the Bayesian network model, the middle event in the fault tree model is mapped to the middle node of the Bayesian network model, and the top event in the fault tree model is mapped to the leaf node of the Bayesian network model;

[0076] Generate a conditional probability table by converting the logical relationship of the intermediate events of the fault tree model into a basic logic gate;

[0077] Through mapping relationships, a directed acyclic graph and conditional probability table are formed, which are combined into a Bayesian network for subsequent risk assessment;

[0078] The root node probability importance and key importance are calculated based on the absolute value of the probability difference between the root node under the conditions of leaf node occurrence and non-occurrence, and the ratio of the leaf node occurrence probability change rate to the root node occurrence probability change rate.

[0079] Specifically, each event unit in the fault tree is mapped to a node in a Bayesian network, forming a directed acyclic graph. Simultaneously, the logical relationships within the fault tree are mapped to the Bayesian network, and a complete conditional probability table is constructed based on the logical relationships within the fault tree. The conditional probability table lists the probability of each node taking the value 1 or 0 under different combinations of parent node values.

[0080] Furthermore, to determine the bottom event probability of the fault tree, we collected data from various sources, including historical records and equipment maintenance logs, and calculated the ratio of the number of specific faults to the total operating time to obtain the bottom event probability. This probability value will serve as the prior probability for the Bayesian network.

[0081] Furthermore, the probabilities of other nodes are inferred based on the prior probability and conditional probability. After the fault tree is converted into a Bayesian network, the conditional probability relationship is constructed based on the logical relationship in the fault tree. For example, if event A and event B are connected to event C through an AND gate in the fault tree, then in the Bayesian network:

[0082] P(C=1|A=1,B=1)=1 、

[0083] P(C=1|A=0,B=1)=0,

[0084] P(C=1|A=1,B=0)=0,

[0085] P(C=1|A=0,B=0)=0;

[0086] Similarly, if event A and event B are connected to event C through an OR gate in the fault tree, then in the Bayesian network:

[0087] P(C=1|A=1,B=1)=1,

[0088] P ( C = 1 | A = 0, B = 1 ) = 1 ,

[0089] P(C=1|A=1,B=0)=1,

[0090] P(C=1|A=0,B=0)=0;

[0091] For more complex logical relationships, a complete conditional probability table is gradually constructed by analyzing the logic gate combinations in the fault tree. Then, the Bayesian network inference algorithm, such as variable elimination or Gibbs sampling, is used to infer the probabilities of other nodes. The specific steps are as follows:

[0092] S1: Confirm the prior probability and conditional probability relationship between the root nodes A, B and the intermediate node C.

[0093] like:

[0094] P(A=1)=0.3,

[0095] P(B=1)=0.4,

[0096] P(C=1|A=0,B=1)=0.3 、

[0097] P(C=1|A=1,B=1)=0.8,

[0098] P(C=1|A=1,B=0)=0.4,

[0099] P(C=1|A=0,B=0)=0.1;

[0100] S2: Calculate the value of P(C=1):

[0101] P(C=1)=P(C=1|A=1,B=1)P(A=1)P(B=1)+

[0102] P(C=1|A=0,B=1)P(A=0)P(B=1)+P(C=1|A=1,B=0)P(A=1)P(B=0)+P(C=1|A=0,B=0)P(A=0)P(B=0).

[0103] S3: Calculate first:

[0104] P(A=0)=1-P(A=1)=0.7,

[0105] P(B=0)=1-P(B=1)=0.6;

[0106] S4: Substituting into the above formula we can get:

[0107] P(C=1)=0.8×0.3×0.4+0.3×0.7×0.4+0.4×0.3×0.6+0.1×0.7×0.6;

[0108] Furthermore, to calculate the probability importance, for the root node Xi and the leaf node T, first calculate the probability of the leaf node occurring P(T=1). Then, while fixing the probabilities of other root nodes, set the probability of Xi to 1 and recalculate P(T=1|Xi=1).

[0109] Probability importance formula:

[0110] Ip(Xi)=∣P(T=1∣Xi=1)-P(T=1)∣;

[0111] Calculate the key importance, let ΔP(Xi) be the small change in the probability of the root node Xi, and ΔP(T) be the corresponding change in the probability of the leaf node T.

[0112] Key importance:

[0113] Ic(Xi)=ΔP(Xi) / P(Xi)ΔP(T) / P(T);

[0114] Furthermore, security risks are assessed based on the probability and importance of each node. If a root node has a high probability importance, it may still be a significant security risk, even if its prior probability is low. Based on the nodes identified as having high security risks, the indicator data associated with these events is compared with the existing risk data database. The same indicator data is retained to form a final risk indicator database. This final data is presented in a visual format, allowing decision makers to clearly understand the system's risk status and formulate appropriate risk management strategies.

[0115] The method provided in the embodiment of the present disclosure also includes:

[0116] Customize the visualization model and delineate the V-HAZOP model into a decision-making VIEW-Glass model and a special management VIEW-Glass. The decision-making VIEW-Glass model includes the enterprise management VIEW-Glass, safety management VIEW-Glass, operations management VIEW-Glass, technical management VIEW-Glass, financial management VIEW-Glass, market management VIEW-Glass, and comprehensive management VIEW-Glass.

[0117] Integrate and analyze risk indicator data and map it to various VIEW-Glass models to achieve data visualization;

[0118] The real-time status of the decision-making VIEW-Glass model and the special management VIEW-Glass is displayed in the form of multi-dimensional charts.

[0119] Specifically, a customized visual V-HAZOP model was developed, encompassing both a decision-making VIEW-Glass and a specialized management VIEW-Glass. The decision-making VIEW-Glass was further divided into the enterprise management VIEW-Glass, the security management VIEW-Glass, the operations management VIEW-Glass, the technical management VIEW-Glass, the financial management VIEW-Glass, the marketing management VIEW-Glass, and the comprehensive management VIEW-Glass. A classification algorithm was used to categorize indicators in the existing risk database. This filtered and classified risk indicator data was then mapped to the VIEW-Glass model using database connection technology or file transfer protocols. Multidimensional charts were used to visualize data, showcasing the data stored in various VIEW-Glass models through the construction of visual charts such as bar charts, pie charts, line charts, radar charts, Gantt charts, and comprehensive scoring dashboards.

[0120] Example 2 is based on the same inventive concept as the port dangerous goods risk assessment method based on the V-HAZOP model in the previous embodiment. Figure 2 As shown, the present disclosure also provides a port dangerous goods risk assessment system based on the V-HAZOP model, the system comprising:

[0121] The risk assessment basic data set establishment module 11 is used to obtain characteristic information and additional information of dangerous goods at the port and establish a risk assessment basic data set. The characteristic information includes the type, quantity, storage conditions, and transportation route of dangerous goods. The additional information includes monitoring data of the storage yard, warehousing, meteorology, and environment.

[0122] The risk data indicator library establishment module 12 is used to pre-process the risk assessment basic data set based on the HAZOP method, divide the HAZOP nodes, integrate the risk indicator data obtained from the HAZOP analysis with the data in the MSDS file library, and establish a risk data indicator library. The HAZOP nodes cover all key links in the operation of the port dangerous goods yard;

[0123] The risk data indicator library optimization module 13 is used to analyze the risk data indicator library based on the association analysis algorithm, mine the associations between item sets, filter out strong association rules based on the set minimum confidence level, analyze the item sets in the strong association rules, filter out indicators with lower risk levels, and update the risk data indicator library. The item sets include indicator data and risk results in the risk data indicator library. The minimum confidence level is a preset minimum confidence threshold. The strong association rules are association rules with a confidence level greater than or equal to the minimum confidence level.

[0124] The Bayesian network risk assessment module 14 is used to obtain the relationship between the fault tree events and the Bayesian network nodes based on the fault tree analysis model to Bayesian network conversion method, map the fault tree logical relationship to the Bayesian network, complete the Bayesian network model construction, and obtain the final risk indicator data. The Bayesian network model further screens the risk indicator data through probability calculation;

[0125] A V-HAZOP model construction module 15 is used to perform visual data modeling based on the final risk indicator database, construct a V-HAZOP model based on the HAZOP analysis and the visual data modeling results, and connect the processed and analyzed indicator data with the V-HAZOP model. The construction of the V-HAZOP model includes determining the structure, nodes, guide words and parameter elements of the model, and establishing connections and relationships between the nodes.

[0126] The data display module 16 is used to use VIEW-Glass multi-dimensional visualization technology to display data obtained from the V-HAZOP model. The V-HAZOP model includes a decision-making VIEW-Glass model and a special management VIEW-Glass model. The decision-making VIEW-Glass model includes key indicators and dynamics, and the special management VIEW-Glass includes key functional indicators and special indicators.

[0127] Furthermore, the risk data indicator library establishment module 12 also includes a HAZOP application module:

[0128] Systematically collect data on dangerous goods at ports, including but not limited to the physical and chemical properties of the goods, storage conditions, and historical accident records;

[0129] Divide the handling processes of dangerous goods loading, unloading, storage and transportation at ports into HAZOP nodes to ensure comprehensive coverage of all possible hazardous scenarios during port transportation;

[0130] Based on the data of dangerous goods at the port, according to the detailed chemical and safety properties of dangerous goods in the MSDS file library, the danger of dangerous goods operation at each node port is analyzed, and a dangerous goods risk data indicator library is constructed.

[0131] Furthermore, the risk data indicator library optimization module 13 also includes an association rule mining module:

[0132] Read the data set in the risk data indicator library, set each risk indicator data as a separate element, calculate the number of occurrences of each element, and sort them from high to low according to the number of occurrences;

[0133] Starting from a single element, gradually generate item sets containing multiple elements;

[0134] Calculate the support of each item set in the data set, which is used to measure the frequency of the item set in the entire data set;

[0135] Filter out frequent item sets according to the preset minimum support, where the frequent item sets are item sets whose support is greater than or equal to the minimum support;

[0136] For each filtered frequent item set, generate an association rule and calculate the confidence, which is the probability that the result part of the right element of the association rule appears under the condition that the condition part of the left element appears;

[0137] Filtering out strong association rules according to a preset minimum confidence, wherein the strong association rules are association rules whose confidence is greater than or equal to the minimum confidence;

[0138] Obtain a strongly associated item set, filter low-risk indicator data, and update the risk data indicator library. The strongly associated item set is an item set with a strong association rule.

[0139] Furthermore, the Bayesian network risk assessment module 14 also includes a Bayesian network construction module based on the fault tree logical relationship:

[0140] Construct a fault tree model, determine the node corresponding to each event in the fault tree in the Bayesian network, convert the logical relationship of the fault tree into a conditional probability relationship in the Bayesian network, and achieve a complete mapping of the node relationship and conditional probability relationship of each event in the fault tree on the Bayesian network, forming a directed acyclic graph and a conditional probability table, and completing the Bayesian network modeling. The directed acyclic graph is a structural representation of the Bayesian network;

[0141] Collect data, calculate the probability value of the bottom event of the fault tree model, obtain the prior probability of the Bayesian network, and infer the probabilities of other nodes based on the prior probability of the root node and the conditional probability relationship. At the same time, train basic data to obtain the importance of the root node of the Bayesian network. Evaluate safety hazards based on the probability and importance of the nodes, identify key operational management risks, and integrate and summarize the final risk indicator data.

[0142] Furthermore, the Bayesian network risk assessment module 14 also includes a fault tree mapping module:

[0143] The bottom event in the fault tree model is mapped to the root node of the Bayesian network model, the middle event in the fault tree model is mapped to the middle node of the Bayesian network model, and the top event in the fault tree model is mapped to the leaf node of the Bayesian network model;

[0144] Generate a conditional probability table by converting the logical relationship of the intermediate events of the fault tree model into a basic logic gate;

[0145] Through mapping relationships, a directed acyclic graph and conditional probability table are formed, which are combined into a Bayesian network for subsequent risk assessment.

[0146] Furthermore, the Bayesian network risk assessment module 14 also includes an importance calculation module:

[0147] The root node probability importance and key importance are calculated based on the absolute value of the probability difference between the root node under the conditions of leaf node occurrence and non-occurrence, and the ratio of the leaf node occurrence probability change rate to the root node occurrence probability change rate.

[0148] Furthermore, the V-HAZOP model building module 15 also includes a VIEW-Glass building display module:

[0149] Customize the visualization model and delineate the V-HAZOP model into a decision-making VIEW-Glass model and a special management VIEW-Glass. The decision-making VIEW-Glass model includes the enterprise management VIEW-Glass, safety management VIEW-Glass, operations management VIEW-Glass, technical management VIEW-Glass, financial management VIEW-Glass, market management VIEW-Glass, and comprehensive management VIEW-Glass.

[0150] Integrate and analyze risk indicator data and map it to various VIEW-Glass models to achieve data visualization;

[0151] The real-time status of the decision-making VIEW-Glass model and the special management VIEW-Glass is displayed in the form of multi-dimensional charts.

[0152] Embodiment 3: Another embodiment of the present invention provides a method and system for port dangerous goods risk assessment based on the V-HAZOP model, including:

[0153] The user end is used to obtain the characteristic information and additional information of dangerous goods at the port, and send the risk assessment basic data set composed of the characteristic information and additional information to the processing end;

[0154] The processing end is used to receive the risk assessment basic data set and perform subsequent data processing, including: pre-processing the risk assessment basic data set based on the HAZOP method to establish a risk data indicator library; analyzing the risk data indicator library based on the association analysis algorithm, excluding and filtering out indicators with low risk levels, and updating the risk data indicator library; obtaining final risk indicator data based on the fault tree analysis model to Bayesian network conversion method; performing visual data modeling based on the final risk indicator data, and connecting the processed and analyzed indicator data with the V-HAZOP model; and finally transmitting the data obtained from the V-HAZOP model to the display end;

[0155] The display terminal is used to display the data obtained from the V-HAZOP model using VIEW-Glass multi-dimensional visualization technology;

[0156] Specifically, the characteristic information obtained by the user end includes the type, quantity, and storage conditions of dangerous goods, and additional information includes monitoring data of the storage yard, storage, weather, and environment;

[0157] Specifically, the processing end pre-processes the risk assessment basic data set based on the HAZOP method, divides the HAZOP nodes, integrates the risk indicator data obtained from the HAZOP analysis with the data in the MSDS file library, and establishes a risk data indicator library. The HAZOP nodes cover all key links in the operation of the port's dangerous goods yard;

[0158] Specifically, the processing end analyzes the risk data indicator library based on an association analysis algorithm, mines associations between item sets, filters out strong association rules based on a set minimum confidence level, analyzes the item sets in the strong association rules, filters out indicators with lower risk levels, and updates the risk data indicator library, wherein the item sets include indicator data and risk results in the risk data indicator library, the minimum confidence level is a preset minimum confidence level threshold, and the strong association rules are association rules with a confidence level greater than or equal to the minimum confidence level;

[0159] Specifically, the processing end, based on the fault tree analysis model to Bayesian network conversion method, obtains the relationship between the fault tree events and the Bayesian network nodes, maps the fault tree logical relationship to the Bayesian network, completes the Bayesian network model construction, and obtains the final risk indicator data. The Bayesian network model further screens the risk indicator data through probability calculation;

[0160] Specifically, the processing end performs visual data modeling based on the final risk indicator database. Based on the HAZOP analysis, the V-HAZOP model is constructed in combination with the visual data modeling results. The processed and analyzed indicator data is then data-connected with the V-HAZOP model. The V-HAZOP model construction involves determining the model structure, nodes, guide words, and parameter elements, and establishing connections and relationships between nodes.

[0161] Specifically, the display end uses VIEW-Glass multi-dimensional visualization technology to display data obtained by the V-HAZOP model. The V-HAZOP model includes a decision-making VIEW-Glass model and a special management VIEW-Glass model. The decision-making VIEW-Glass model includes key indicators and dynamics, and the special management VIEW-Glass includes key functional indicators and thematic indicators.

Claims

1. The risk assessment method for dangerous goods at ports based on the V-HAZOP model is characterized by: The method comprises: Obtain characteristic information and additional information of dangerous goods at ports to establish a basic data set for risk assessment. The characteristic information includes the type, quantity, storage conditions, and transportation routes of dangerous goods. The additional information includes monitoring data of storage yards, warehousing, meteorology, and the environment. Preprocessing the risk assessment basic data set based on the HAZOP method, dividing HAZOP nodes, integrating the risk indicator data obtained from the HAZOP analysis with the data in the MSDS file library, and establishing a risk data indicator library. The HAZOP nodes cover all key links in the operation of the port dangerous goods yard; The risk data indicator library is analyzed based on an association analysis algorithm to mine associations between item sets, strong association rules are screened out according to a set minimum confidence level, item sets in the strong association rules are analyzed, indicators with lower risk levels are filtered out, and the risk data indicator library is updated, wherein the item sets include indicator data and risk results in the risk data indicator library, the minimum confidence level is a preset minimum confidence level threshold, and the strong association rules are association rules with a confidence level greater than or equal to the minimum confidence level, wherein the association analysis algorithm further includes: Read the data set in the risk data indicator library, set each risk indicator data as a separate element, calculate the number of occurrences of each element, and sort them from high to low according to the number of occurrences; Starting from a single element, gradually generate item sets containing multiple elements; Calculate the support of each item set in the data set, which is used to measure the frequency of the item set in the entire data set; Filter out frequent item sets according to the preset minimum support, where the frequent item sets are item sets whose support is greater than or equal to the minimum support; For each filtered frequent item set, generate an association rule and calculate the confidence, which is the probability that the result part of the right element of the association rule appears under the condition that the condition part of the left element appears; Filtering out strong association rules according to a preset minimum confidence, wherein the strong association rules are association rules whose confidence is greater than or equal to the minimum confidence; Obtaining a strongly correlated item set, filtering low-risk indicator data, and updating a risk data indicator library, wherein the strongly correlated item set is an item set with a strong association rule; Based on the method of transforming the fault tree analysis model into the Bayesian network, the relationship between the fault tree events and the Bayesian network nodes is obtained, the fault tree logical relationship is mapped to the Bayesian network, the Bayesian network model is constructed, and the final risk indicator data is obtained. The Bayesian network model once again screens the risk indicator data through probability calculation; Performing visual data modeling based on the final risk indicator database, building a V-HAZOP model based on the HAZOP analysis and combining the visual data modeling results, and connecting the processed and analyzed indicator data with the V-HAZOP model. Building the V-HAZOP model involves determining the model's structure, nodes, guide words, and parameter elements, and establishing connections and relationships between nodes. The data obtained by the V-HAZOP model is displayed using VIEW-Glass multi-dimensional visualization technology. The V-HAZOP model includes a decision-making VIEW-Glass model and a special management VIEW-Glass model. The decision-making VIEW-Glass model includes key indicators and dynamics, and the special management VIEW-Glass includes key functional indicators and thematic indicators.

2. The method according to claim 1, wherein The HAZOP method also includes: Systematically collect data on dangerous goods at ports, including but not limited to the physical and chemical properties of the goods, storage conditions, and historical accident records; Divide the handling processes of dangerous goods loading, unloading, storage and transportation at ports into HAZOP nodes to ensure comprehensive coverage of all possible hazardous scenarios during port transportation; Based on the data of dangerous goods at the port, according to the detailed chemical and safety properties of dangerous goods in the MSDS file library, the danger of dangerous goods operation at each node port is analyzed, and a dangerous goods risk data indicator library is constructed.

3. The method according to claim 1, wherein The method for converting the fault tree analysis model into a Bayesian network also includes: Construct a fault tree model, determine the node corresponding to each event in the fault tree in the Bayesian network, convert the logical relationship of the fault tree into a conditional probability relationship in the Bayesian network, and achieve a complete mapping of the node relationship and conditional probability relationship of each event in the fault tree on the Bayesian network, forming a directed acyclic graph and a conditional probability table, and completing the Bayesian network modeling. The directed acyclic graph is a structural representation of the Bayesian network; Collect data, calculate the probability value of the bottom event of the fault tree model, obtain the prior probability of the Bayesian network, and infer the probabilities of other nodes based on the prior probability of the root node and the conditional probability relationship. At the same time, train basic data to obtain the importance of the root node of the Bayesian network. Evaluate safety hazards based on the probability and importance of the nodes, identify key operational management risks, and integrate and summarize the final risk indicator data.

4. The method according to claim 3, wherein The Bayesian network modeling also includes: The bottom event in the fault tree model is mapped to the root node of the Bayesian network model, the middle event in the fault tree model is mapped to the middle node of the Bayesian network model, and the top event in the fault tree model is mapped to the leaf node of the Bayesian network model; Generate a conditional probability table by converting the logical relationship of the intermediate events of the fault tree model into a basic logic gate; Through mapping relationships, a directed acyclic graph and conditional probability table are formed, which are combined into a Bayesian network for subsequent risk assessment.

5. The method according to claim 3, wherein The obtaining of the importance of the Bayesian network root node also includes: The root node probability importance and key importance are calculated based on the absolute value of the probability difference between the root node under the conditions of leaf node occurrence and non-occurrence, and the ratio of the leaf node occurrence probability change rate to the root node occurrence probability change rate.

6. The method according to claim 1, wherein The V-HAZOP model also includes: Customize the visualization model and delineate the V-HAZOP model into a decision-making VIEW-Glass model and a special management VIEW-Glass. The decision-making VIEW-Glass model includes the enterprise management VIEW-Glass, safety management VIEW-Glass, operations management VIEW-Glass, technical management VIEW-Glass, financial management VIEW-Glass, market management VIEW-Glass, and comprehensive management VIEW-Glass. Integrate and analyze risk indicator data and map it to various VIEW-Glass models to achieve data visualization; The real-time status of the decision-making VIEW-Glass model and the special management VIEW-Glass is displayed in the form of multi-dimensional charts.

7. The port dangerous goods risk assessment system based on the V-HAZOP model is characterized by: A system for implementing the port dangerous goods risk assessment method based on the V-HAZOP model according to any one of claims 1 to 6, comprising: A risk assessment basic data set establishment module is used to obtain characteristic information and additional information of dangerous goods at ports and establish a risk assessment basic data set. The characteristic information includes the type, quantity, storage conditions, and transportation route of dangerous goods, and the additional information includes monitoring data of storage yards, warehousing, meteorology, and the environment; A risk data indicator library establishment module is used to pre-process the risk assessment basic data set based on the HAZOP method, divide the HAZOP nodes, integrate the risk indicator data obtained from the HAZOP analysis with the data in the MSDS file library, and establish a risk data indicator library. The HAZOP nodes cover all key links in the operation of the port dangerous goods yard; A risk data indicator library optimization module is configured to analyze the risk data indicator library based on an association analysis algorithm, mine associations between item sets, screen out strong association rules based on a set minimum confidence level, analyze the item sets in the strong association rules, filter out indicators with lower risk levels, and update the risk data indicator library. The item sets include indicator data and risk results in the risk data indicator library. The minimum confidence level is a preset minimum confidence level threshold. The strong association rules are association rules with a confidence level greater than or equal to the minimum confidence level. The risk data indicator library optimization module further includes an association rule mining module to perform the following steps: Read the data set in the risk data indicator library, set each risk indicator data as a separate element, calculate the number of occurrences of each element, and sort them from high to low according to the number of occurrences; Starting from a single element, gradually generate item sets containing multiple elements; Calculate the support of each item set in the data set, which is used to measure the frequency of the item set in the entire data set; Filter out frequent item sets according to the preset minimum support, where the frequent item sets are item sets whose support is greater than or equal to the minimum support; For each filtered frequent item set, generate an association rule and calculate the confidence, which is the probability that the result part of the right element of the association rule appears under the condition that the condition part of the left element appears; Filtering out strong association rules according to a preset minimum confidence, wherein the strong association rules are association rules whose confidence is greater than or equal to the minimum confidence; Obtaining a strongly correlated item set, filtering low-risk indicator data, and updating a risk data indicator library, wherein the strongly correlated item set is an item set with a strong association rule; A Bayesian network risk assessment module is used to obtain the relationship between fault tree events and Bayesian network nodes based on the fault tree analysis model to Bayesian network conversion method, map the fault tree logical relationship to the Bayesian network, complete the Bayesian network model construction, and obtain the final risk indicator data. The Bayesian network model further screens the risk indicator data through probability calculation; A V-HAZOP model construction module is used to perform visual data modeling based on the final risk indicator database, construct a V-HAZOP model based on the HAZOP analysis and the visual data modeling results, and connect the processed and analyzed indicator data with the V-HAZOP model. The construction of the V-HAZOP model includes determining the model structure, nodes, guide words and parameter elements, and establishing connections and relationships between nodes; A data display module is used to display data obtained from the V-HAZOP model using VIEW-Glass multi-dimensional visualization technology. The V-HAZOP model includes a decision-making VIEW-Glass model and a special management VIEW-Glass model. The decision-making VIEW-Glass model includes key indicators and dynamics, and the special management VIEW-Glass includes key functional indicators and thematic indicators.

8. The port dangerous goods risk assessment system based on the V-HAZOP model according to claim 7 is characterized in that: Also included are HAZOP application modules to perform the following steps: Systematically collect data on dangerous goods at ports, including but not limited to the physical and chemical properties of the goods, storage conditions, and historical accident records; Divide the handling processes of dangerous goods loading, unloading, storage and transportation at ports into HAZOP nodes to ensure comprehensive coverage of all possible hazardous scenarios during port transportation; Based on the data of dangerous goods at the port, according to the detailed chemical and safety properties of dangerous goods in the MSDS file library, the danger of dangerous goods operation at each node port is analyzed, and a dangerous goods risk data indicator library is constructed.

9. The port dangerous goods risk assessment system based on the V-HAZOP model according to claim 7, characterized in that: It also includes a Bayesian network building module based on the fault tree logic relationship to perform the following steps: Construct a fault tree model, determine the node corresponding to each event in the fault tree in the Bayesian network, convert the logical relationship of the fault tree into a conditional probability relationship in the Bayesian network, and achieve a complete mapping of the node relationship and conditional probability relationship of each event in the fault tree on the Bayesian network, forming a directed acyclic graph and a conditional probability table, and completing the Bayesian network modeling. The directed acyclic graph is a structural representation of the Bayesian network; Collect data, calculate the probability value of the bottom event of the fault tree model, obtain the prior probability of the Bayesian network, and infer the probabilities of other nodes based on the prior probability of the root node and the conditional probability relationship. At the same time, train basic data to obtain the importance of the root node of the Bayesian network. Evaluate safety hazards based on the probability and importance of the nodes, identify key operational management risks, and integrate and summarize the final risk indicator data.

10. The port dangerous goods risk assessment system based on the V-HAZOP model according to claim 9, characterized in that: A Fault Tree Mapping module is also included to perform the following steps: The bottom event in the fault tree model is mapped to the root node of the Bayesian network model, the middle event in the fault tree model is mapped to the middle node of the Bayesian network model, and the top event in the fault tree model is mapped to the leaf node of the Bayesian network model; Generate a conditional probability table by converting the logical relationship of the intermediate events of the fault tree model into a basic logic gate; Through mapping relationships, a directed acyclic graph and conditional probability table are formed, which are combined into a Bayesian network for subsequent risk assessment.

11. The port dangerous goods risk assessment system based on the V-HAZOP model according to claim 9, characterized in that: It also includes an importance calculation module to perform the following steps: The root node probability importance and key importance are calculated based on the absolute value of the probability difference between the root node under the conditions of leaf node occurrence and non-occurrence, and the ratio of the leaf node occurrence probability change rate to the root node occurrence probability change rate.

12. The port dangerous goods risk assessment system based on the V-HAZOP model according to claim 7, characterized in that: Also included is the VIEW - Glass build display module to perform the following steps: Customize the visualization model and delineate the V-HAZOP model into a decision-making VIEW-Glass model and a special management VIEW-Glass. The decision-making VIEW-Glass model includes the enterprise management VIEW-Glass, safety management VIEW-Glass, operations management VIEW-Glass, technical management VIEW-Glass, financial management VIEW-Glass, market management VIEW-Glass, and comprehensive management VIEW-Glass. Integrate and analyze risk indicator data and map it to various VIEW-Glass models to achieve data visualization; The real-time status of the decision-making VIEW-Glass model and the special management VIEW-Glass is displayed in the form of multi-dimensional charts.

Images