Dangerous chemical accident emergency management cloud data platform based on affair atlas

Abstract

The invention discloses a hazardous chemical accident emergency management cloud data platform based on a affair atlas, and relates to the field of hazardous chemical accident emergency management. According to the method, the accident investigation report database is analyzed based on a principal component analysis method, and the risk key influence factors of the dangerous chemical accident are determined; establishing a fuzzy interpretation structure model of the dangerous chemical accident risk influence factors; according to self-defined risk division grades, obtaining a risk grade evaluation result; building a Bayesian network model of dangerous chemical accident risks according to the fuzzy interpretation structure model, and evaluating accident risk influence factors to obtain an accident risk influence factor evaluation result; and performing early warning based on the risk level assessment result and the accident risk influence factor assessment result.

Description

Technical Field

[0001] The present invention relates to the field of emergency management of accidents involving hazardous chemicals, and in particular to a cloud data platform for emergency management of accidents involving hazardous chemicals based on a causal graph. Background Art

[0002] Current research on the causal relationship of hazardous chemical accidents lacks analysis of the details of the accidents and the causal evolution process. A causal graph is a logical knowledge base that reveals the evolution laws between events. It includes a directed cyclic graph in which nodes represent events and directed edges represent the sequential relationship, causal relationship, etc. between events. It is a special form of knowledge graph.

[0003] The present invention designs a cloud data platform for emergency management of hazardous chemical accidents based on a causal map, which aims to mine causal relationships through historical accident investigation report data, construct a causal map, and intuitively display the evolution laws and patterns between events. The event probability calculation function provided by the platform, combined with the customized professional accident probability risk level classification, can realize real-time evaluation of the probability of accident occurrence. When the probability of an accident reaches the warning level, the platform will mark the accident node in red and pop up a warning window, so as to achieve effective evaluation and warning of hazardous chemical accidents. By combing the development context of the accident and analyzing the law of accident development, on the one hand, it can assist in the investigation of the cause of the accident and evaluate the probability of the event. On the other hand, it provides decision support for future accident evaluation, prediction, prevention, and warning, which is helpful for accident cause investigation, prediction, prevention, and warning, and avoids secondary injuries as much as possible, and reduces casualties and losses caused by accidents. Summary of the invention

[0004] The purpose of the present invention is to solve the problems in the prior art and to propose a cloud data platform for emergency management of hazardous chemical accidents based on a cause-and-effect graph.

[0005] A cloud data platform for emergency management of hazardous chemical accidents based on event graph, including:

[0006] The user login and event receiving UI module guides the user to the system main page through the verification of the user name and password; the system main page includes a hazardous chemical accident investigation report library, which divides accidents into four categories: explosion, fire, poisoning, and suffocation, and has the functions of adding, deleting, and querying;

[0007] A module for generating a matter map, wherein the matter map is generated based on a single, multiple or entire investigation report library, specifically comprising the following steps:

[0008] S1: Collect and process the data of hazardous chemical accidents. Based on the accident investigation report text input by the user, generate sentences with logical semantics through denoising and sentence segmentation. Segment the words and determine whether the sentence contains causal prompt words. If so, extract the explicit causal relationship, otherwise extract the implicit causal relationship.

[0009] S2: The cloud data platform is oriented to the text data of hazardous chemical accidents. It divides the causal relationship extraction of accidents into explicit causal relationship extraction and implicit causal relationship extraction. It adopts the pattern matching-based method and the machine learning-based method respectively. On the basis of the causal relationship event pairs obtained by causal relationship extraction, different expressions of the same event are merged through event co-reference elimination to generate a causal relationship graph for each accident. Then, based on the causal relationship graph of a single accident, a causal relationship event map in the field of hazardous chemical accidents is generated.

[0010] S3: Abstract specific events through event and domain ontology mapping, further form a matter-of-fact graph, and mark the event transition probability on the edge of the matter-of-fact graph, and finally display it in the form of a matter-of-fact graph.

[0011] A risk warning module, wherein the risk warning module analyzes the accident investigation report database based on the principal component analysis method to determine the key risk influencing factors of hazardous chemical accidents; establishes a fuzzy interpretation structural model of risk influencing factors of hazardous chemical accidents; obtains risk level assessment results according to the customized risk classification level; constructs a Bayesian network model of the risk of hazardous chemical accidents based on the fuzzy interpretation structural model, evaluates the risk influencing factors of accidents, obtains the assessment results of the risk influencing factors of accidents, and issues a warning based on the risk level assessment results and the assessment results of the risk influencing factors of accidents;

[0012] User module; includes an independent risk factor library, which is constructed through user-side products, wherein the risk factor library can obtain the causal relationship diagram corresponding to the node through the input event node, and give its corresponding recommended measures, and generate a visual causal relationship diagram.

[0013] In the above-mentioned hazardous chemicals accident emergency management cloud data platform based on the causal graph, the method for extracting the explicit causal relationship is: according to the text characteristics of the accident investigation report, the syntactic pattern is summarized, the extraction template is constructed, and the extraction rules are designed to extract the cause clause and the result clause, and the dependency syntactic analysis is used to obtain the event tuples contained in the clauses, thereby obtaining explicit causal relationship event tuple pairs. The method for extracting implicit causal relationships is: using a combination of part-of-speech filtering and dependency syntactic analysis to extract the events contained in each sentence. The events are represented in the form of tuples and paired in pairs. A bidirectional LSTM method based on the self-attention mechanism is used, and only the sentence set with the event trigger word position identifier is used as input to classify the relationship between the event tuple pairs, thereby obtaining implicit causal relationship event tuple pairs.

[0014] In the above-mentioned cloud data platform for emergency management of hazardous chemicals accidents based on the event graph, the cloud data platform also includes a forum module, which includes official website link navigation, official policy announcements, hazardous chemicals accident announcements and industry information, and connects the above information to the cloud data terminal, and has functions such as user communication, question and answer, etc.

[0015] In the above-mentioned hazardous chemicals accident emergency management cloud data platform based on the event map, the cloud data platform also includes a mall module, which has functions such as attracting investment and self-operated sales of hazardous chemicals related products, creating a comprehensive platform for the hazardous chemicals industry that integrates systematic information, services, materials, and equipment.

[0016] In the above-mentioned hazardous chemicals accident emergency management cloud data platform based on the event graph, the cloud data platform also includes an education and training module.

[0017] The present invention innovatively provides an early warning function. Users can query the causal relationship graph corresponding to a certain event node by inputting this node. Users can use the cloud data platform to analyze, evaluate and warn of hazardous chemical accidents. The causal relationship of the hazardous chemical accident is analyzed through the generated graph to obtain the probability of an accident. The cloud data platform evaluates the probability of the accident and divides it into probability levels according to the cloud data platform's customized probability level. When the probability of an accident reaches the warning level, the cloud data platform will mark the accident node in red and issue a pop-up warning, thereby achieving the evaluation and early warning of hazardous chemical accidents. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 This is an overall framework diagram of the platform in the hazardous chemicals accident emergency management cloud data platform based on the event graph proposed by the present invention.

[0019] Figure 2 This is a workflow diagram of a hazardous chemical accident reporting library in a hazardous chemical accident emergency management cloud data platform based on a causal graph proposed by the present invention.

[0020] Figure 3 This is a platform module diagram of a hazardous chemicals accident emergency management cloud data platform based on a matter-of-fact graph proposed by the present invention.

[0021] Figure 4 The present invention proposes a causal relationship extraction process in a hazardous chemicals accident emergency management cloud data platform based on a causal graph.

[0022] Figure 5 This is a path diagram for generating a causal graph in a hazardous chemicals accident emergency management cloud data platform based on a causal graph proposed by the present invention.

[0023] Figure 6 This invention proposes a technical roadmap for the hazardous chemicals accident emergency management cloud data platform based on the event graph. DETAILED DESCRIPTION

[0024] Reference Figure 1-6 , a cloud data platform for emergency management of hazardous chemical accidents based on event graph, including:

[0025] The user login and event receiving UI module guides the user to the system main page through the verification of the user name and password; the system main page includes a hazardous chemical accident investigation report library, which can upload hazardous chemical accident investigation reports by adding or directly uploading the web link of the accident investigation report. The cloud data platform can directly obtain the accident investigation report through the link, thereby forming a hazardous chemical accident investigation report library built by the user. The library has the functions of adding, deleting and querying, and divides accidents into four categories: explosion, fire, poisoning and suffocation. Users can also search for accident reports by adding and deleting. Users can operate the accident investigation report or directly enter keywords and related parameters to accurately match the expected hazardous chemical accident investigation report. The hazardous chemical accident investigation report library can be managed through these functions;

[0026] A module for generating a matter map, wherein the matter map is generated based on a single, multiple or entire investigation report library, specifically comprising the following steps:

[0027] S1: Collect and process the data of hazardous chemical accidents. Based on the accident investigation report text input by the user, generate sentences with logical semantics through denoising and sentence segmentation. Segment the words and determine whether the sentence contains causal prompt words. If so, extract the explicit causal relationship, otherwise extract the implicit causal relationship.

[0028] S2: Based on the in-depth analysis of the text data of hazardous chemical accidents, this cloud data platform extracts causal relationships of hazardous chemical accidents, which is divided into two parts: explicit causal relationship extraction and implicit causal relationship extraction. Causal relationship extraction is an important prerequisite for causal relationship analysis, a basic work for the construction of a causal graph, and one of the core tasks of the construction of a causal graph. Causal relationships can be divided into explicit causal relationships and implicit causal relationships. The extraction of causal relationships of hazardous chemical accidents includes the extraction of explicit causal relationships of hazardous chemical accidents and the extraction of implicit causal relationships of hazardous chemical accidents. The causal relationship extraction process is as follows: Figure 4 As shown in the figure, a pattern matching-based method and a machine learning-based method are used respectively. On the basis of the causal event pairs obtained by causal relationship extraction, different expressions of the same event are merged through event co-reference elimination, and a causal relationship graph is generated for each accident to describe the process of the accident. Then, based on the causal relationship graph of a single accident, a causal relationship graph in the field of hazardous chemicals accidents is generated. The research idea of ​​generating the graph is as follows Figure 5 As shown;

[0029] The specific extraction method of explicit causal relationship is: according to the text characteristics of accident investigation report, the syntactic pattern is summarized, the extraction template is constructed, and the extraction rules are designed to extract the cause clause and the result clause. The event tuple contained in the clause is obtained by dependency syntactic analysis, so as to obtain the explicit causal relationship event tuple pair. The specific extraction method of implicit causal relationship is: the event contained in each sentence is extracted by combining part-of-speech filtering and dependency syntactic analysis. The events are represented in the form of tuples and paired in pairs. The bidirectional LSTM method based on self-attention mechanism is adopted, and only the sentence set with event trigger word position identifier is used as input. The relationship of event tuple pairs is classified, so as to obtain implicit causal relationship event tuple pairs.

[0030] S3: Through event and domain ontology mapping, specific events are abstracted and further formed into a causal map to reveal the laws and patterns of accident development. The event transfer probability is marked on the edge of the causal map and finally displayed in the form of a causal map. The generated causal map can be used to analyze the cause and effect relationship of a hazardous chemical accident and obtain the probability of an accident occurring.

[0031] The risk warning module analyzes the accident investigation report database based on the principal component analysis method to determine the key risk influencing factors of hazardous chemical accidents; establishes a fuzzy interpretation structural model of risk influencing factors of hazardous chemical accidents; obtains risk level assessment results according to the customized risk classification level; constructs a Bayesian network model of hazardous chemical accident risk based on the fuzzy interpretation structural model, evaluates the accident risk influencing factors, obtains the accident risk influencing factor assessment results, and issues a warning based on the risk level assessment results and the accident risk influencing factors assessment results. The specific implementation of the warning process is as follows: Figure 6 shown.

[0032] The cloud data platform designed in this application also includes:

[0033] User module; includes an independent risk factor library, which is constructed through user-side products, wherein the risk factor library can obtain the causal relationship diagram corresponding to the node through the input event node, and give its corresponding recommended measures, and generate a visual causal relationship diagram.

[0034] Forum module, the forum module includes official website link navigation, official policy announcements, hazardous chemicals accident announcements and industry information, and connects the above information to the cloud data terminal, and also has functions such as user communication, question and answer, etc.

[0035] The mall module of this cloud data platform can also attract investment and sell hazardous chemicals-related equipment and materials on its own, creating a comprehensive platform for the hazardous chemicals industry that integrates systematic information, services, materials, and equipment. The mall module can be customized according to different users, and can provide technical support and consulting services, emergency resource libraries, supervision and compliance tools, product catalogs and searches, online ordering and trading systems, and other functions.

[0036] The education and training module can provide a comprehensive service education and training center for colleges and governments, which not only supports education and training, but also strengthens supervision and emergency management capabilities, thereby improving the level of hazardous chemicals safety management in the entire society.

[0037] It is known from common technical knowledge that the present invention can be implemented by other embodiments that do not deviate from its spirit or essential features. Therefore, the above disclosed embodiments are only illustrative in all respects and are not exclusive. All changes within the scope of the present invention or within the scope equivalent to the present invention are included in the present invention.

Claims

1. A cloud data platform for emergency management of hazardous chemical accidents based on event graph, characterized in that: include: The user login and event receiving UI module guides the user to the system main page through the verification of the user name and password; the system main page includes a hazardous chemical accident investigation report library, which divides accidents into four categories: explosion, fire, poisoning, and suffocation, and has the functions of adding, deleting, and querying; A module for generating a matter map, wherein the matter map is generated based on a single, multiple or entire investigation report library, specifically comprising the following steps: S1: Collect and process the data of hazardous chemical accidents. Based on the accident investigation report text input by the user, generate sentences with logical semantics through denoising and sentence segmentation. Segment the words and determine whether the sentence contains causal prompt words. If so, extract the explicit causal relationship, otherwise extract the implicit causal relationship. S2: The cloud data platform is oriented to the text data of hazardous chemical accidents. It divides the causal relationship extraction of accidents into explicit causal relationship extraction and implicit causal relationship extraction. It adopts the pattern matching-based method and the machine learning-based method respectively. On the basis of the causal relationship event pairs obtained by causal relationship extraction, different expressions of the same event are merged through event co-reference elimination to generate a causal relationship graph for each accident. Then, based on the causal relationship graph of a single accident, a causal relationship event map in the field of hazardous chemical accidents is generated. S3: Abstract specific events through event and domain ontology mapping, further form a matter-of-fact graph, and mark the event transition probability on the edge of the matter-of-fact graph, and finally display it in the form of a matter-of-fact graph. A risk warning module, wherein the risk warning module analyzes the accident investigation report database based on the principal component analysis method to determine the key risk influencing factors of hazardous chemical accidents; establishes a fuzzy interpretation structural model of risk influencing factors of hazardous chemical accidents; obtains risk level assessment results according to the customized risk classification level; constructs a Bayesian network model of the risk of hazardous chemical accidents based on the fuzzy interpretation structural model, evaluates the risk influencing factors of accidents, obtains the assessment results of the risk influencing factors of accidents, and issues a warning based on the risk level assessment results and the assessment results of the risk influencing factors of accidents; User module; includes an independent risk factor library, which is constructed through user-side products, wherein the risk factor library can obtain the causal relationship diagram corresponding to the node through the input event node, and give its corresponding recommended measures, and generate a visual causal relationship diagram.

2. According to claim 1, a cloud data platform for emergency management of hazardous chemical accidents based on a matter-of-fact graph is characterized by: The method for extracting explicit causal relationships is as follows: according to the text characteristics of the accident investigation report, the syntactic patterns are summarized, the extraction templates are constructed, and the extraction rules are designed to extract the cause clauses and the result clauses, and the event tuples contained in the clauses are obtained by using dependency syntactic analysis, so as to obtain explicit causal relationship event tuple pairs. The method for extracting implicit causal relationships is as follows: the events contained in each sentence are extracted by combining part-of-speech filtering with dependency syntactic analysis, and the events are represented in the form of tuples and paired in pairs. A bidirectional LSTM method based on a self-attention mechanism is adopted, and only a set of sentences with event trigger word position identifiers is used as input to classify the relationships between event tuple pairs, so as to obtain implicit causal relationship event tuple pairs.

3. According to claim 1, a cloud data platform for emergency management of hazardous chemical accidents based on a matter-of-fact graph is characterized by: The cloud data platform also includes a forum module, which includes official website link navigation, official policy announcements, hazardous chemicals accident announcements and industry information, and connects the above information to the cloud data terminal. It also has functions such as user communication, question and answer.

4. According to claim 1, a cloud data platform for emergency management of hazardous chemical accidents based on a matter-of-fact graph is characterized by: The cloud data platform also includes a mall module, which has functions such as attracting investment and self-operated sales of hazardous chemicals related products, creating a comprehensive platform for the hazardous chemicals industry that integrates systematic information, services, materials, and equipment.

5. According to claim 4, a cloud data platform for emergency management of hazardous chemical accidents based on a matter-of-fact graph is characterized by: The cloud data platform also includes an education and training module.

Images