Accident reason classification and grading method and system

Abstract

The invention discloses an accident reason classification and grading method and system, and solves the technical problems of poor relevance between direct reasons and indirect reasons, unclear logicality and inaccurate accident reason analysis in part of historical case accident investigation reports. Comprising the steps of building an accident cause classification and grading framework, performing classification and grading identification on accident causes, performing semantic understanding and professional marking, and performing system development and function realization. The accident cause classification and grading method and system are formed by setting up the accident cause classification and grading framework and constructing the accident cause dictionary, the accident causes are clearly, comprehensively and systematically sorted, rapid classification and grading display of the accident causes of historical accidents is achieved, intelligent matching recommendation of the accident causes of new accidents is supported, and the accident cause classification and grading efficiency is improved. The system has the characteristics of automation, intelligence, high efficiency, accuracy and the like, and has important application value and popularization prospect in the aspects of improving accident investigation efficiency, guaranteeing safety production and the like.

Description

Technical Field

[0001] The present invention relates to the technical field of accident classification, and in particular to a method and system for classifying and grading accident causes. Background Art

[0002] Since 2018, the overall national production safety situation has improved and the number of accidents has dropped significantly, but there are still more than 10,000 accidents on average each year. It is still important to find out the causes of accidents, formulate targeted measures, and prevent accidents. The causes of accidents are complex and diverse. Based on traditional theoretical models such as Heinrich theory, trajectory intersection theory, and Swiss cheese model, it is difficult to comprehensively and accurately analyze the causes of complex and diverse accidents in China. At present, the analysis of the causes of accidents in domestic accident investigations often relies mainly on expert experience. Experts also often use theoretical methods such as fault tree analysis and root cause analysis to analyze the correlation between the causes of accidents. However, different experts have strong subjectivity. Especially when facing complex and multi-factor accidents, the constructed accident trees also have their own characteristics, making it difficult to logically, comprehensively and systematically determine the direct causes, indirect causes and their correlations of accidents.

[0003] In view of the above problems, the present invention provides an automated, intelligent, efficient and accurate accident cause classification and grading method and system. Summary of the invention

[0004] The purpose of the present invention is to provide an automated, intelligent, efficient and accurate method and system for classifying and grading accident causes, so as to solve the technical problems in some historical case accident investigation reports that the direct and indirect causes are not strongly correlated, the logic is unclear, and the accident cause analysis is not accurate enough.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] In a first aspect, the present invention provides a method for classifying and grading accident causes, comprising the following steps:

[0007] S1. Construct a classification and grading framework for accident causes, classify and analyze the direct causes of accidents according to unsafe behaviors of people, unsafe conditions of objects, and unsafe factors of the environment, and grade the indirect causes according to the correlation between the causes of accidents;

[0008] S2. Combine big data identification technology to identify the direct and indirect causes of historical cases according to the accident cause classification and grading framework;

[0009] S3. Combine semantic understanding technology to professionally annotate the identified content and build a dictionary of accident causes;

[0010] S4. Combine the accident cause classification and grading framework with the construction of an accident cause dictionary to form an accident cause classification and grading analysis method and develop a corresponding system to achieve rapid classification and grading display of accident causes of historical accidents. The system is equipped with an intelligent matching model for accident causes to support intelligent matching recommendations for accident causes of new accidents.

[0011] Furthermore, the S1 specifically includes the following steps:

[0012] S11. Comprehensively apply the intersection theory of human and material causes to form accidents and the Haddon matrix theory of human, equipment and environmental factors interacting to form accidents. Use charts to classify the direct causes of accidents according to unsafe human behavior, unsafe material conditions and unsafe environmental factors.

[0013] S12. Divide the specific hazard factors of unsafe human behavior, unsafe conditions of objects, and unsafe environmental factors into Level 2 factor indicators and Level 3 factor indicators;

[0014] S13. Based on the root cause analysis method and the accident cause structure of domestic accident investigation reports, the indirect causes are divided into factors that cause direct causes, enterprise management causes, and system causes. For each direct cause, the inducement is found. For each inducement, the enterprise management cause is identified. For the enterprise management cause, the system cause is analyzed.

[0015] S14. Combined with the above accident cause classification and grading results, a framework for accident cause classification and grading is constructed based on the FTA fault tree analysis method and the MORT parsing tree analysis method.

[0016] Furthermore, the S2 specifically includes the following steps:

[0017] S21, using CRF word segmentation and TextRank keyword extraction to implement word segmentation of accident causes in historical cases and identify characteristic words of accident causes;

[0018] S22, using HMM role-labeled named entity recognition to achieve recognition of person names, place names, and unit name entities;

[0019] S23, using word2vec word vector algorithm and deep learning technology to extract the relationship between entities, and match each accident entity with the characteristic words of the accident cause;

[0020] S24. Based on the accident cause classification and grading framework rules, the direct and indirect causes of historical cases are classified and graded.

[0021] Furthermore, the S3 specifically includes the following steps:

[0022] S31. Use natural language processing, machine learning and corpus tools to understand the semantics of the identified content and to label it professionally according to the accident cause classification and grading framework;

[0023] S32. After the annotation is completed, a review is conducted to ensure the accuracy and professionalism of the annotation and improve the quality of the accident cause dictionary;

[0024] S33. Combine the accident cause classification and grading framework and the reviewed annotation content to build a complete accident cause dictionary, which should include the definition, classification, grading and relevant examples and explanations of various accident causes;

[0025] S34. Combine the accident cause classification and grading framework with the accident cause dictionary to form an accident cause classification and grading analysis method.

[0026] Furthermore, the S4 specifically includes the following steps:

[0027] S41. Based on the accident cause classification and grading analysis method, develop an accident cause classification and grading framework module to display the accident cause in the form of an accident tree page;

[0028] S42. Develop a cause classification and grading identification module to realize the classification and grading identification of accident causes, develop an accident cause dictionary library, and classify and store accident causes according to professional labels;

[0029] S43. Develop ES search function to realize intelligent association search of accident cases by accident type, accident location, accident unit, and accident cause dictionary content, and realize the page display of accident causes in the form of accident tree;

[0030] S44. Based on the TF-IDF similarity calculation model, natural language processing and Bayesian probability model, develop an intelligent matching model for similar accident causes to realize the intelligent matching function for the causes of new accidents.

[0031] In a second aspect, the present invention provides an accident cause classification and grading system, comprising:

[0032] Accident cause classification and grading framework module: used for the relationship diagram component in the echarts front-end data display library, and develops an accident cause classification and grading framework module with a page displaying the accident cause in the form of an accident tree;

[0033] Accident cause classification and grading identification module: Use CRF word segmentation, TextRank keywords, HMM role labeling, word2vec word vector algorithm and deep learning technology to develop an accident cause classification and grading identification module to achieve accident cause classification and grading identification;

[0034] Semantic understanding and professional annotation module: used for natural language processing, machine learning and corpus tools to achieve professional annotation of accident causes;

[0035] Accident cause dictionary module: used for Mysql database and Elasticsearch search engine, to develop an accident cause dictionary function module, to classify and store accident causes according to professional annotations, and to realize intelligent associated retrieval of accident cases according to accident type, accident location, accident unit, and accident cause dictionary content;

[0036] Intelligent matching model for accident causes: Based on the classification and grading analysis method of accident causes, the TF-IDF similarity calculation model, natural language processing and Bayesian probability model are applied to develop an intelligent matching model for accident causes, and realize the intelligent matching function for the accident causes of new accidents.

[0037] In a third aspect, the present invention provides a storage medium, in which instructions are stored. When a computer reads the instructions, the computer executes the method for classifying and grading the causes of accidents.

[0038] In a third aspect, the present invention provides an electronic device, comprising a processor and a storage medium, wherein the processor executes instructions in the storage medium.

[0039] The present invention has the following advantages: the present invention establishes an accident cause classification and grading framework, constructs an accident cause dictionary, forms an accident cause classification and grading method and system, clearly, comprehensively and systematically sorts out the accident causes, realizes rapid classification and grading display of the accident causes of historical accidents, and supports intelligent matching and recommendation of the accident causes of new accidents. It has the characteristics of automation, intelligence, high efficiency and accuracy, and has important application value and promotion prospects in improving accident investigation efficiency and ensuring safe production. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 It is the technical roadmap of the present invention;

[0041] Figure 2 This is an example diagram of the accident cause classification and grading analysis method of the present invention;

[0042] Figure 3 It is the accident analysis model diagram of the present invention. DETAILED DESCRIPTION

[0043] The technical solutions in the embodiments of the present invention will be described clearly and completely below. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention. In addition, the technical solutions between the various embodiments can be combined with each other, but they must be based on the ability of ordinary technicians in this field to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

[0044] The present invention provides a method for classifying and grading accident causes, including constructing a classification and grading framework for accident causes, identifying the classification and grading of accident causes, semantic understanding and professional annotation, and system development and function implementation, such as Figure 1 As shown, the specific steps include:

[0045] S1. Construct a classification and grading framework for accident causes, such as Figure 2 As shown, the direct causes of the accident are classified and analyzed according to unsafe human behavior, unsafe conditions of objects, and unsafe factors of the environment, and the indirect causes are graded according to the correlation between the causes of the accident.

[0046] Among them, unsafe behaviors of people include command errors, operation errors, monitoring errors, protection errors, emergency errors, etc. Unsafe conditions of things: including unsafe conditions of materials, unsafe conditions of equipment and facilities, etc. Unsafe factors of the environment: including unsafe factors indoors, outdoors, and underground, etc.

[0047] S1 specifically includes the following steps:

[0048] S11. Comprehensively apply the intersection theory of human and material causes to form accidents and the Haddon matrix theory of human, equipment and environmental factors interacting to form accidents. Use charts to classify the direct causes of accidents according to unsafe human behavior, unsafe material conditions and unsafe environmental factors.

[0049] S12. Divide the specific hazard factors of unsafe human behavior, unsafe conditions of objects, and unsafe environmental factors into Level 2 factor indicators and Level 3 factor indicators;

[0050] S13. Based on the root cause analysis method and the accident cause structure of domestic accident investigation reports, the indirect causes are divided into factors (inducements) that cause direct causes, enterprise management causes, and system causes. For each direct cause, the inducement is found. For each inducement, the enterprise management cause is identified. For the enterprise management cause, the system cause is analyzed.

[0051] Among them, the inducing factors are the subordinate unit reasons (factors of the perpetrator himself, the unit and other personnel factors) and related unit reasons that are directly related to the direct cause and induce the direct cause. Enterprise management reasons include no safety production education and training / inadequate safety production education and training, failure to establish and improve / (strictly) implement safety operating procedures, failure to (strictly) implement the "three simultaneous" safety facilities of construction projects, inadequate management of major hazardous sources, inadequate management of equipment and facilities, etc. System reasons include government issues, regulations and standards issues, social issues, etc.

[0052] S14. Combined with the above accident cause classification and grading results, a framework for accident cause classification and grading is constructed based on the FTA fault tree analysis method and the MORT parsing tree analysis method.

[0053] S2. Combine big data identification technology to identify the direct and indirect causes of historical cases according to the accident cause classification and grading framework.

[0054] S2 specifically includes the following steps:

[0055] S21, using CRF word segmentation and TextRank keyword extraction to implement word segmentation of accident causes in historical cases and identify characteristic words of accident causes;

[0056] S22, using HMM role-labeled named entity recognition to achieve recognition of person names, place names, and unit name entities;

[0057] S23, using word2vec word vector algorithm and deep learning technology to extract the relationship between entities, and match each accident entity with the characteristic words of the accident cause;

[0058] S24. Based on the accident cause classification and grading framework rules, the direct and indirect causes of historical cases are classified and graded.

[0059] S3. Combine semantic understanding technology to professionally annotate the identified content and build a dictionary of accident causes.

[0060] S3 specifically includes the following steps:

[0061] S31. Use natural language processing, machine learning and corpus tools to understand the semantics of the identified content and to label it professionally according to the accident cause classification and grading framework;

[0062] S32. After the annotation is completed, a review is conducted to ensure the accuracy and professionalism of the annotation and improve the quality of the accident cause dictionary;

[0063] S33. Combine the accident cause classification and grading framework and the reviewed annotation content to build a complete accident cause dictionary, which should include the definition, classification, grading and relevant examples and explanations of various accident causes;

[0064] S34. Combine the accident cause classification and grading framework with the accident cause dictionary to form an accident cause classification and grading analysis method.

[0065] S4. Combine the accident cause classification and grading framework with the construction of an accident cause dictionary to form an accident cause classification and grading analysis method and develop a corresponding system to achieve rapid classification and grading display of accident causes of historical accidents. The system is equipped with an intelligent matching model for accident causes to support intelligent matching recommendations for accident causes of new accidents.

[0066] S4 specifically includes the following steps:

[0067] S41. Based on the accident cause classification and grading analysis method, develop an accident cause classification and grading framework module to display the accident cause in the form of an accident tree page;

[0068] S42. Develop a cause classification and grading identification module to realize the classification and grading identification of accident causes, develop an accident cause dictionary library, and classify and store accident causes according to professional labels;

[0069] S43. Develop ES search function to realize intelligent association search of accident cases by accident type, accident location, accident unit, and accident cause dictionary content, and realize the page display of accident causes in the form of accident tree;

[0070] S44. Based on the TF-IDF similarity calculation model, natural language processing and Bayesian probability model, develop an intelligent matching model for similar accident causes to realize the intelligent matching function for the causes of new accidents.

[0071] The present invention provides an accident cause classification and grading system, comprising:

[0072] Accident cause classification and grading framework module: used for the relationship diagram component in the echarts front-end data display library, and develops an accident cause classification and grading framework module with a page displaying the accident cause in the form of an accident tree;

[0073] Accident cause classification and grading identification module: Use CRF word segmentation, TextRank keywords, HMM role labeling, word2vec word vector algorithm and deep learning technology to develop an accident cause classification and grading identification module to achieve accident cause classification and grading identification;

[0074] Semantic understanding and professional annotation module: used for natural language processing, machine learning and corpus tools to achieve professional annotation of accident causes;

[0075] Accident cause dictionary module: used for Mysql database and Elasticsearch search engine, to develop an accident cause dictionary function module, to classify and store accident causes according to professional annotations, and to realize intelligent associated retrieval of accident cases according to accident type, accident location, accident unit, and accident cause dictionary content;

[0076] Intelligent matching model for accident causes: Based on the classification and grading analysis method of accident causes, the TF-IDF similarity calculation model, natural language processing and Bayesian probability model are applied to develop an intelligent matching model for accident causes, and realize the intelligent matching function for the accident causes of new accidents.

[0077] In this embodiment, the characteristics of cause analysis in domestic accident investigations and well-known domestic and foreign accident cause analysis theories such as fault tree analysis (FTA), event number analysis (ETA), BowTie analysis, and 2-4 model are combined to establish a cause analysis model suitable for accident investigations in my country. This analysis model includes two first-level indicators and six second-level indicators, namely direct causes (unsafe human behavior, unsafe conditions of objects, and unsafe factors in the environment) and indirect causes (inducing factors, enterprise safety management reasons, and system reasons). It further divides multiple third-level indicators and fourth-level indicators according to the characteristics of the accident process and causes of multiple accidents, providing theoretical support for cause analysis in accident investigations. The accident cause analysis model is as follows: Figure 3 shown.

[0078] Although the specific embodiments of the present invention are described in detail in conjunction with the accompanying drawings, it should not be understood as limiting the scope of protection of the present invention. Within the scope described in the claims, various modifications and variations that can be made by those skilled in the art without creative work still belong to the scope of protection of the present invention.

Claims

1. A method for classifying and grading accident causes, characterized in that: The following steps are involved: S1. Construct a classification and grading framework for accident causes, classify and analyze the direct causes of accidents according to unsafe behaviors of people, unsafe conditions of objects, and unsafe factors of the environment, and grade the indirect causes according to the correlation between the causes of accidents; S2. Combine big data identification technology to identify the direct and indirect causes of historical cases according to the accident cause classification and grading framework; S3. Combine semantic understanding technology to professionally annotate the identified content and build a dictionary of accident causes; S4. Combine the accident cause classification and grading framework with the construction of an accident cause dictionary to form an accident cause classification and grading analysis method and develop a corresponding system to achieve rapid classification and grading display of accident causes of historical accidents. The system is equipped with an intelligent matching model for accident causes to support intelligent matching recommendations for accident causes of new accidents.

2. The accident cause classification and grading method according to claim 1, characterized in that: The S1 specifically includes the following steps: S11. Comprehensively apply the intersection theory of human and material causes to form accidents and the Haddon matrix theory of human, equipment and environmental factors interacting to form accidents. Use charts to classify the direct causes of accidents according to unsafe human behavior, unsafe material conditions and unsafe environmental factors. S12. Divide the specific hazard factors of unsafe human behavior, unsafe conditions of objects, and unsafe environmental factors into Level 2 factor indicators and Level 3 factor indicators; S13. Based on the root cause analysis method and the accident cause structure of domestic accident investigation reports, the indirect causes are divided into factors that cause direct causes, enterprise management causes, and system causes. For each direct cause, the inducement is found. For each inducement, the enterprise management cause is identified. For the enterprise management cause, the system cause is analyzed. S14. Combined with the above accident cause classification and grading results, a framework for accident cause classification and grading is constructed based on the FTA fault tree analysis method and the MORT parsing tree analysis method.

3. The accident cause classification and grading method according to claim 1, characterized in that: The S2 specifically includes the following steps: S21, using CRF word segmentation and TextRank keyword extraction to implement word segmentation of accident causes in historical cases and identify characteristic words of accident causes; S22, using HMM role-labeled named entity recognition to achieve recognition of person names, place names, and unit name entities; S23, using word2vec word vector algorithm and deep learning technology to extract the relationship between entities, and match each accident entity with the characteristic words of the accident cause; S24. Based on the accident cause classification and grading framework rules, the direct and indirect causes of historical cases are classified and graded.

4. The accident cause classification and grading method according to claim 1, characterized in that: The S3 specifically includes the following steps: S31. Use natural language processing, machine learning and corpus tools to understand the semantics of the identified content and to label it professionally according to the accident cause classification and grading framework; S32. After the annotation is completed, a review is conducted to ensure the accuracy and professionalism of the annotation and improve the quality of the accident cause dictionary; S33. Combine the accident cause classification and grading framework and the reviewed annotation content to build a complete accident cause dictionary, which should include the definition, classification, grading and relevant examples and explanations of various accident causes; S34. Combine the accident cause classification and grading framework with the accident cause dictionary to form an accident cause classification and grading analysis method.

5. The accident cause classification and grading method according to claim 1, characterized in that: The S4 specifically comprises the following steps: S41. Based on the accident cause classification and grading analysis method, develop an accident cause classification and grading framework module to display the accident cause in the form of an accident tree page; S42. Develop a cause classification and grading identification module to realize the classification and grading identification of accident causes, develop an accident cause dictionary library, and classify and store accident causes according to professional labels; S43. Develop ES search function to realize intelligent association search of accident cases by accident type, accident location, accident unit, and accident cause dictionary content, and realize the page display of accident causes in the form of accident tree; S44. Based on the TF-IDF similarity calculation model, natural language processing and Bayesian probability model, develop an intelligent matching model for similar accident causes to realize the intelligent matching function for the causes of new accidents.

6. An accident cause classification and grading system, characterized in that: include: Accident cause classification and grading framework module: used for the relationship diagram component in the echarts front-end data display library, and develops an accident cause classification and grading framework module with a page displaying the accident cause in the form of an accident tree; Accident cause classification and grading identification module: Use CRF word segmentation, TextRank keywords, HMM role labeling, word2vec word vector algorithm and deep learning technology to develop an accident cause classification and grading identification module to achieve accident cause classification and grading identification; Semantic understanding and professional annotation module: used for natural language processing, machine learning and corpus tools to achieve professional annotation of accident causes; Accident cause dictionary module: used for Mysql database and Elasticsearch search engine, to develop an accident cause dictionary function module, to classify and store accident causes according to professional annotations, and to realize intelligent associated retrieval of accident cases according to accident type, accident location, accident unit, and accident cause dictionary content; Intelligent matching model for accident causes: Based on the classification and grading analysis method of accident causes, the TF-IDF similarity calculation model, natural language processing and Bayesian probability model are applied to develop an intelligent matching model for accident causes, and realize the intelligent matching function for the accident causes of new accidents.

7. A storage medium, characterized in that: The storage medium stores instructions, and when a computer reads the instructions, the computer executes an accident cause classification and grading method as described in any one of claims 1 to 5.

8. An electronic device, characterized in that: The invention comprises a processor and the storage medium as claimed in claim 7, wherein the processor executes instructions in the storage medium.

Images