| As society continues to develop,the country has entered a stage of high-quality growth.Petroleum enterprises play a crucial role in driving economic development,making safety production in the petrochemical industry increasingly important.It is essential to uncover and address potential safety hazards to ensure the health,stability,and sustainable development of the petroleum industry.This article is based on the rich unstructured textual data contained within the QHSE(Quality,Health,Safety,and Environment)management system of petroleum enterprises.It aims to extract valuable information from historical data and summarize potential safety risks and hazards.The article utilizes historical inspection data from 2018 to 2022,with cleaned data including five key elements: the existence of issues,root cause analysis,severity,prevalence,and repetition of problems.The primary focus of the analysis in this article is the short and medium-length textual data under the "existence of issues" column in the inspections.After preprocessing steps such as stop-word removal,custom dictionary creation,and tokenization,common LDA(Latent Dirichlet Allocation)and BTM(Biterm Topic Model)topic analysis models,commonly used in topic mining,were applied to classify and qualitatively analyze the data.After comparing the experimental results,the LDA topic analysis model with better outcomes was selected and validated,resulting in the identification of eight risk topics.These eight risk topics can be divided into two categories: direct hazards that can lead to accidents and indirect risks with a certain lag period.Subsequent visualization and analysis of non-structured data related to root cause analysis,severity,prevalence,and repetition of problems were conducted,ultimately leading to conclusions.The research findings reveal that risks exhibit characteristics of randomness and specificity,with the inadequacy of the transmission of execution and inspection supervision being a contributing factor to their occurrence.In terms of risk levels,the highest-risk issues account for 0.97%,primarily comprising direct risks with a very low repetition rate that can directly lead to accidents.The next highest risk level,serious problems,accounts for 2.43% and mainly pertains to risks related to system processes,job safety,and construction sites,warranting significant attention.The remaining issues account for 96.6%,and as indicated by the word cloud,they represent indirect risks that can be rectified with some buffering.From a temporal perspective,key terms show minimal year-to-year changes,indicating existing issues with rectification efforts.Inspections are becoming more detailed,with the number of problems increasing over the years.There is a growing trend in the proportion of issues related to construction sites and assessment training,while the number of issues related to oil depot equipment safety is decreasing.In the future,it is advisable to establish a database for risk issues at various levels and conduct corresponding specialized inspections and training.Daily efforts should focus on strengthening supervision of system processes,conducting inspections related to job safety and construction sites,which can effectively enhance safety levels.Currently,there is a clear upward trend in issues related to system processes and assessment training,underscoring the need for greater attention to the execution and implementation of work tasks.In the era of digitalization,a peoplecentric approach should be maintained to construct a safety management system and cultivate employee safety awareness to effectively address various safety risks. |
PDF Full Text Request