Research On Human Risk Evaluation And Prediction Model Of Community Indoor Gas

In recent years,indoor gas accidents have occurred frequently,and the retrospective results of many gas accidents indicate that human risk is the main cause of indoor gas accidents.Traditional indoor gas risk assessment methods mostly start from the perspective of the safety status of gas equipment,while ignoring the user’s human risk factors.At the same time,the existing human risk analysis methods mainly focus on special populations such as safe operation and emergency evacuation.Research on the human risk of ordinary users in indoor gas scenarios is rare.Moreover,the current machine learning algorithms used to build human risk prediction models are also seldom applied and have defects such as low prediction accuracy.Based on this,a method for quantitative evaluation of indoor gas risk based on human factors is proposed,and the indoor gas human risk prediction model based on integrated learning is established through the evaluation results,which solves the subjectivity of prediction results caused by relying on expert experience in the past.Problems such as excessive strength have good research application value and significance.First,research and analyze the human risk index system and comprehensive evaluation methods for indoor gas.From the perspective of human risk,analyze the occurrence process of indoor gas accidents,and establish an AHP human factor index system based on this,and propose an indoor gas human risk assessment method based on AHP and fuzzy comprehensive evaluation,which is useful to the community Quantitative risk assessment and analysis of gas human-caused risks.Then,research and establish the indoor gas human-caused risk prediction model.According to the fuzzy comprehensive evaluation results of human risk,three integrated learning algorithms of random forest,adaboost and XGboost are used to establish risk prediction models respectively,compare the prediction effects of the three models under different sample sizes,and apply the best model to the community For gas users,predict and visualize human risk.Finally,a visual display of the indoor gas human-caused risk prediction model is carried out.Establish a front-end and back-end software framework for data collection and storage.The back-end calls the risk prediction model to calculate the human-caused risk prediction results,and the front-end visualizes the prediction results.The results show that the risk prediction model has good interpretability and feasibility.