Study On Risk Analysis Method Of Petrochemical Plant Based On Text Analysis

Petrochemical industry is an important foundation and pillar industry for the development of the national economy.But in recent years,major safety accidents of petrochemical plants in China still occur from time to time.HAZOP,LOPA,SIL and other risk analysis techniques have been applied more widely,in order to ensure the safety of petrochemical plant production.However,the analysts are highly dependent on their knowledge background and relevant experience in risk analysis,which makes risk analysis subjective and uncertain.The analysts need to learn from the experience knowledge of previous risk analysis.Meanwhile,the experience knowledge of finished risk analysis project reports contain a large number of scattered text content,which makes it impossible for analysts to quickly and accurately find useful experience knowledge for risk analysis.Classification of a large number of disorderly risk analysis texts can quickly extract effective information in texts and make more efficient use of experience knowledge in risk analysis,therefore a risk analysis method of petrochemical plant based on text analysis is proposed in this paper.This topic build a complete process of HAZOP analysis and SIL assessment,and conduct risk analysis on several large petrochemical plant.Then the paper build experience database of risk analysis for petrochemical plants by collecting risk analysis reports.This method carry out text analysis of risk analysis experience knowledge by R,so that experience knowledge with similar characteristics is divided into the same group,which simplifies a large number of experience data into different categories.Finally,the classified experience knowledge is constructed as a risk analysis classification database based on text analysis.The classification database can assist the analysts to achieve risk analysis on petrochemical plant by query of database,and reduce the subjectivity and uncertainty of the risk analysis.The main research contents are as follows:(1)HAZOP analysis and SIL identification method are used to carry out risk analysis of petrochemical plant and build experience database.Take a petrochemical n-hexane plant as an example to illustrate the risk analysis process.This paper collects the risk analysis reports of several large petrochemical plants in China,and the HAZOP experience database and SIL experience database of petrochemical plants are constructed respectively.(2)Text analysis of device name,guide word,cause,result and suggestion in HAZOP experience database is conducted.In order to classify the experience database for convenience of query,the paper separately analyzes the cause,result and suggestion in the experience knowledge based on text analysis.The K-means clustering method is used for the text clustering of the cause,result and suggestion experience knowledge,and the experience data is automatically divided into different classification.Taking HAZOP analysis data of gasoline grounding device as new experience knowledge,KNN classification method is used to classify the new experience data,and the new experience data is classified into the categories after clustering.(3)Text analysis of result scene and initial event in SIL experience database is conducted.The result scene description and initial event description in experience database are word segmented by R.Then drawing the wordcloud by word frequency statistics,it's easy to identify common result scene and initial event in SIL experience database.Through text clustering of SIL result scene and initial event experience knowledge,the SIL experience data is effectively classified.(4)Construct classification database of petrochemical plant risk analysis based on text analysis.The query function of the experience classification database is realized by FineReport software.HAZOP experience data and SIL experience data of text classification are constructed into experience database of petrochemical plant risk analysis,and the query is made through the webpage.Combining the text clustering results of HAZOP experience data with the equipment category,it's realized to query the cause,result and suggestion of experience database quickly and effectively.The clustered SIL result scene category is associated with initial event category,and query is made according to the result scene and the corresponding initial event.(5)Case application of risk analysis method for petrochemical plant based on text analysis.HAZOP analysis and SIL identification of butadiene plant have been taken as an example to illustrate the application method of experience classification database.The paper established reasonable classification for experience database through text analysis of experience knowledge of petrochemical plant risk analysis.The method is effective to assist analysts to use experience database for risk analysis and improve the reliability of risk analysis of petrochemical plant.