Dangerous chemical tanker transportation safety risk monitoring and early warning system and method

Abstract

The invention discloses a dangerous chemical tanker transportation safety risk monitoring and early warning system and method based on a knowledge graph, and mainly solves the problems that in the prior art, monitoring parameters are single, safety data lacks an effective mining means, and an accident case library cannot be directly used for accident early warning. The method comprises the following modules: a hazardous chemical substance tank car transportation data acquisition module, a hazardous chemical substance tank car transportation law and regulation, standard specification and historical accident data acquisition module, a data preprocessing module, a knowledge graph module, a hazardous chemical substance tank car transportation safety risk situation analysis module and a dynamic updating and feedback module. A hazardous chemical substance tank car transportation risk situation visualization module; according to the method, safety risk analysis and accident cause mining of dangerous chemical tank car transportation can be realized, the intelligence and accuracy of dangerous chemical tank car transportation safety risk management and control can be obviously enhanced, the possibility of accidents is reduced, and the safety and stability of overall production are improved.

Description

Technical Field

[0001] This application relates to the technical field of hazardous chemical transportation management, and particularly to a safety risk monitoring and early warning system and method for hazardous chemical tank truck transportation. Background Art

[0002] Due to their special properties such as flammability, explosiveness, toxicity, instability, etc., hazardous chemicals make it easy for hazardous chemical tank trucks to have serious traffic accidents such as combustion, explosion, and poisoning during the transportation process. Currently, most of the monitoring systems on the market only monitor a small number of parameters such as the speed, acceleration, and attitude of the tank truck, and do not comprehensively and real-time dynamically monitor multiple parameters of "people-vehicle-cargo-environment".

[0003] Currently, most monitoring systems simply judge whether the current operating state of the hazardous chemical tank truck is normal through thresholds, with relatively low accuracy and reliability. And there is a lack of scientific risk assessment, prediction, and early warning models in the existing technology, and a lack of intelligent and scientific risk analysis and early warning means to provide scientific decision-making support for major risk control. Therefore, in order to realize the functions of intelligent, automated, and precise risk factor identification, multi-dimensional risk assessment, and risk early warning under the intelligent supervision mode, it is necessary to construct a theoretical framework structure of a safety risk intelligent supervision system to provide an effective supervision method for the intelligent supervision mode of the safety risk of hazardous chemical tank truck transportation, thereby improving the safety risk supervision efficiency of the hazardous chemical tank truck transportation industry. Summary of the Invention

[0004] This application provides a safety risk monitoring and early warning system and method for hazardous chemical tank truck transportation, and its advantage is that it can significantly enhance the intelligence, accuracy, and comprehensiveness of the safety risk monitoring and early warning of hazardous chemical tank truck transportation, reduce the possibility of accidents, and improve the safety and stability of the overall transportation.

[0005] The technical solution of this application is as follows:

[0006] On the one hand, this application provides a safety risk monitoring and early warning system for hazardous chemical tank truck transportation based on a knowledge graph, which is characterized by including:

[0007] A hazardous chemical tank truck transportation data collection module, which is used to collect data related to the safety of hazardous chemical tank truck transportation in real time, including vehicle status data, driver behavior data, environmental condition data, real-time monitoring data of hazardous chemicals, and safety data during transportation, so as to obtain heterogeneous multi-source hazardous chemical tank truck transportation data;

[0008] The transportation laws, regulations, standards and specifications, and historical accident data collection module for hazardous chemical tank trucks is used to collect data from authoritative data sources according to various published transportation standards and specifications for hazardous chemical tank trucks and historical accident data, and achieve dynamic real-time updates; obtain multi-source transportation safety standards and specifications for hazardous chemical tank trucks;

[0009] The data preprocessing module is used to clean and standardize the collected safe transportation data and standard specification data, and then store them uniformly;

[0010] The dynamic knowledge graph module constructs a dynamic safety risk knowledge graph for hazardous chemical tank truck transportation based on the causal relationship and data correlation of hazardous chemical tank truck transportation data, hazardous chemical tank truck transportation safety standards and specifications, and hazardous chemical tank truck transportation accident data; adopts a causal chain reasoning engine to quantify the causal relationship strength of risk events based on a dynamic Bayesian network; predicts the risk propagation path through a spatio-temporal graph attention network; and dynamically updates the knowledge graph through triples <risk source, causal strength, scope of influence>.

[0011] The hazardous chemical tank truck transportation safety risk situation analysis module conducts risk situation analysis on the real-time collected hazardous chemical tank truck data based on the dynamic safety risk knowledge graph for hazardous chemical tank truck transportation. When the transportation data is greater than the acceptable range of the risk situation, risk events are marked; the risk factor weights are dynamically adjusted based on a sliding window mechanism; and a real-time risk index is generated by combining a multi-sensor data alignment algorithm.

[0012] The dynamic update and feedback module updates the knowledge graph according to the real-time data during transportation, and optimizes the risk factor weight allocation mechanism and the real-time risk index generation algorithm based on the feedback results;

[0013] The hazardous chemical tank truck safety risk situation visualization module combines EGIS map technology to display the global safety risk situation in real time.

[0014] Furthermore, the transportation safety data collection module for hazardous chemical tank trucks obtains key information on transportation and environmental monitoring from different vehicles through multi-source data integration and real-time collection; the integrated data includes environmental parameters such as real-time data from sensors and monitored tank trucks, vehicle performance parameters, personnel and vehicle positioning, atmospheric monitoring, and temperature monitoring.

[0015] Furthermore, the dynamic knowledge graph module analyzes accident causality, safety laws, regulations, standards, and specifications for the transportation of hazardous chemical tank trucks, extracts knowledge from semi-structured data using a rule-based method; extracts knowledge from unstructured data based on the BRT-CRF model; then uses a template matching method based on causal cue words to extract the causes and results of accidents; calculates the weights of the causal chains of risk events based on the dynamic Bayesian network; combines the spatio-temporal graph attention network to predict the risk propagation path; stores the extracted risk information in the Neo4j graph database in the form of triples, and dynamically updates the knowledge graph through the triples <risk source, causal intensity, scope of influence>.

[0016] Furthermore, the safety risk situation analysis module for the transportation of hazardous chemical tank trucks marks risk events including the time of event occurrence, the location of event occurrence, and the spatio-temporal positioning information of the personnel involved in the event.

[0017] On the other hand, the present application provides a method for monitoring and warning the safety risks of transporting hazardous chemical tank trucks based on a knowledge graph, which is characterized by: using the above-mentioned system; the specific steps include:

[0018] S101: Data collection: Collect various data including the safety data of transporting hazardous chemical tank trucks, as well as the relevant laws, regulations, standards, and historical transportation accident data for the transportation of hazardous chemical tank trucks.

[0019] S102: Data preprocessing: Remove noise data and invalid data from the collected data; standardize data from different sources and types to make them have a unified measurement standard; store the processed data in a relational database or a NoSQL database.

[0020] S103: Dynamic knowledge graph construction: Identify safety risk targets based on the formation elements of safety risks, and identify target triples one by one for the safety risk targets, and then construct a dynamic knowledge graph of safety risks.

[0021] S104: Safety risk situation analysis, identify safety risks based on the safety risk knowledge graph for the collected real-time data, mark them, evaluate the identified risks, and provide decision-making suggestions.

[0022] S105: Update the knowledge graph according to the transportation data, and optimize the risk factor weight allocation mechanism and the real-time risk index generation algorithm based on the feedback results.

[0023] S106: Visualization of the risk situation, intuitively display the risk situation.

[0024] Furthermore, step S103 specifically includes:

[0025] Preprocessing of data: First, preprocess laws, regulations, industry standards or specifications, accident reports, and operation logs, including word segmentation, stop word removal, and named entity annotation operations;

[0026] Entity extraction: Use the pre-trained BERT model to identify and extract key entities from the text;

[0027] Domain dictionary matching: Construct a special dictionary for the field of hazardous chemical tank truck transportation, and assist in identifying specific professional terms and entities through dictionary matching;

[0028] Relationship recognition: Machine learning algorithm: Train a relationship classification model. Through the supervised learning method, use the labeled data to train the model to identify the relationships between entities;

[0029] Rule-based relationship extraction: Define rules, and automatically identify the relationships between entities through pre-set templates or pattern matching;

[0030] Knowledge graph construction: Graph database: Map the extracted entities and relationships to the graph database to construct nodes and edges;

[0031] Calculate the weights of the causal chains of risk events based on the dynamic Bayesian network DBN;

[0032] Predict the risk propagation path by combining the spatio-temporal graph attention network;

[0033] Dynamically update the knowledge graph through the triple <risk source, causal intensity, influence range>.

[0034] Data storage and query: Store the constructed knowledge graph in the graph database, and users can perform complex queries and analyses through graph query languages.

[0035] Furthermore, step S104 specifically includes:

[0036] Risk assessment:

[0037] Adopt the dynamic time warping algorithm to align the multi-source sensor time series data;

[0038] Introduce a sliding window mechanism in the analytic hierarchy process to dynamically adjust the risk factor weights according to the real-time environmental data;

[0039] Dynamically calculate the risk factor weights based on the sliding window mechanism. The formula is:

[0040]

[0041] Where Δxi(t) is the change rate of the i-th factor within the time window t, and α is the expert experience weight;

[0042] Risk prediction:

[0043] Use a spatio-temporal graph convolutional recurrent network to predict the spatio-temporal propagation path of risks in the transportation road network;

[0044] Dynamically switch the prediction model parameters according to real-time environmental data to adapt to special scenarios such as heavy rain and night;

[0045] Generate a risk prediction explanation report based on the causal chain of the knowledge graph, including key causal paths and weights;

[0046] Decision support:

[0047] Generate response plans based on the evaluation results, including preventive measures, emergency plans, and resource allocation suggestions;

[0048] Provide a visual decision support interface to help managers make quick and accurate decisions.

[0049] Furthermore, step S105 specifically includes:

[0050] Real-time data fusion:

[0051] Continuously collect the latest data from sensors and information systems to update the knowledge graph in real time;

[0052] Achieve instant synchronization of data and the graph, so that the knowledge graph always reflects the latest transportation status;

[0053] Feedback mechanism:

[0054] Verify and adjust the prediction model according to the occurrence of actual accidents or events;

[0055] Use the feedback data to optimize machine learning and deep learning models, and improve the prediction accuracy and reaction speed of the system.

[0056] Furthermore, step S106 specifically includes: Visual interface design:

[0057] Through a Web-based visual interface, display the knowledge graph, risk warning information, and decision support suggestions;

[0058] Adopt an interactive graph and dashboard method to intuitively present the key indicators and risk status during the transportation of hazardous chemical tank trucks;

[0059] Visualization function implementation:

[0060] Use a front-end visualization library to draw the knowledge graph;

[0061] Calculate key indicators through the data analysis module and display them in the form of charts and dashboards on the visual interface;

[0062] Integrate risk warning information and decision-making suggestions and present them to users;

[0063] Visualization effect optimization:

[0064] Continuously optimize the layout, interaction method, and information presentation of the visualization interface according to user feedback;

[0065] Adopt color matching and icon visual design means to improve the aesthetics and readability of the interface;

[0066] By providing a visualization interface, users can intuitively grasp the safety status of the transportation process of hazardous chemical tank trucks and make more timely and accurate decisions based on the visualization information.

[0067] In summary, the beneficial effects of this application are: it can significantly enhance the intelligence, accuracy, and comprehensiveness of the safety risk monitoring and early warning of hazardous chemical tank truck transportation, reduce the possibility of accidents, and improve the safety and stability of the overall transportation. Brief Description of the Drawings

[0068] Figure 1 It is the overall flowchart of the method for monitoring and early warning of the safety risk of hazardous chemical tank truck transportation based on the knowledge graph of the present invention;

[0069] Figure 2 It is the module composition diagram of the system for monitoring and early warning of the safety risk of hazardous chemical tank truck transportation based on the knowledge graph of the present invention. Detailed Description of the Preferred Embodiment

[0070] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0071] A specific embodiment of this application provides a system for monitoring and early warning of the safety risk of hazardous chemical tank truck transportation based on the knowledge graph. Refer to Figure 1-2 , and specifically includes the following modules:

[0072] The hazardous chemical tank truck transportation data collection module A1 is used to collect data related to the safety of hazardous chemical tank truck transportation in real time, including vehicle status data, driver behavior data, environmental condition data, real-time monitoring data of hazardous chemicals, and safety data during the transportation process, and obtain heterogeneous multi-source hazardous chemical tank truck transportation data;

[0073] A module for collecting laws, regulations, standards and historical accident data of hazardous chemical tank truck transportation, which is used to collect data from authoritative data sources according to various published standards and historical accident data of hazardous chemical tank truck transportation, and realize dynamic real-time update; obtain multi-source safety standards and specifications for hazardous chemical tank truck transportation;

[0074] Data preprocessing module A3, which is used to clean and standardize the collected safe transportation data and standard specification data, and then store them uniformly;

[0075] Dynamic knowledge graph module A4, which uses a causal chain inference engine based on causal relationships and data correlations to quantify the causal relationship strength of risk events for hazardous chemical tank truck transportation data, safety standards and specifications for hazardous chemical tank truck transportation, and accident data of hazardous chemical tank truck transportation; predicts the risk propagation path through the Spatio-Temporal Graph Attention Network (Spatio-Temporal GAT);

[0076] Hazardous chemical tank truck transportation safety risk situation analysis module A5, which is based on the dynamic knowledge graph of hazardous chemical tank truck transportation safety risks, analyzes the risk situation of the real-time collected hazardous chemical tank truck data, and automatically marks risk events when the transportation data is greater than the acceptable range of the risk situation; dynamically adjusts the risk factor weights based on the sliding window mechanism; generates a real-time risk index by combining the multi-sensor data alignment algorithm (DTW).

[0077] Dynamic update and feedback module A6, which updates the knowledge graph according to the real-time data during transportation, and optimizes the risk factor weight allocation mechanism and the real-time risk index generation algorithm based on the feedback results;

[0078] Hazardous chemical tank truck safety risk situation visualization module A7, which combines EGIS map technology to display the global safety risk situation in real time.

[0079] Preferably, the chemical production safety data collection module obtains key information on transportation and environmental monitoring from different vehicles through multi-source data integration and real-time collection; the integrated data includes real-time data from sensors and monitoring tank trucks, and environmental parameters such as vehicle performance parameters, personnel and vehicle positioning, atmospheric monitoring, and temperature monitoring.

[0080] The knowledge graph module analyzes accident causality, safety laws, regulations, standards, and specifications for the transportation of hazardous chemical tank trucks, and extracts knowledge from semi-structured data using a rule-based method; extracts knowledge from unstructured data based on the BRT-CRF model, and sets up multiple groups of comparative experiments to verify the effectiveness of the model; then uses a template matching method based on causal cue words to extract the causes and results of accidents. Finally, through knowledge fusion, the extracted risk information is stored in the Neo4j graph database in the form of triples. A single risk information triple includes the cause of the risk event, the risk relationship, and the result of the risk event, where the risk relationship includes direct and indirect relationships; thus, the construction of the safety risk knowledge graph for the transportation of hazardous chemical tank trucks is completed.

[0081] The safety risk situation analysis module for the transportation of hazardous chemical tank trucks automatically marks risk events, including the time of event occurrence, the location of event occurrence, and the spatio-temporal positioning information of the personnel involved in the event.

[0082] Another specific embodiment of this application provides a safety risk monitoring and early warning method for the transportation of hazardous chemical tank trucks based on a knowledge graph, referring to Figure 1-2 , and adopting the system as described above; the specific steps include: S101: Data collection: Collect various types of data including the safe transportation data of hazardous chemical tank trucks, as well as relevant laws, regulations, standards, and historical transportation accident data for the safe transportation of hazardous chemical tank trucks. In this embodiment, preferably, a sensor network is used to collect the operating state data of the transportation of hazardous chemical tank trucks in real time, such as temperature, leakage, pressure, vibration, etc.;

[0083] Obtain other data related to the transportation of hazardous chemical tank trucks from the database of the enterprise or the transportation control platform for hazardous chemical tank trucks, such as unstructured approval documents, planning documents, etc.;

[0084] Extract historical accident and incident reports from the enterprise's internal database;

[0085] Scrape relevant laws, regulations, standards, and specifications for the safe transportation of hazardous chemical tank trucks from authoritative websites;

[0086] Use environmental monitoring equipment to collect external environmental data, such as air temperature, humidity, wind speed, etc.;

[0087] Collect personnel operation records and maintenance logs;

[0088] S102: Data preprocessing: Remove noise data and invalid data from the collected data; such as abnormal data caused by sensor failure; standardize data from different sources and types to make them have a unified measurement standard; store the processed data in a relational database or a NoSQL database.

[0089] S103: Knowledge Graph Construction: Identify security risk targets based on security risk formation elements, and identify target triples one by one for the security risk targets, and then construct a security risk knowledge graph;

[0090] In this embodiment, preferably, step S103 specifically includes:

[0091] Preprocessing of data:

[0092] First, preprocess laws and regulations, industry standards or specifications, accident reports, and operation logs, including word segmentation, stop word removal, and named entity annotation operations; through preprocessing, the data can be cleaned and normalized to provide a better data basis for subsequent entity extraction;

[0093] Entity extraction:

[0094] Natural Language Processing (NLP) technology: Use existing NLP technologies, such as named entity recognition (NER) models, to identify and extract key entities from text. For example, from the sentence "Today, a tanker truck carrying methanol departs from Shanghai and is expected to arrive in Beijing at 10 am tomorrow. The driver Zhang San has 5 years of driving experience, and the vehicle number is Shanghai A12345, which meets the latest safety standards.", entities such as "methanol", "Shanghai A12345", "Zhang San", "Shanghai", "Beijing", "Today", "10 am tomorrow", and "meets the latest safety standards" can be extracted.

[0095] Domain dictionary matching: Construct a special dictionary for the chemical production field, and assist in identifying specific professional terms and entities through dictionary matching; this method has high accuracy for proper nouns in specific fields.

[0096] Relationship recognition:

[0097] Machine learning algorithm: Train a relationship classification model. Through supervised learning methods, use the labeled data to train the model to identify the relationships between entities. For example, there is a "depart from..." relationship between "Shanghai A12345" and "Shanghai", and a "drive" relationship between "Zhang San" and "Shanghai A12345".

[0098] Rule-based relationship extraction: Define a series of rules, and automatically identify the relationships between entities through pre-set templates or pattern matching. This method is simple and efficient, but the rule base needs to be maintained and updated.

[0099] Implementation of causal chain inference engine:

[0100] Use a dynamic Bayesian network (DBN) to quantify the causal relationship strength between risk events. Calculate the causal transfer probability and conditional entropy based on historical accident data. The formula is:

[0101] Causal Strength = P(Effect∣Cause) × (1 - H(Effect∣Cause))

[0102] Among them, H is the conditional entropy, reflecting the certainty of the causal relationship.

[0103] Code example (dynamic causal weight update):

[0104]

[0105] Spatio-Temporal Graph Attention Network (Spatio-Temporal GAT) integration:

[0106] Construct a road network topology graph, where the nodes are road segments and the edges are the connection relationships between road segments. The weights include real-time traffic flow, accident rate, etc.

[0107] Predict the risk propagation path through the spatio-temporal graph attention network. The model structure is as follows:

[0108]

[0109] Triple dynamic update logic:

[0110] Neo4j operation example: Supplementary Cypher query statement for triple update:

[0111] / / Insert causal chain triple

[0112] CREATE (c:RiskSource{name: "Tank corrosion"})

[0113] CREATE (e:RiskImpact{name: "Leakage"})

[0114] CREATE (c)-[r:CAUSES{strength: 0.85, scope: "Road section G4-K203"}]->(e)

[0115] Update mechanism: It is explained that the system scans the real-time data every 10 minutes to dynamically update the causal chain weights and influence ranges.

[0116] Data storage and query: Store the constructed knowledge graph in a graph database, and users can perform complex queries and analyses through graph query languages (such as Cypher). For example, one can query the common causes of a certain type of accident, or the work records and types of accidents handled by a certain driver.

[0117] S104: Safety risk situation analysis, identifying and marking safety risks for the collected real-time data based on the safety risk knowledge graph, evaluating the identified risks by combining expert experience and system analysis, and providing decision-making suggestions;

[0118] In this embodiment, preferably, step S104 specifically includes:

[0119] Implementation of the DTW algorithm:

[0120] Algorithm steps:

[0121] Input: Multisensor time-series data (such as pressure and temperature sequences).

[0122] Output: Aligned time-series matrix.

[0123] Code example:

[0124]

[0125] Introduce a sliding window mechanism in the Analytic Hierarchy Process (AHP) to dynamically adjust the weights of risk factors according to real-time environmental data.

[0126] Dynamically calculate the weights of risk factors based on the sliding window mechanism. The formula is:

[0127]

[0128] where Δxi(t) is the change rate of the i-th factor within the time window t, and α is the weight of expert experience.

[0129] Code implementation:

[0130] def dynamic_weight(alpha,ahp_weight,delta_x):

[0131] total_delta=sum(delta_x.values())

[0132] return alpha*ahp_weight+(1-alpha)*(delta_x / total_delta)

[0133] Risk prediction:

[0134] Spatio-temporal graph convolutional recurrent network (ST-GCRN):

[0135] Use the spatio-temporal graph convolutional recurrent network (ST-GCRN) to predict the spatio-temporal propagation path of risks in the transportation road network. Construct spatio-temporal training samples using historical accident data. The input is the sensor data (time window) of the previous 6 hours and the road network topology graph (spatial association), and the output is the risk probability for the next 1 hour.

[0136] Dynamic scene adaptation:

[0137] Dynamically switch the prediction model parameters according to real-time environmental data to adapt to special scenes such as heavy rain and night; when the system detects that the visibility is <100 meters, the foggy weather optimization model is automatically loaded, and the F1-score of this model on the foggy weather test set reaches 0.91, which is 39% higher than the baseline model.

[0138] Scene adaptation model switching logic:

[0139] When heavy rain (rainfall > 50 mm / h) is detected, load the pre-trained heavy rain optimization model (storm_gat.pth).

[0140] When the night mode (light intensity < 10 lux) is activated, switch to the night time series model (night_gru.pth).

[0141] Code example:

[0142]

[0143] Performance comparison:

[0144] Add the accuracy comparison between the scene adaptation model and the traditional model:

[0145]

[0146] Causal interpretability report generation:

[0147] Generate a risk prediction explanation report based on the causal chain of the knowledge graph, including the key causal path and weight. For example: "The leakage risk of the predicted section S12-K55+200 is high risk (88%), main causal chain:

[0148] 1. Tank corrosion (weight 0.45) → stress concentration (weight 0.72) → crack propagation (weight 0.91)

[0149] 2. Driver's sudden braking (weight 0.63) → liquid sloshing in the tank (weight 0.58) → sudden pressure increase (weight 0.85)"

[0150] Decision support:

[0151] Based on the evaluation results, generate response plans, including preventive measures, emergency plans, and resource allocation suggestions;

[0152] Provide a visual decision support interface to help managers make quick and accurate decisions.

[0153] S105: Update the knowledge graph according to the production data, and optimize the risk factor weight allocation mechanism and the real-time risk index generation algorithm based on the feedback results;

[0154] In this embodiment, preferably, step S105 specifically includes: Real-time data fusion:

[0155] Continuously collect the latest data from sensors and information systems to update the knowledge graph in real time;

[0156] Achieve instant synchronization of data and the graph, so that the knowledge graph always reflects the latest production status.

[0157] Feedback mechanism:

[0158] Verify and adjust the prediction model according to the occurrence of actual accidents or events;

[0159] Use the feedback data to optimize the graph algorithm and deep learning model, and improve the prediction accuracy and reaction speed of the system.

[0160] S106: Visualize the risk situation and intuitively display the risk situation.

[0161] In this embodiment, preferably, step S106 specifically includes: Visualization interface design:

[0162] Develop a Web-based visualization interface to display the knowledge graph, risk warning information, and decision support suggestions;

[0163] Adopt an interactive graph and dashboard method to intuitively present the key indicators and risk status during the transportation of hazardous chemical storage tanks;

[0164] Visualization function implementation:

[0165] Use a front-end visualization library (such as D3.js, ECharts, etc.) to draw the knowledge graph and support interactive operations such as zooming and dragging;

[0166] Calculate the key indicators through the data analysis module and display them in the form of charts, dashboards, etc. on the visualization interface;

[0167] Integrate risk warning information and decision-making suggestions and present them to users in a prominent manner.

[0168] Visualization effect optimization:

[0169] Continuously optimize the layout, interaction method, and information presentation of the visualization interface according to user feedback;

[0170] Adopt visual design means such as color matching and icons to improve the aesthetics and readability of the interface;

[0171] By providing a visual interface, users can intuitively grasp the safety status of the transportation process of hazardous chemical storage tanks and make more timely and accurate decisions based on the visual information.

[0172] Although the embodiments of the present invention have been shown and described (see the above detailed description), for those of ordinary skill in the art, it can be understood that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A hazardous chemicals tanker transportation safety risk monitoring and early warning system based on knowledge graph, characterized by: include: The hazardous chemicals tanker transportation data collection module is used to collect real-time hazardous chemicals tanker transportation safety-related data, including vehicle status data, driver behavior data, environmental condition data, real-time monitoring data of hazardous chemicals, and safety data during transportation. Obtain heterogeneous multi-source hazardous chemicals tanker transportation data; The module for collecting data on laws, regulations, standards and historical accidents for hazardous chemicals tanker transportation is used to collect data from authoritative data sources based on the published standards and historical accident data for various hazardous chemicals tanker transportation, and to achieve dynamic real-time updates; obtain multi-source safety standards and specifications for hazardous chemicals tanker transportation; The data preprocessing module is used to clean and standardize the collected safe transportation data and standard specification data, and then store them in a unified manner; The dynamic knowledge graph module constructs a dynamic knowledge graph of hazardous chemicals tanker transportation safety risks based on the causal relationship and data correlation of hazardous chemicals tanker transportation data, hazardous chemicals tanker transportation safety standards and specifications, and hazardous chemicals tanker transportation accident data; Using a causal chain reasoning engine, we can quantify the causal strength of risk events based on a dynamic Bayesian network. Predict risk propagation paths through spatiotemporal graph attention networks; dynamically update knowledge graphs through triples of <risk source, causal strength, impact range>. The hazardous chemicals tanker transportation safety risk situation analysis module conducts risk situation analysis on the real-time collected hazardous chemicals tanker data based on the dynamic hazardous chemicals tanker transportation safety risk knowledge graph. When the transportation data is greater than the acceptable range of the risk situation, a risk event is marked. Dynamically adjust risk factor weights based on a sliding window mechanism; Combine multi-sensor data alignment algorithm to generate real-time risk index. Dynamic update and feedback module, which updates the knowledge graph according to the real-time data in the transportation process, and optimizes the risk factor weight allocation mechanism and real-time risk index generation algorithm based on the feedback results; The hazardous chemicals tanker safety risk situation visualization module, combined with EGIS map technology, displays the global safety risk situation in real time.

2. According to claim 1, a hazardous chemicals tanker transportation safety risk monitoring and early warning system based on knowledge graph is characterized by: The hazardous chemicals tanker transportation safety data acquisition module obtains key information on transportation and environmental monitoring from different vehicles by implementing multi-source data integration and real-time acquisition; the integrated data includes real-time data of sensors and monitoring tankers, vehicle performance parameters, personnel and vehicle positioning, atmospheric monitoring, and environmental parameters including temperature monitoring.

3. According to claim 1, a hazardous chemicals tanker transportation safety risk monitoring and early warning system based on knowledge graph is characterized by: The dynamic knowledge graph module analyzes the cause-effect relationship of the accident, the laws and regulations on the safety of hazardous chemicals tanker transportation, and the standards and specifications, and extracts knowledge from semi-structured data using a rule-based method; extracts knowledge from unstructured data based on the BRT-CRF model; and then uses a template matching method based on causal prompt words to extract the causes and consequences of the accident. Calculate the weight of the causal chain of risk events based on dynamic Bayesian networks; Combined with the spatiotemporal graph attention network, the risk propagation path is predicted; the extracted risk information is stored in the Neo4j graph database in the form of triples, and the knowledge graph is dynamically updated through the triples <risk source, causal strength, impact range>.

4. According to claim 1, a hazardous chemicals tanker transportation safety risk monitoring and early warning system based on knowledge graph is characterized by: The hazardous chemicals tanker transportation safety risk situation analysis module performs risk event marking including the time of the event, the location of the event, and the spatiotemporal positioning information of the people involved in the event.

5. A method for monitoring and early warning of safety risks in hazardous chemicals tanker transportation based on knowledge graph, characterized in that: The system according to any one of claims 1 to 4 is used; the specific steps include: S101: Data collection: Collect various data including hazardous chemicals tanker transportation safety data and relevant laws and regulations, standards and specifications for hazardous chemicals tanker transportation safety, and historical transportation accident data; S102: Data preprocessing: remove noise data and invalid data from the collected data; standardize data from different sources and types to make them have a unified metric; store the processed data in a relational database or a NoSQL database; S103: Dynamic knowledge graph construction: Identify security risk targets based on security risk formation factors, and identify target triples for security risk targets one by one, and then construct a security risk dynamic knowledge graph; S104: Security risk situation analysis: identify and mark security risks based on the security risk knowledge graph based on the collected real-time data, evaluate the identified risks, and provide decision-making suggestions; S105: Update the knowledge graph according to the transportation data, and optimize the risk factor weight allocation mechanism and real-time risk index generation algorithm based on the feedback results; S106: Visualize risk situation and display risk situation directly.

6. According to claim 5, a method for monitoring and early warning of safety risks in hazardous chemicals tanker transportation based on knowledge graph is characterized in that: Step S103 specifically includes: Data preprocessing: First, preprocess laws and regulations, industry standards or specifications, accident reports, and operation logs, including word segmentation, stop word removal, and named entity tagging operations; Entity extraction: Use the pre-trained BERT model to identify and extract key entities from text; Domain dictionary matching: Build a special dictionary for hazardous chemicals tanker transportation, and use dictionary matching to assist in identifying specific professional terms and entities; Relationship identification: Machine learning algorithms: train relationship classification models, use supervised learning methods to train models using labeled data to identify relationships between entities; Rule-based relationship extraction: define rules and automatically identify the relationship between entities through pre-set templates or pattern matching; Graph construction: Graph database: Map the extracted entities and relationships into the graph database to build nodes and edges; Calculate the weight of the causal chain of risk events based on the dynamic Bayesian network DBN; Combined with spatiotemporal graph attention network to predict risk propagation path; The knowledge graph is dynamically updated through the triple <risk source, causal strength, impact range>. Data storage and query: The constructed knowledge graph is stored in the graph database, and users can perform complex queries and analyses through the graph query language.

7. According to claim 5, a method for monitoring and early warning of safety risks in hazardous chemicals tanker transportation based on knowledge graph is characterized in that: Step S104 specifically includes: risk assessment: Dynamic time warping algorithm is used to align multi-source sensor time series data; Introducing a sliding window mechanism into the AHP method, dynamically adjusting the risk factor weights based on real-time environmental data; The risk factor weights are dynamically calculated based on the sliding window mechanism. The formula is: Where Δxi(t) is the rate of change of the i-th factor in the time window t, and α is the expert experience weight; Risk Profile: Use spatiotemporal graph convolutional recurrent networks to predict the spatiotemporal propagation path of risks in transportation networks; Dynamically switch prediction model parameters based on real-time environmental data to adapt to special scenarios such as heavy rain and nighttime; Generate risk prediction explanation report based on knowledge graph causal chain, including key causal paths and weights; Decision support: Based on the assessment results, generate response plans, including preventive measures, emergency plans, and resource allocation recommendations; Provides a visual decision support interface to help managers make fast and accurate decisions.

8. According to claim 5, a method for monitoring and early warning of safety risks in hazardous chemicals tanker transportation based on knowledge graph is characterized in that: Step S105 specifically includes: Real-time data fusion: Continuously collect the latest data from sensors and information systems to update the knowledge graph in real time; Realize instant synchronization of data and graphs, so that the knowledge graph always reflects the latest transportation status; Feedback Mechanism: Verify and adjust the prediction model based on actual accidents or incidents; Use feedback data to optimize machine learning and deep learning models to improve the system's predictive accuracy and response speed.

9. According to claim 5, a method for monitoring and early warning of safety risks in hazardous chemicals tanker transportation based on knowledge graph is characterized in that: Step S106 specifically includes: Visual interface design: Display knowledge graphs, risk warning information, and decision support suggestions through a web-based visual interface; Interactive graphics and dashboards are used to intuitively present key indicators and risk status during the transportation of hazardous chemicals tankers; Visualization function realization: Use the front-end visualization library to draw knowledge graphs; Calculate key indicators through the data analysis module and display them on the visual interface in the form of charts and dashboards; Integrate risk warning information and decision-making suggestions and present them to users; Visualization effect optimization: Continuously optimize the layout, interaction mode and information presentation of the visualization interface based on user feedback; Use color matching and icon visual design methods to improve the aesthetics and readability of the interface; By providing a visual interface, users can intuitively understand the safety status of the hazardous chemicals tanker transportation process and make more timely and accurate decisions based on the visual information.

Images